Why Your Production Schedule Is Wrong Before It Even Starts

1
00:00:00,000 --> 00:00:03,080
Your production schedule looks solid when it leaves the ERP system.

2
00:00:03,080 --> 00:00:06,320
Order dates line up, rooting times make sense, capacity seems available,

3
00:00:06,320 --> 00:00:08,440
and material status says you're good to go.

4
00:00:08,440 --> 00:00:09,560
But here's the problem.

5
00:00:09,560 --> 00:00:11,080
That plan is still an assumption.

6
00:00:11,080 --> 00:00:14,440
It treats yesterday's data as if it describes the factory right now.

7
00:00:14,440 --> 00:00:17,120
Before the first operator even starts the first order.

8
00:00:17,120 --> 00:00:19,520
The real question is what changed since the last planning run?

9
00:00:19,520 --> 00:00:20,760
You don't need another dashboard.

10
00:00:20,760 --> 00:00:23,000
You need a feedback loop that adjusts the decision

11
00:00:23,000 --> 00:00:24,720
when the factory actually changes.

12
00:00:24,720 --> 00:00:28,280
Let's follow one order from the plan to the shop floor and see what happens.

13
00:00:28,280 --> 00:00:31,360
The familiar Friday afternoon scenario.

14
00:00:31,360 --> 00:00:34,400
Picture a plant running mixed products across three shifts.

15
00:00:34,400 --> 00:00:36,600
Shared machines, some specialist tools,

16
00:00:36,600 --> 00:00:39,840
and people who know from experience that a rooting can look clean in the system

17
00:00:39,840 --> 00:00:41,600
while causing real trouble on the floor.

18
00:00:41,600 --> 00:00:45,040
Let's Friday afternoon, the ERP system already released the schedule

19
00:00:45,040 --> 00:00:47,920
based on the demanded new the rooting times in master data,

20
00:00:47,920 --> 00:00:52,000
material status from purchasing, and the capacity calendar for the week.

21
00:00:52,000 --> 00:00:53,280
Nobody did anything wrong.

22
00:00:53,280 --> 00:00:55,600
That was just the information available when it ran.

23
00:00:55,600 --> 00:00:56,720
Now here's the catch.

24
00:00:56,720 --> 00:00:59,240
Much of that information already belongs to yesterday.

25
00:00:59,240 --> 00:01:02,040
A delivery that looked on track this morning now arrives late.

26
00:01:02,040 --> 00:01:04,360
The material for one order may come through the gate,

27
00:01:04,360 --> 00:01:07,080
but it won't reach the point of use when the schedule expects it.

28
00:01:07,080 --> 00:01:09,840
Receiving needs to check it, quality needs to inspect it,

29
00:01:09,840 --> 00:01:14,360
and the work order can't run just because a purchase order line says the material is coming.

30
00:01:14,360 --> 00:01:17,360
At the same time, a tool issue shows up at a shared machine.

31
00:01:17,360 --> 00:01:21,240
Maybe it needs adjustment, or the team needs a different tool from the tool room.

32
00:01:21,240 --> 00:01:25,720
Either way, a job that looked ready in the ERP system can't start at the plant time.

33
00:01:25,720 --> 00:01:28,920
Then an experienced operator calls in sick for the evening shift.

34
00:01:28,920 --> 00:01:31,840
The replacement can run several machines, but not all of them,

35
00:01:31,840 --> 00:01:33,960
and not every product family without support.

36
00:01:33,960 --> 00:01:36,800
That detail lives in someone's head, a skills matrix,

37
00:01:36,800 --> 00:01:38,560
or a spreadsheet on a shared drive.

38
00:01:38,560 --> 00:01:41,120
It usually doesn't travel into the released schedule.

39
00:01:41,120 --> 00:01:44,160
Meanwhile, a short stop hits a machine upstream.

40
00:01:44,160 --> 00:01:48,600
Nothing dramatic, a sensor fault, a jam, a reset, a few checks.

41
00:01:48,600 --> 00:01:49,960
Then production starts again.

42
00:01:49,960 --> 00:01:52,200
On its own, it takes a small slice out of the shift,

43
00:01:52,200 --> 00:01:55,920
but that machine feeds a shared cue that supports work due later in the day.

44
00:01:55,920 --> 00:01:57,760
Small changes build on each other.

45
00:01:57,760 --> 00:02:00,280
The planner now faces a schedule that still looks official,

46
00:02:00,280 --> 00:02:04,000
even though the sequence has drifted away from what people can physically execute.

47
00:02:04,000 --> 00:02:06,240
One order needs material that isn't at the line,

48
00:02:06,240 --> 00:02:08,000
another needs a tool that isn't ready,

49
00:02:08,000 --> 00:02:09,760
a third needs a person who isn't on shift,

50
00:02:09,760 --> 00:02:14,440
and the upstream delay threatens a job that will later need the same resource as a higher priority order.

51
00:02:14,440 --> 00:02:15,760
So the planner opens Excel.

52
00:02:15,760 --> 00:02:18,760
Nobody should be surprised by that Excel responds fast.

53
00:02:18,760 --> 00:02:24,560
A planner can move orders, add notes, color cells, call a supervisor, and send a revised file within minutes.

54
00:02:24,560 --> 00:02:26,960
When a plant needs to recover from a real constraint,

55
00:02:26,960 --> 00:02:29,320
waiting for a slow system update isn't an option.

56
00:02:29,320 --> 00:02:32,480
On the floor, supervisors start doing what good supervisors do.

57
00:02:32,480 --> 00:02:33,520
They use local knowledge.

58
00:02:33,520 --> 00:02:36,600
They know which order can run without a long changeover, which operator can help,

59
00:02:36,600 --> 00:02:39,200
and that a customer order looking urgent may actually be missing apart

60
00:02:39,200 --> 00:02:41,880
while another job keeps the bottleneck busy for the next hour.

61
00:02:41,880 --> 00:02:45,040
That local knowledge keeps production moving, but it can create a gap.

62
00:02:45,040 --> 00:02:47,360
The ERP still shows the customer promise date.

63
00:02:47,360 --> 00:02:49,560
Sales still see the same committed shipment,

64
00:02:49,560 --> 00:02:52,560
management still sees a plan assuming the original sequence,

65
00:02:52,560 --> 00:02:56,760
but the people closest to the work already know the plan has changed.

66
00:02:56,760 --> 00:02:59,560
Changes don't always get recorded in the same place.

67
00:02:59,560 --> 00:03:01,800
A sequence change goes through a quick conversation,

68
00:03:01,800 --> 00:03:03,520
a material issue shows up in an email,

69
00:03:03,520 --> 00:03:05,640
a revised priority lands in a spreadsheet,

70
00:03:05,640 --> 00:03:08,120
a supervisor tells the next shift during handover.

71
00:03:08,120 --> 00:03:11,240
By Monday, several people each have a reasonable version of what happened,

72
00:03:11,240 --> 00:03:12,800
and those versions don't fully match.

73
00:03:12,800 --> 00:03:15,280
This isn't a criticism of planners or supervisors.

74
00:03:15,280 --> 00:03:19,200
It's the normal result of a factory reacting faster than the systems that describe it.

75
00:03:19,200 --> 00:03:22,960
And yes, this is where Excel becomes the unofficial manufacturing control system.

76
00:03:22,960 --> 00:03:26,320
It fills the space between a static schedule and a changing shop floor

77
00:03:26,320 --> 00:03:28,760
because it moves faster than the formal process.

78
00:03:28,760 --> 00:03:32,200
Excel holds the exceptions, the calls, and the practical decisions

79
00:03:32,200 --> 00:03:34,880
that the formal process couldn't capture quickly enough.

80
00:03:34,880 --> 00:03:37,880
Through all of this, the customer date in ERP remains visible,

81
00:03:37,880 --> 00:03:41,080
but it starts to look less like a commitment and more like a hopeful memory

82
00:03:41,080 --> 00:03:42,280
from the last planning run.

83
00:03:42,280 --> 00:03:43,280
Think about that gap.

84
00:03:43,280 --> 00:03:46,040
The first schedule didn't contain bad intentions or bad logic.

85
00:03:46,040 --> 00:03:48,560
The problem is that a schedule document can't react on its own

86
00:03:48,560 --> 00:03:51,040
when material tools, people, and machines change around it.

87
00:03:51,040 --> 00:03:54,240
So we need to separate two things that often get treated as the same thing,

88
00:03:54,240 --> 00:03:58,040
the schedule you release, and the scheduling process that keeps decisions current

89
00:03:58,040 --> 00:04:00,080
once work begins.

90
00:04:00,080 --> 00:04:03,000
A production schedule is a decision about the future.

91
00:04:03,000 --> 00:04:06,040
You look at a production schedule and see columns,

92
00:04:06,040 --> 00:04:10,000
order number, due date, quantity, maybe a machine assignment, straight forward enough.

93
00:04:10,000 --> 00:04:13,160
But each one of those rows is actually a bet on what should happen next.

94
00:04:13,160 --> 00:04:15,920
Even if it just looks like a table someone exported from ERP,

95
00:04:15,920 --> 00:04:18,680
a line on the schedule says, "Job A runs before job B."

96
00:04:18,680 --> 00:04:21,360
It picks a start time, chooses a machine or work center,

97
00:04:21,360 --> 00:04:25,920
names the material, and counts on having the right people with the right skills available.

98
00:04:25,920 --> 00:04:29,920
All those promises sit inside a single decision to take a single order, due next week.

99
00:04:29,920 --> 00:04:32,840
The schedule might place its first operation on a machine at 8 in the morning,

100
00:04:32,840 --> 00:04:35,920
send it to inspection, then push it to the next process later that day.

101
00:04:35,920 --> 00:04:38,000
For that to actually work, the machine has to be free.

102
00:04:38,000 --> 00:04:40,040
The previous operation needs to finish on time.

103
00:04:40,040 --> 00:04:41,760
Material has to arrive at the right spot,

104
00:04:41,760 --> 00:04:44,960
and someone qualified has to be standing there ready to go.

105
00:04:44,960 --> 00:04:47,520
And the schedule also assumes that order won't collide

106
00:04:47,520 --> 00:04:50,240
with another one that needs the same resource more urgently.

107
00:04:50,240 --> 00:04:52,880
It assumes the cycle time you plan is still realistic.

108
00:04:52,880 --> 00:04:54,960
It assumes the setup fits in the time you allow it,

109
00:04:54,960 --> 00:04:57,840
and the next operation won't sit in a queue longer than expected.

110
00:04:57,840 --> 00:05:00,920
That's a lot of hidden assumptions packed into one line of work.

111
00:05:00,920 --> 00:05:05,240
People often talk about the schedule as if it reports what production is actually going to do.

112
00:05:05,240 --> 00:05:06,840
Here's a better way to think about it.

113
00:05:06,840 --> 00:05:09,680
The schedule chooses what production should try to do,

114
00:05:09,680 --> 00:05:12,440
based on whatever facts and rules were available at that moment.

115
00:05:12,440 --> 00:05:14,760
That distinction changes the whole conversation,

116
00:05:14,760 --> 00:05:17,720
because a decision about the future always runs into constraints.

117
00:05:17,720 --> 00:05:20,080
A machine can only process one job at a time.

118
00:05:20,080 --> 00:05:22,400
A fixture might only fit one product family,

119
00:05:22,400 --> 00:05:24,720
an operator may need a specific qualification,

120
00:05:24,720 --> 00:05:26,480
a batch might have to stay together.

121
00:05:26,480 --> 00:05:29,680
And a long changeover can make a perfectly sensible due date sequence

122
00:05:29,680 --> 00:05:31,360
impossible to run in practice.

123
00:05:31,360 --> 00:05:33,440
Capacity isn't an empty block on a calendar,

124
00:05:33,440 --> 00:05:36,680
it's usable time on real resources under real conditions.

125
00:05:36,680 --> 00:05:37,880
And those conditions shift,

126
00:05:37,880 --> 00:05:41,000
a plan start time gets less certain when the prior job runs long.

127
00:05:41,000 --> 00:05:44,040
A material promise gets less certain when that material needs inspection.

128
00:05:44,040 --> 00:05:48,760
A resource assignment gets less certain when the work centers actual state changes mid shift.

129
00:05:48,760 --> 00:05:50,080
None of this means planning failed.

130
00:05:50,080 --> 00:05:52,680
It means planning deals with the future that hasn't happened yet.

131
00:05:52,680 --> 00:05:54,320
There are also different levels of planning

132
00:05:54,320 --> 00:05:56,480
and mixing them together creates confusion.

133
00:05:56,480 --> 00:05:58,400
A plan usually works at a broader level.

134
00:05:58,400 --> 00:05:59,600
It answers questions like,

135
00:05:59,600 --> 00:06:00,880
what demand do we need to cover?

136
00:06:00,880 --> 00:06:03,160
Which week should we produce this product family?

137
00:06:03,160 --> 00:06:06,280
And do we have enough capacity and material across a longer horizon?

138
00:06:06,280 --> 00:06:08,280
A detailed schedule moves closer to the floor.

139
00:06:08,280 --> 00:06:11,080
It decides which operation runs on which resource in what order

140
00:06:11,080 --> 00:06:13,000
and within a more precise time window.

141
00:06:13,000 --> 00:06:14,760
Now you're dealing with finite capacity,

142
00:06:14,760 --> 00:06:16,520
not just broad demand coverage.

143
00:06:16,520 --> 00:06:17,760
Then you have a dispatch list.

144
00:06:17,760 --> 00:06:19,440
That's the immediate instruction for the floor.

145
00:06:19,440 --> 00:06:21,520
It answers a much more practical question,

146
00:06:21,520 --> 00:06:23,000
given the current situation,

147
00:06:23,000 --> 00:06:26,000
what should this person or this machine run next?

148
00:06:26,000 --> 00:06:27,680
Those are related decisions,

149
00:06:27,680 --> 00:06:29,360
but they aren't the same decision.

150
00:06:29,360 --> 00:06:32,000
The plan may say an order belongs in this week's production.

151
00:06:32,000 --> 00:06:35,240
The detailed schedule may reserve time for it on a resource tomorrow.

152
00:06:35,240 --> 00:06:38,560
The dispatch list may need to decide whether it can run in the next hour.

153
00:06:38,560 --> 00:06:40,760
If you treat all three as one fixed document,

154
00:06:40,760 --> 00:06:43,520
the floor ends up carrying the burden of resolving the gaps.

155
00:06:43,520 --> 00:06:45,080
A static schedule still has a place.

156
00:06:45,080 --> 00:06:46,160
You need a starting point.

157
00:06:46,160 --> 00:06:48,720
You need a shared intent about demand, priorities,

158
00:06:48,720 --> 00:06:50,560
capacity and customer commitments.

159
00:06:50,560 --> 00:06:52,760
Without that, every shift starts from scratch

160
00:06:52,760 --> 00:06:55,320
and nobody wants to run a factory by collective guesswork.

161
00:06:55,320 --> 00:06:57,640
But a static schedule is weak as a control loop.

162
00:06:57,640 --> 00:06:59,480
A control loop takes a planned action,

163
00:06:59,480 --> 00:07:02,240
watches what happens, compares the result with the intent

164
00:07:02,240 --> 00:07:04,680
and changes the next action when conditions call for it.

165
00:07:04,680 --> 00:07:06,960
A schedule that only gets published does the first part.

166
00:07:06,960 --> 00:07:08,000
It creates an instruction.

167
00:07:08,000 --> 00:07:10,360
It doesn't have a built-in way to revise that instruction

168
00:07:10,360 --> 00:07:12,320
when the factory produces new facts.

169
00:07:12,320 --> 00:07:13,880
That leaves people with an awkward choice.

170
00:07:13,880 --> 00:07:16,720
They can follow a schedule they no longer fits the work.

171
00:07:16,720 --> 00:07:20,120
Or they can adapt locally, which often keeps production moving

172
00:07:20,120 --> 00:07:22,520
but can detach daily decisions from the wider plan.

173
00:07:22,520 --> 00:07:24,480
So when execution changes who changes the decision,

174
00:07:24,480 --> 00:07:27,080
is it the planner after hearing about the issue later?

175
00:07:27,080 --> 00:07:28,680
Is it the supervisor at the resource?

176
00:07:28,680 --> 00:07:31,320
Is it a spreadsheet someone updates between calls?

177
00:07:31,320 --> 00:07:34,080
Or does the operating model give the right people current information

178
00:07:34,080 --> 00:07:37,640
clear constraints and a controlled way to issue the next dispatch decision?

179
00:07:37,640 --> 00:07:40,240
That question takes us directly to the limits of ERP planning.

180
00:07:40,240 --> 00:07:43,120
An ERP system can plan demand and commitments really well,

181
00:07:43,120 --> 00:07:46,080
but the factory changes at a pace and level of detail

182
00:07:46,080 --> 00:07:48,720
that broad planning cycles don't always capture.

183
00:07:48,720 --> 00:07:52,360
What ERP knows and what it cannot know alone.

184
00:07:52,360 --> 00:07:54,800
ERP earns its place at the centre of the business.

185
00:07:54,800 --> 00:07:57,720
It holds demand, customer orders, bills of material,

186
00:07:57,720 --> 00:08:00,560
purchase status, due dates, inventory records

187
00:08:00,560 --> 00:08:04,360
and the commercial commitments that teleplant why a given order matters.

188
00:08:04,360 --> 00:08:07,640
Without that information, production planning turns into guesswork.

189
00:08:07,640 --> 00:08:10,160
When an ERP planning run creates a proposed schedule,

190
00:08:10,160 --> 00:08:12,600
it works from the data the business has agreed to use.

191
00:08:12,600 --> 00:08:14,840
It reads lead times, it reads routing standards,

192
00:08:14,840 --> 00:08:17,080
it reads work centre calendars, material records,

193
00:08:17,080 --> 00:08:19,880
order priorities and the planned relationships between components

194
00:08:19,880 --> 00:08:20,920
and finished goods.

195
00:08:20,920 --> 00:08:24,000
That logic can be sound, but it only knows what reaches the model.

196
00:08:24,000 --> 00:08:27,280
Here's the thing, a routing might allow 40 minutes for an operation

197
00:08:27,280 --> 00:08:29,600
because that figure entered the system years ago

198
00:08:29,600 --> 00:08:33,080
or because it represents a standard rather than the current day-to-day result.

199
00:08:33,080 --> 00:08:36,200
The calendar might show eight hours of capacity for a work centre

200
00:08:36,200 --> 00:08:39,120
while the real resource needs a long setup, a tool check

201
00:08:39,120 --> 00:08:42,000
and an experienced operator who only works one shift.

202
00:08:42,000 --> 00:08:45,080
Those aren't small details when you're trying to decide what runs next.

203
00:08:45,080 --> 00:08:48,240
ERP can tell you that a purchase order has a due date.

204
00:08:48,240 --> 00:08:51,520
It can show that material has been ordered, received or allocated

205
00:08:51,520 --> 00:08:53,040
according to the business process,

206
00:08:53,040 --> 00:08:56,240
but the system may not know whether that material sits at the point of use

207
00:08:56,240 --> 00:09:00,560
waits for inspection, belongs to the correct batch, or can support the exact operation

208
00:09:00,560 --> 00:09:01,840
that needs it right now.

209
00:09:01,840 --> 00:09:03,520
The same gap appears with capacity.

210
00:09:03,520 --> 00:09:06,680
A work centre in ERP often represents a useful planning unit.

211
00:09:06,680 --> 00:09:09,560
It might refer to one machine, several similar machines

212
00:09:09,560 --> 00:09:13,560
or a broader area of the plant that helps with demand and broad load planning.

213
00:09:13,560 --> 00:09:17,720
Still, it doesn't automatically describe the current queue in front of each resource,

214
00:09:17,720 --> 00:09:19,520
the real setup already underway,

215
00:09:19,520 --> 00:09:22,200
or the tool condition that could stop the next job.

216
00:09:22,200 --> 00:09:24,040
And it usually can't see the human detail

217
00:09:24,040 --> 00:09:26,280
unless someone has built and maintained that connection.

218
00:09:26,280 --> 00:09:28,760
Who on this shift holds the right qualification?

219
00:09:28,760 --> 00:09:31,400
Who knows the process well enough to run a difficult product?

220
00:09:31,400 --> 00:09:33,800
Who has already been pulled to solve a problem elsewhere?

221
00:09:33,800 --> 00:09:35,720
Those facts change what capacity means.

222
00:09:35,720 --> 00:09:37,240
This isn't an ERP defect.

223
00:09:37,240 --> 00:09:38,680
It describes a system boundary.

224
00:09:38,680 --> 00:09:40,840
ERP manages the business view of production.

225
00:09:40,840 --> 00:09:44,840
It connects customer demand, supply, finance, inventory, purchasing,

226
00:09:44,840 --> 00:09:46,120
and order commitments.

227
00:09:46,120 --> 00:09:49,160
That role matters because production can't decide an isolation

228
00:09:49,160 --> 00:09:52,200
from the orders it needs to fulfill or the material it needs to consume,

229
00:09:52,200 --> 00:09:54,360
but shop flow control deals with a different clock.

230
00:09:54,360 --> 00:09:56,600
Planning may run overnight, once per shift,

231
00:09:56,600 --> 00:09:58,520
or at defined points during the day.

232
00:09:58,520 --> 00:10:01,880
A factory can face a disturbance between two machine cycles.

233
00:10:01,880 --> 00:10:03,560
A tool can wear out during a run.

234
00:10:03,560 --> 00:10:06,600
A material issue can appear when a batch reaches inspection.

235
00:10:06,600 --> 00:10:08,840
An operator can discover that the actual setup

236
00:10:08,840 --> 00:10:10,200
doesn't match the rooting assumption.

237
00:10:10,200 --> 00:10:13,160
The planning run doesn't become wrong because the ERP failed.

238
00:10:13,160 --> 00:10:14,600
It becomes older than the factory.

239
00:10:14,600 --> 00:10:17,160
That sounds obvious, but many operating models still act

240
00:10:17,160 --> 00:10:20,760
as if the released plant stays authoritative until the next formal update.

241
00:10:20,760 --> 00:10:22,680
People then create their own workarounds

242
00:10:22,680 --> 00:10:25,240
because the physical process can't wait for the next cycle.

243
00:10:25,240 --> 00:10:27,160
I think the more useful view is simple.

244
00:10:27,160 --> 00:10:28,840
ERP owns the business intent.

245
00:10:28,840 --> 00:10:31,720
It tells you what demand exists, what commitments matter,

246
00:10:31,720 --> 00:10:33,640
what material should support the work,

247
00:10:33,640 --> 00:10:35,720
and what the organization plan to produce.

248
00:10:35,720 --> 00:10:37,000
It should remain the backbone.

249
00:10:37,000 --> 00:10:40,360
It just shouldn't pretend to be the person standing next to the machine.

250
00:10:40,360 --> 00:10:42,120
For a schedule to stay usable,

251
00:10:42,120 --> 00:10:45,400
the business intent needs to meet the current production state.

252
00:10:45,400 --> 00:10:48,760
That means moving closer to where the schedule turns into physical work,

253
00:10:48,760 --> 00:10:51,160
where orders enter cues people set up resources

254
00:10:51,160 --> 00:10:53,880
and the factory starts producing facts rather than plans.

255
00:10:53,880 --> 00:10:57,160
Execute where the schedule meets the real factory.

256
00:10:57,160 --> 00:11:00,200
Here's where the plan hits reality.

257
00:11:00,200 --> 00:11:01,800
Once an order lands on the shop floor,

258
00:11:01,800 --> 00:11:03,400
planning hands off to execution.

259
00:11:03,400 --> 00:11:06,440
And that's when the MES, the operators, the machines,

260
00:11:06,440 --> 00:11:10,280
quality teams and maintenance all converge around the same physical work.

261
00:11:10,280 --> 00:11:12,920
The MES releases the work order and tracks progress.

262
00:11:12,920 --> 00:11:15,880
The operator reads the instructions, grabs the material,

263
00:11:15,880 --> 00:11:18,280
sets up the machine, and starts the operation.

264
00:11:18,280 --> 00:11:20,920
The machine runs, stops, alarms or waits.

265
00:11:20,920 --> 00:11:23,800
Quality inspects the first off-part or puts a batch on hold.

266
00:11:23,800 --> 00:11:26,520
Maintenance steps in because a fault needs attention

267
00:11:26,520 --> 00:11:28,120
before production can continue.

268
00:11:28,120 --> 00:11:29,160
None of that is theory.

269
00:11:29,160 --> 00:11:31,560
That's the factory deciding, minute by minute,

270
00:11:31,560 --> 00:11:33,720
whether the plan can actually turn into product.

271
00:11:33,720 --> 00:11:36,360
Think about one work order entering a machining area.

272
00:11:36,360 --> 00:11:37,960
The schedule says it starts at 10,

273
00:11:37,960 --> 00:11:41,320
but the prior order finishes late because the actual cycle time drifted.

274
00:11:41,320 --> 00:11:43,720
The operator needs a few extra minutes for setup.

275
00:11:43,720 --> 00:11:45,320
The fixture needs adjustment.

276
00:11:45,320 --> 00:11:47,160
Then the first part goes to quality,

277
00:11:47,160 --> 00:11:49,400
and quality spots something that needs clarification

278
00:11:49,400 --> 00:11:50,920
before the full batch can run.

279
00:11:50,920 --> 00:11:52,520
So the order didn't start at 10.

280
00:11:52,520 --> 00:11:54,600
It started when the conditions allowed it to start.

281
00:11:54,600 --> 00:11:57,880
That sounds simple, but it changes how you should think about execution data.

282
00:11:57,880 --> 00:12:01,880
Those actual start and finish times aren't just time stamps for a report later.

283
00:12:01,880 --> 00:12:04,360
Their facts about the capacity, the factory really used.

284
00:12:04,360 --> 00:12:08,200
The queue it created, and the choice is still available for every order behind it.

285
00:12:08,200 --> 00:12:09,800
Scrap changes the picture to.

286
00:12:09,800 --> 00:12:11,960
A quantity that looked available in the plan

287
00:12:11,960 --> 00:12:14,200
might no longer cover the next operation.

288
00:12:14,200 --> 00:12:18,360
A quality hold can pause work that technically looks complete from the machine's perspective.

289
00:12:18,360 --> 00:12:21,400
Rework can send material back through part of the routing,

290
00:12:21,400 --> 00:12:24,120
where it competes for the same resources as planned work.

291
00:12:24,120 --> 00:12:25,800
Changeovers tell a similar story.

292
00:12:25,800 --> 00:12:29,160
A schedule might reserve a standard amount of time for a product change,

293
00:12:29,160 --> 00:12:31,800
but the actual change depends on the product pair,

294
00:12:31,800 --> 00:12:33,800
the tool state, the material,

295
00:12:33,800 --> 00:12:35,560
and sometimes the person doing the work.

296
00:12:35,560 --> 00:12:39,080
A short delay in a changeover can shift the whole queue

297
00:12:39,080 --> 00:12:41,480
when the resource is already running near its limit.

298
00:12:41,480 --> 00:12:43,320
And then there are the small interruptions.

299
00:12:43,320 --> 00:12:45,640
A jam gets cleared, a sensor needs a reset.

300
00:12:45,640 --> 00:12:48,760
An operator waits for a forklift, someone checks an unusual part.

301
00:12:48,760 --> 00:12:51,640
Each event may look too small to trigger an urgent meeting,

302
00:12:51,640 --> 00:12:54,840
but the accumulated time can change what the shift can actually finish.

303
00:12:54,840 --> 00:12:57,960
That's why a work order produces facts the planning model couldn't fully predict.

304
00:12:57,960 --> 00:12:59,240
Planning estimates.

305
00:12:59,240 --> 00:13:00,600
Execution reveals.

306
00:13:00,600 --> 00:13:02,840
Execution has hard physical limits.

307
00:13:02,840 --> 00:13:04,600
One machine can't run two jobs at once,

308
00:13:04,600 --> 00:13:07,080
even if both look urgent in a priority report.

309
00:13:07,080 --> 00:13:09,480
One fixture can't support two parts at once.

310
00:13:09,480 --> 00:13:12,920
One qualified operator can't stand at two work centers during the same half hour,

311
00:13:12,920 --> 00:13:15,000
no matter how optimistic the calendar looks.

312
00:13:15,000 --> 00:13:17,640
The shop floor resolves those collisions every single day.

313
00:13:17,640 --> 00:13:19,720
Picture a supervisor at the start of a shift.

314
00:13:19,720 --> 00:13:21,160
Two orders need the same machine.

315
00:13:21,160 --> 00:13:22,760
One order has an earlier due date,

316
00:13:22,760 --> 00:13:24,840
but its material hasn't arrived at the line.

317
00:13:24,840 --> 00:13:26,200
The other order can run now,

318
00:13:26,200 --> 00:13:27,560
uses the same setup family,

319
00:13:27,560 --> 00:13:29,240
and keeps the resource productive.

320
00:13:29,240 --> 00:13:31,400
The supervisor has to decide whether to wait,

321
00:13:31,400 --> 00:13:32,920
switch, or prep something else.

322
00:13:32,920 --> 00:13:35,240
That decision usually happens in a call with the planner,

323
00:13:35,240 --> 00:13:36,360
a note on a whiteboard,

324
00:13:36,360 --> 00:13:38,440
or a quick discussion next to the machine.

325
00:13:38,440 --> 00:13:40,680
In many plans, that's not poor discipline.

326
00:13:40,680 --> 00:13:44,920
Its experience people handling conditions the formal scheduled in capture in time.

327
00:13:44,920 --> 00:13:47,000
Experience carries practical detail.

328
00:13:47,000 --> 00:13:49,800
A supervisor might know a machine can technically process an order

329
00:13:49,800 --> 00:13:51,640
but will struggle with the current tool.

330
00:13:51,640 --> 00:13:55,080
An operator might know a product needs extra care after a recent issue.

331
00:13:55,080 --> 00:13:57,160
Maintenance might know a resource can run,

332
00:13:57,160 --> 00:13:59,720
but only until a planned inspection later in the shift.

333
00:13:59,720 --> 00:14:03,160
The problem comes when those decisions disappear after the conversation ends.

334
00:14:03,160 --> 00:14:05,800
If a revised dispatch choice only lives on a whiteboard,

335
00:14:05,800 --> 00:14:08,360
the next shift may not know why the sequence changed.

336
00:14:08,360 --> 00:14:11,400
If a quality hold reaches the operator through a verbal message,

337
00:14:11,400 --> 00:14:14,600
the planner may not see the capacity effect until much later.

338
00:14:14,600 --> 00:14:17,240
If maintenance changes the usable condition of a resource,

339
00:14:17,240 --> 00:14:20,120
the schedule may keep assuming capacity that no longer exists.

340
00:14:20,120 --> 00:14:22,920
So execution needs more than a released work order.

341
00:14:22,920 --> 00:14:25,480
It needs a way to capture what actually happened around that work order

342
00:14:25,480 --> 00:14:27,800
with enough detail to support the next decision.

343
00:14:27,800 --> 00:14:29,800
That doesn't mean recording every human movement

344
00:14:29,800 --> 00:14:31,960
or flooding the cloud with raw machine signals.

345
00:14:31,960 --> 00:14:33,960
Factories already generate enough data

346
00:14:33,960 --> 00:14:36,360
to keep several consultants busy for years.

347
00:14:36,360 --> 00:14:39,640
The point is to capture the events that change your ability to execute work.

348
00:14:39,640 --> 00:14:41,400
A job started late, a machine stopped,

349
00:14:41,400 --> 00:14:42,840
a quality check blocked release,

350
00:14:42,840 --> 00:14:44,600
a change over took longer than expected.

351
00:14:44,600 --> 00:14:47,320
A maintenance action removed a resource from service.

352
00:14:47,320 --> 00:14:49,000
Those aren't just production records.

353
00:14:49,000 --> 00:14:50,680
They change the schedule's assumptions,

354
00:14:50,680 --> 00:14:53,160
while the schedule is still supposed to guide the day.

355
00:14:53,160 --> 00:14:55,080
Execution produces the signals,

356
00:14:55,080 --> 00:14:57,560
signals alone don't close the loop.

357
00:14:57,560 --> 00:14:59,720
Measure means more than collecting machine data.

358
00:14:59,720 --> 00:15:02,440
A machine can produce a lot of signals

359
00:15:02,440 --> 00:15:04,920
without telling you much about the production decision.

360
00:15:04,920 --> 00:15:07,080
It might report speed, temperature, vibration,

361
00:15:07,080 --> 00:15:08,680
current draw alarms, run state,

362
00:15:08,680 --> 00:15:11,640
and dozens of tags that matter to maintenance or process control.

363
00:15:11,640 --> 00:15:13,160
That telemetry has a purpose.

364
00:15:13,160 --> 00:15:15,640
But a machine event only becomes a production fact

365
00:15:15,640 --> 00:15:19,000
when you can connect it to the work that event affected,

366
00:15:19,000 --> 00:15:20,920
who or what perform that work,

367
00:15:20,920 --> 00:15:23,880
and where that work sits in the current flow of orders.

368
00:15:23,880 --> 00:15:25,960
Say a machine reports a 20-minute stop

369
00:15:25,960 --> 00:15:28,360
on its own that tells you a resource lost time.

370
00:15:28,360 --> 00:15:31,240
But you still can't tell whether the stop affected a rush order,

371
00:15:31,240 --> 00:15:33,160
a low priority replenishment run,

372
00:15:33,160 --> 00:15:34,600
a batch waiting downstream,

373
00:15:34,600 --> 00:15:37,000
or a job that had enough buffer to absorb the delay.

374
00:15:37,000 --> 00:15:39,240
The production system needs that surrounding detail.

375
00:15:39,240 --> 00:15:40,920
Tie the event to the work order,

376
00:15:40,920 --> 00:15:42,920
tie it to the operation within the routing,

377
00:15:42,920 --> 00:15:44,520
the product and batch involved,

378
00:15:44,520 --> 00:15:45,720
the specific resource,

379
00:15:45,720 --> 00:15:47,720
the shift, and a reason code people can trust.

380
00:15:47,720 --> 00:15:52,600
Then that same 20-minute stop starts to answer practical questions.

381
00:15:52,600 --> 00:15:54,840
Which order lost capacity did the stop interrupt

382
00:15:54,840 --> 00:15:56,200
a setup or a production run?

383
00:15:56,200 --> 00:15:58,760
Did it affect one batch that needs to stay together?

384
00:15:58,760 --> 00:16:01,080
Did it happen during a shift with limited labor?

385
00:16:01,080 --> 00:16:02,520
Was the cause of planned change?

386
00:16:02,520 --> 00:16:05,640
Material waiting, a quality check, a machine fault, or something else?

387
00:16:05,640 --> 00:16:07,640
That may sound like data modeling detail.

388
00:16:07,640 --> 00:16:08,360
It's not.

389
00:16:08,360 --> 00:16:10,440
It determines whether a planner can act on the event

390
00:16:10,440 --> 00:16:13,320
without calling half the plant to reconstruct the story.

391
00:16:13,320 --> 00:16:15,960
Think about an operator marking an order complete.

392
00:16:15,960 --> 00:16:17,480
Complete can mean many things,

393
00:16:17,480 --> 00:16:19,320
unless the event definition is clear.

394
00:16:19,320 --> 00:16:21,000
It might mean the machine cycle ended,

395
00:16:21,000 --> 00:16:23,320
it might mean the full quantity passed inspection,

396
00:16:23,320 --> 00:16:25,080
it might mean the operator finished their step

397
00:16:25,080 --> 00:16:26,760
and moved the batch to a queue.

398
00:16:26,760 --> 00:16:29,640
Each meaning changes the state of the next operation.

399
00:16:29,640 --> 00:16:31,800
So event definitions need agreement.

400
00:16:31,800 --> 00:16:34,040
Not perfect agreement in a workshop slide deck,

401
00:16:34,040 --> 00:16:36,120
but agreement that survives a Friday shift,

402
00:16:36,120 --> 00:16:38,600
a system update, and a handover between teams.

403
00:16:38,600 --> 00:16:39,560
Time matters too.

404
00:16:39,560 --> 00:16:42,200
You need timestamps that identify when an event actually occurred,

405
00:16:42,200 --> 00:16:44,040
not just when someone entered it later.

406
00:16:44,040 --> 00:16:45,320
If a job started at 9,

407
00:16:45,320 --> 00:16:48,040
stopped at 10, resumed at 11, and finished at 1.

408
00:16:48,040 --> 00:16:50,840
That sequence tells you far more than a single completion entry

409
00:16:50,840 --> 00:16:52,440
posted at the end of the shift.

410
00:16:52,440 --> 00:16:53,560
It shows actual runtime.

411
00:16:53,560 --> 00:16:54,840
It shows lost time.

412
00:16:54,840 --> 00:16:57,240
It may expose a weight before the job began.

413
00:16:57,240 --> 00:16:59,000
And it helps separate a routing estimate

414
00:16:59,000 --> 00:17:01,080
from the time the resource truly consumed.

415
00:17:01,080 --> 00:17:04,760
This is where many OEE conversations become incomplete.

416
00:17:04,760 --> 00:17:07,880
Overall equipment effectiveness can help you see availability,

417
00:17:07,880 --> 00:17:09,800
performance, and quality at a resource.

418
00:17:09,800 --> 00:17:11,800
It can show that a machine lost time,

419
00:17:11,800 --> 00:17:13,640
ran below its expected pace,

420
00:17:13,640 --> 00:17:15,480
or produced reject parts.

421
00:17:15,480 --> 00:17:16,200
That's useful.

422
00:17:16,200 --> 00:17:17,400
Nobody should dismiss it.

423
00:17:17,400 --> 00:17:20,040
But OEE doesn't decide which order should run next.

424
00:17:20,040 --> 00:17:22,040
A machine can show a good OEE result

425
00:17:22,040 --> 00:17:24,200
while the plant still misses a customer commitment

426
00:17:24,200 --> 00:17:26,440
because the wrong work set in front of the bottleneck.

427
00:17:26,440 --> 00:17:28,360
It can also show a poor result for a reason

428
00:17:28,360 --> 00:17:30,760
that has little impact on the current delivery risk.

429
00:17:30,760 --> 00:17:32,920
OEE measures how a resource performed.

430
00:17:32,920 --> 00:17:35,240
Scheduling needs to decide how limited resources

431
00:17:35,240 --> 00:17:37,160
should be used across competing work.

432
00:17:37,160 --> 00:17:38,840
For that, you need measures at the level

433
00:17:38,840 --> 00:17:40,280
where decisions happen.

434
00:17:40,280 --> 00:17:43,480
Q-time tells you how long work waits between operations.

435
00:17:43,480 --> 00:17:45,640
Actual cycle time tells you whether the routing

436
00:17:45,640 --> 00:17:47,160
still fits the process.

437
00:17:47,160 --> 00:17:48,920
Yield tells you whether the expected output

438
00:17:48,920 --> 00:17:50,040
will reach the next step.

439
00:17:50,040 --> 00:17:53,000
Availability tells you whether a resource can take work.

440
00:17:53,000 --> 00:17:55,640
Constraint use tells you whether the resource limiting flow

441
00:17:55,640 --> 00:17:58,360
spends its usable time on the work that matters most.

442
00:17:58,360 --> 00:18:01,240
Each measure answers a different part of the same question.

443
00:18:01,240 --> 00:18:03,000
Can this schedule still happen?

444
00:18:03,000 --> 00:18:05,800
A long queue may point to a blocked downstream resource.

445
00:18:05,800 --> 00:18:08,360
A cycle time drift may quietly consume the capacity

446
00:18:08,360 --> 00:18:09,880
the next order is expected.

447
00:18:09,880 --> 00:18:12,680
No yield may turn a material plan into a shortage.

448
00:18:12,680 --> 00:18:14,600
A resource might show available time on paper

449
00:18:14,600 --> 00:18:17,080
while a required tool, person, or support process

450
00:18:17,080 --> 00:18:18,760
limits the work in practice.

451
00:18:18,760 --> 00:18:20,280
Measure at the grain of the decision

452
00:18:20,280 --> 00:18:24,280
that usually means order, operation, resource, and time

453
00:18:24,280 --> 00:18:26,440
with enough context to explain the variance later.

454
00:18:26,440 --> 00:18:28,840
You don't need every signal from every device for that.

455
00:18:28,840 --> 00:18:31,720
You need reliable events with clear names, clear time logic,

456
00:18:31,720 --> 00:18:33,640
and a link to the production work affected.

457
00:18:33,640 --> 00:18:35,800
If operators, maintenance, quality, and planning

458
00:18:35,800 --> 00:18:37,800
all record the same event differently,

459
00:18:37,800 --> 00:18:41,080
the data moves faster, but the factory doesn't understand more.

460
00:18:41,080 --> 00:18:42,920
And that leads to the next problem.

461
00:18:42,920 --> 00:18:45,960
Data without shared meaning creates faster confusion.

462
00:18:45,960 --> 00:18:48,200
The data problem is usually a context problem.

463
00:18:48,200 --> 00:18:52,040
Here's what most factories actually run into.

464
00:18:52,040 --> 00:18:55,800
Plenty of data, but no shared way to describe what any of it refers to.

465
00:18:55,800 --> 00:18:56,760
Take one machine.

466
00:18:56,760 --> 00:18:59,960
In ERP, it shows up as a work center with some business-friendly code,

467
00:18:59,960 --> 00:19:03,640
while MES calls it a resource ID, maintenance tracks it by asset number,

468
00:19:03,640 --> 00:19:06,600
and the IoT layer sends signals under whatever controller tag

469
00:19:06,600 --> 00:19:09,960
or gateway name the engineer used when the system went live.

470
00:19:09,960 --> 00:19:12,200
Each of those systems is correct inside its own world,

471
00:19:12,200 --> 00:19:14,120
but none of those names talk to each other.

472
00:19:14,120 --> 00:19:15,400
And when the names don't connect,

473
00:19:15,400 --> 00:19:17,400
you cannot reliably link a production delay

474
00:19:17,400 --> 00:19:19,960
to the machine condition, the maintenance record,

475
00:19:19,960 --> 00:19:21,800
and the orders that were waiting for that machine.

476
00:19:21,800 --> 00:19:23,800
So people end up doing the joining manually.

477
00:19:23,800 --> 00:19:26,200
A planner sees an overloaded work center in ERP,

478
00:19:26,200 --> 00:19:29,240
a maintenance technician sees an alarm against an asset ID,

479
00:19:29,240 --> 00:19:31,800
and an engineer sees a stop event from a tag stream.

480
00:19:31,800 --> 00:19:35,160
Then, someone who knows the plant translates between all three,

481
00:19:35,160 --> 00:19:36,520
usually over the phone.

482
00:19:36,520 --> 00:19:39,640
That person essentially becomes part of the integration architecture.

483
00:19:39,640 --> 00:19:41,240
It works fine until they take a holiday.

484
00:19:41,240 --> 00:19:42,760
The naming problem is only one piece.

485
00:19:42,760 --> 00:19:44,840
The harder issue is that words that seem obvious

486
00:19:44,840 --> 00:19:47,880
end up meaning different things once multiple systems have to act on them.

487
00:19:47,880 --> 00:19:49,400
Take the word complete.

488
00:19:49,400 --> 00:19:52,920
To a machine, complete means the last cycle finished.

489
00:19:52,920 --> 00:19:55,800
To mess, it means the operator recorded the produced quantity.

490
00:19:55,800 --> 00:19:58,920
To quality, it means inspection cleared the batch.

491
00:19:58,920 --> 00:20:01,080
To the warehouse material reach storage.

492
00:20:01,080 --> 00:20:04,040
To planning, it means the next operation can actually start.

493
00:20:04,040 --> 00:20:05,400
Those are all different states,

494
00:20:05,400 --> 00:20:08,120
and treating them as one creates false confidence.

495
00:20:08,120 --> 00:20:09,960
Available creates the same problem.

496
00:20:09,960 --> 00:20:13,560
A machine may show as available because it isn't in an active production run,

497
00:20:13,560 --> 00:20:15,880
while maintenance has already marked it for a check.

498
00:20:15,880 --> 00:20:19,320
Material may show as available in inventory while it sits in quarantine.

499
00:20:19,320 --> 00:20:21,960
An operator may appear available in the workforce system

500
00:20:21,960 --> 00:20:24,920
while they lack the certification for the order waiting at the machine.

501
00:20:24,920 --> 00:20:27,160
And planned can mean a demand forecast,

502
00:20:27,160 --> 00:20:28,520
a released production order,

503
00:20:28,520 --> 00:20:29,880
a slot in a detailed schedule,

504
00:20:29,880 --> 00:20:32,280
or a supervises decision about the next job.

505
00:20:32,280 --> 00:20:34,600
If a team uses the same word for all those things,

506
00:20:34,600 --> 00:20:37,400
people can agree on the word while disagreeing on the work.

507
00:20:37,400 --> 00:20:40,760
This is why more event data doesn't automatically fix the problem.

508
00:20:40,760 --> 00:20:43,800
You can stream thousands of events from machines into Azure,

509
00:20:43,800 --> 00:20:45,480
store them in Microsoft Fabric,

510
00:20:45,480 --> 00:20:47,480
and build reports that refresh quickly.

511
00:20:47,480 --> 00:20:50,120
Yet a stream of timestamps and machine states

512
00:20:50,120 --> 00:20:52,760
can't tell you what an event means for the schedule,

513
00:20:52,760 --> 00:20:55,000
unless it connects to the product, the process,

514
00:20:55,000 --> 00:20:56,600
and the resource involved.

515
00:20:56,600 --> 00:20:59,240
Let's look at that stop event from a planning perspective.

516
00:20:59,240 --> 00:21:03,560
A system receives a message that resource A17 entered a fault state at 1412.

517
00:21:03,560 --> 00:21:06,440
That fact matters, but it doesn't yet answer the real planning question.

518
00:21:06,440 --> 00:21:08,280
Which operation was actually running?

519
00:21:08,280 --> 00:21:09,560
Which order sat in the queue?

520
00:21:09,560 --> 00:21:11,560
Which products can use another resource?

521
00:21:11,560 --> 00:21:13,800
Does the next operation depend on a batch rule?

522
00:21:13,800 --> 00:21:15,720
Is a specialist tool installed elsewhere?

523
00:21:15,720 --> 00:21:18,520
Does moving the work break a sequence rule that protects quality

524
00:21:18,520 --> 00:21:20,120
or reduces setup time?

525
00:21:20,120 --> 00:21:21,560
The event stream gives you a signal,

526
00:21:21,560 --> 00:21:23,560
but context gives you a decision path.

527
00:21:23,560 --> 00:21:25,960
In practical terms, the common language for that context

528
00:21:25,960 --> 00:21:28,760
sits in product, process, and resource relationships.

529
00:21:28,760 --> 00:21:30,680
Product means what you're trying to produce,

530
00:21:30,680 --> 00:21:34,200
including the material, batch, variant, and specification

531
00:21:34,200 --> 00:21:36,120
where those details affect the work.

532
00:21:36,120 --> 00:21:37,960
Process means the route through production,

533
00:21:37,960 --> 00:21:39,480
which operation comes first,

534
00:21:39,480 --> 00:21:41,240
what conditions release the next step,

535
00:21:41,240 --> 00:21:43,560
where rework returns and what sequence rules apply.

536
00:21:43,560 --> 00:21:46,440
Resource means more than a machine name.

537
00:21:46,440 --> 00:21:49,000
It can include the machine, the tool, the fixture,

538
00:21:49,000 --> 00:21:51,240
the operator skill, the inspection station,

539
00:21:51,240 --> 00:21:53,640
and the time when each one can support the work.

540
00:21:53,640 --> 00:21:55,320
When you connect those relationships,

541
00:21:55,320 --> 00:21:57,800
the factory starts to describe itself in a form

542
00:21:57,800 --> 00:21:59,640
that both people and systems can use.

543
00:21:59,640 --> 00:22:01,800
Now, a production order needs a particular operation

544
00:22:01,800 --> 00:22:03,960
that operation can run on certain resources,

545
00:22:03,960 --> 00:22:05,240
understated rules,

546
00:22:05,240 --> 00:22:08,200
and those resources need tools, materials, and qualified people.

547
00:22:08,200 --> 00:22:11,080
The order also connects to a due date

548
00:22:11,080 --> 00:22:13,480
and other orders competing for the same capacity.

549
00:22:13,480 --> 00:22:15,880
So when a stop event comes in, you can route it through those links.

550
00:22:15,880 --> 00:22:18,120
Instead of asking machine A17 stopped,

551
00:22:18,120 --> 00:22:19,640
should we turn the status red,

552
00:22:19,640 --> 00:22:22,520
you can ask, which released work faces a risk?

553
00:22:22,520 --> 00:22:24,600
What alternative work can run now?

554
00:22:24,600 --> 00:22:27,400
And which commitments change if this resource stays down

555
00:22:27,400 --> 00:22:29,160
for the rest of the shift?

556
00:22:29,160 --> 00:22:30,600
That's a much better question.

557
00:22:30,600 --> 00:22:32,120
It doesn't guarantee an easy answer.

558
00:22:32,120 --> 00:22:33,400
Factories have real limits,

559
00:22:33,400 --> 00:22:35,400
but it turns the issue from a disconnected alarm

560
00:22:35,400 --> 00:22:37,880
into a planning problem you can actually reason about.

561
00:22:37,880 --> 00:22:38,680
Here's the thing,

562
00:22:38,680 --> 00:22:40,600
this doesn't require one giant model

563
00:22:40,600 --> 00:22:42,520
of the entire company before you start.

564
00:22:42,520 --> 00:22:44,680
That approach can turn into a long data program

565
00:22:44,680 --> 00:22:46,920
that doesn't help the people running today's orders.

566
00:22:46,920 --> 00:22:50,120
Instead, begin where a recurring decision depends on facts,

567
00:22:50,120 --> 00:22:51,320
scattered across systems,

568
00:22:51,320 --> 00:22:53,560
then build the relationships needed for that decision.

569
00:22:53,560 --> 00:22:54,840
The model needs to be clear enough

570
00:22:54,840 --> 00:22:57,080
that a planner, supervisor, quality engineer,

571
00:22:57,080 --> 00:23:00,120
and data team can all recognize the same production state.

572
00:23:00,120 --> 00:23:03,160
If they can't, the system will only automate confusion faster.

573
00:23:03,160 --> 00:23:05,720
Once the factory can connect an event to the work at effects,

574
00:23:05,720 --> 00:23:07,960
it can do more than just react to each disruption.

575
00:23:07,960 --> 00:23:10,760
It can start asking what each disruption

576
00:23:10,760 --> 00:23:12,360
should teach the next planning decision.

577
00:23:13,400 --> 00:23:16,360
Learn, turn events into better planning assumptions.

578
00:23:16,360 --> 00:23:20,360
Once you can connect events to the work they affected,

579
00:23:20,360 --> 00:23:22,200
the factory can start learning from the gap

580
00:23:22,200 --> 00:23:23,960
between the plan and reality.

581
00:23:23,960 --> 00:23:26,040
Not by pointing fingers after a late order,

582
00:23:26,040 --> 00:23:29,080
but by asking which assumptions keep failing in the same way.

583
00:23:29,080 --> 00:23:31,720
One late job might come from a one-off problem.

584
00:23:31,720 --> 00:23:33,400
A supplier truck breaks down,

585
00:23:33,400 --> 00:23:34,600
a tool gets damaged,

586
00:23:34,600 --> 00:23:36,760
someone enters the wrong quantity, things happen.

587
00:23:36,760 --> 00:23:38,280
But patents tell you more,

588
00:23:38,280 --> 00:23:40,200
say an operation repeatedly takes longer

589
00:23:40,200 --> 00:23:41,800
than the routing time in ERP,

590
00:23:41,800 --> 00:23:44,360
not once, but across shifts, operators, and weeks.

591
00:23:44,360 --> 00:23:45,720
The standard might still describe

592
00:23:45,720 --> 00:23:47,720
how the process ran when the product launched,

593
00:23:47,720 --> 00:23:49,800
or how long it takes under ideal conditions,

594
00:23:49,800 --> 00:23:51,640
the right material, a clean setup,

595
00:23:51,640 --> 00:23:53,160
no normal interruptions.

596
00:23:53,160 --> 00:23:56,040
But production doesn't run under ideal conditions.

597
00:23:56,040 --> 00:23:57,720
If the routing expects 30 minutes

598
00:23:57,720 --> 00:23:59,720
and the work regularly consumes 50,

599
00:23:59,720 --> 00:24:01,320
every schedule built on that routing

600
00:24:01,320 --> 00:24:03,560
starts with capacity that doesn't exist.

601
00:24:03,560 --> 00:24:05,080
The planner sees a feasible week,

602
00:24:05,080 --> 00:24:06,760
the supervisor sees a queue that grows

603
00:24:06,760 --> 00:24:08,520
before the second shift begins.

604
00:24:08,520 --> 00:24:10,520
Both people may be working from correct information

605
00:24:10,520 --> 00:24:11,560
in their own system,

606
00:24:11,560 --> 00:24:14,360
but the planning assumption no longer matches the process.

607
00:24:14,360 --> 00:24:15,640
That gap needs a response.

608
00:24:15,640 --> 00:24:17,560
Sometimes the routing time needs updating,

609
00:24:17,560 --> 00:24:19,400
sometimes the work needs a different standard

610
00:24:19,400 --> 00:24:20,680
for a certain product variant,

611
00:24:20,680 --> 00:24:22,280
tool condition, or resource.

612
00:24:22,280 --> 00:24:24,520
Sometimes the issue isn't cycle time at all.

613
00:24:24,520 --> 00:24:26,520
The operation may wait for inspection,

614
00:24:26,520 --> 00:24:27,560
material movement,

615
00:24:27,560 --> 00:24:30,200
or an approval that sits outside the formal routing.

616
00:24:30,200 --> 00:24:32,920
Learning means finding the condition behind the gap.

617
00:24:32,920 --> 00:24:35,080
Set-up loss often exposes the same issue.

618
00:24:35,080 --> 00:24:37,240
A system may reserve a fixed change over time

619
00:24:37,240 --> 00:24:38,280
for a work center,

620
00:24:38,280 --> 00:24:40,680
while the real effort depends on the sequence of products,

621
00:24:40,680 --> 00:24:42,520
the tools needed, the cleaning task,

622
00:24:42,520 --> 00:24:45,320
or the amount of adjustment after the first piece.

623
00:24:45,320 --> 00:24:47,720
If the same product transitions keep taking longer,

624
00:24:47,720 --> 00:24:49,400
that's not just a bad day on the floor.

625
00:24:49,400 --> 00:24:51,720
It's a planning fact that hasn't reached planning yet.

626
00:24:51,720 --> 00:24:54,120
Quality creates another source of repeated loss.

627
00:24:54,120 --> 00:24:56,600
A hold can look like an isolated quality event

628
00:24:56,600 --> 00:24:58,760
until you see that the same product family,

629
00:24:58,760 --> 00:25:01,320
operation, tool, or material source

630
00:25:01,320 --> 00:25:03,560
keeps creating work that waits for review.

631
00:25:03,560 --> 00:25:05,880
Then the schedule needs to account for more than expected yield.

632
00:25:05,880 --> 00:25:07,800
It needs to account for the time and capacity

633
00:25:07,800 --> 00:25:09,240
tied up while the work remains blocked.

634
00:25:09,960 --> 00:25:12,600
Resource reliability needs the same treatment.

635
00:25:12,600 --> 00:25:14,120
A machine doesn't have to fail completely

636
00:25:14,120 --> 00:25:15,640
to create planning trouble.

637
00:25:15,640 --> 00:25:18,760
A resource that regularly loses short periods of usable time

638
00:25:18,760 --> 00:25:21,800
may still show broad availability in a capacity calendar

639
00:25:21,800 --> 00:25:23,240
while the work assigned to it

640
00:25:23,240 --> 00:25:24,840
keeps slipping in practice.

641
00:25:24,840 --> 00:25:26,840
That's the difference between a resource existing

642
00:25:26,840 --> 00:25:29,080
and a resource delivering usable capacity.

643
00:25:29,080 --> 00:25:30,120
Be careful though.

644
00:25:30,120 --> 00:25:32,120
Factories contain normal variation.

645
00:25:32,120 --> 00:25:34,120
A process may take longer because one order

646
00:25:34,120 --> 00:25:36,040
includes an unusual specification.

647
00:25:36,040 --> 00:25:38,760
A setup may run late because a new operator needs support.

648
00:25:38,760 --> 00:25:41,160
A quality hold may follow a genuine rare event.

649
00:25:41,160 --> 00:25:43,240
If you change planning data after every variance,

650
00:25:43,240 --> 00:25:45,000
the system starts chasing noise.

651
00:25:45,000 --> 00:25:46,120
The question isn't,

652
00:25:46,120 --> 00:25:48,040
did actual differ from planned?

653
00:25:48,040 --> 00:25:49,000
It will.

654
00:25:49,000 --> 00:25:50,360
The real question is,

655
00:25:50,360 --> 00:25:52,280
does this difference repeat often enough

656
00:25:52,280 --> 00:25:53,800
under clear enough conditions

657
00:25:53,800 --> 00:25:55,800
that the next plan should treat it differently?

658
00:25:55,800 --> 00:25:59,000
You need people from different parts of the factory

659
00:25:59,000 --> 00:26:00,120
to answer that well.

660
00:26:00,120 --> 00:26:01,800
The planner sees the effect on orders,

661
00:26:01,800 --> 00:26:03,640
capacity, and customer commitments.

662
00:26:03,640 --> 00:26:05,240
Production sees the work as it happens

663
00:26:05,240 --> 00:26:07,720
and can explain where a standard no longer fits.

664
00:26:07,720 --> 00:26:09,560
Maintenance understands whether lost time

665
00:26:09,560 --> 00:26:10,920
points to equipment, condition,

666
00:26:10,920 --> 00:26:12,680
or ordinary operating behavior.

667
00:26:12,680 --> 00:26:14,280
Quality knows whether a hold signals

668
00:26:14,280 --> 00:26:16,280
a repeat issue or an exception.

669
00:26:16,280 --> 00:26:18,360
Engineering can determine whether the process,

670
00:26:18,360 --> 00:26:20,440
tooling, or routing itself needs revision.

671
00:26:20,440 --> 00:26:23,320
No single team owns the whole learning loop.

672
00:26:23,320 --> 00:26:25,640
That shared ownership also stops a familiar argument.

673
00:26:25,640 --> 00:26:27,800
Planning says the floor didn't follow the schedule.

674
00:26:27,800 --> 00:26:29,640
The floor says the schedule ignored the work.

675
00:26:29,640 --> 00:26:31,560
Both statements can contain part of the story

676
00:26:31,560 --> 00:26:33,000
but neither repairs the assumption

677
00:26:33,000 --> 00:26:34,840
that caused the next schedule to drift.

678
00:26:34,840 --> 00:26:37,000
A useful review stays close to the decision.

679
00:26:37,000 --> 00:26:39,320
Take a repeated plan versus actual gap.

680
00:26:39,320 --> 00:26:40,840
Trace the conditions around it,

681
00:26:40,840 --> 00:26:43,880
agree what should change and record who owns that change.

682
00:26:43,880 --> 00:26:45,880
Maybe the response updates a routing standard,

683
00:26:45,880 --> 00:26:48,440
maybe a change is how a resource gets treated in planning.

684
00:26:48,440 --> 00:26:50,040
Maybe it exposes a process issue

685
00:26:50,040 --> 00:26:53,080
that needs engineering work before anyone edits master data.

686
00:26:53,080 --> 00:26:56,120
Either way, learning needs a path back into the next decision.

687
00:26:56,120 --> 00:26:58,040
Otherwise, the factory collects lessons,

688
00:26:58,040 --> 00:26:58,920
files them away,

689
00:26:58,920 --> 00:27:01,320
and rebuilds tomorrow's schedule from the same assumptions.

690
00:27:01,320 --> 00:27:04,680
Replan is not copy the spreadsheet and try again.

691
00:27:06,120 --> 00:27:08,520
Here's the thing about learning and manufacturing.

692
00:27:08,520 --> 00:27:11,320
It only matters when it changes what you actually do next.

693
00:27:11,320 --> 00:27:12,680
When a material delay hits,

694
00:27:12,680 --> 00:27:14,600
a machine drops out, quality holds a batch,

695
00:27:14,600 --> 00:27:16,040
or demand shifts overnight.

696
00:27:16,040 --> 00:27:18,440
The real question isn't whether someone can shuffle a few rows

697
00:27:18,440 --> 00:27:19,320
in a spreadsheet.

698
00:27:19,320 --> 00:27:21,320
It's whether the factory can produce a new,

699
00:27:21,320 --> 00:27:24,040
feasible decision from the state it knows right now.

700
00:27:24,040 --> 00:27:26,520
Picture a machine that fails halfway through a shift.

701
00:27:26,520 --> 00:27:28,920
The first reaction is usually simple.

702
00:27:28,920 --> 00:27:30,680
Move the work to another machine.

703
00:27:30,680 --> 00:27:32,040
But what about the second machine?

704
00:27:32,040 --> 00:27:34,360
It might already be loaded with work that's due sooner,

705
00:27:34,360 --> 00:27:37,240
need a different fixture, or have its required tool,

706
00:27:37,240 --> 00:27:39,080
still sitting on the failed machine.

707
00:27:39,080 --> 00:27:42,600
And the person available might not even hold the qualification for that product.

708
00:27:42,600 --> 00:27:45,960
Moving an order only solves a problem if the new sequence still works.

709
00:27:45,960 --> 00:27:47,640
The same logic applies to material.

710
00:27:47,640 --> 00:27:49,640
Say a late component blocks one order.

711
00:27:49,640 --> 00:27:52,280
A planner might pull another order forward to fill the gap,

712
00:27:52,280 --> 00:27:55,000
but that replacement order needs enough released material,

713
00:27:55,000 --> 00:27:56,440
the right batch status,

714
00:27:56,440 --> 00:27:59,480
and a route that can continue after the first operation.

715
00:27:59,480 --> 00:28:01,880
Otherwise, the plan just moves the waiting point downstream

716
00:28:01,880 --> 00:28:03,240
without solving anything.

717
00:28:03,240 --> 00:28:05,880
Quality holds create an even harder version of this.

718
00:28:05,880 --> 00:28:07,640
A batch can't simply disappear from the schedule

719
00:28:07,640 --> 00:28:09,080
because inspection stopped it.

720
00:28:09,080 --> 00:28:10,760
It might need a decision from quality,

721
00:28:10,760 --> 00:28:12,520
a rework path, more machine time,

722
00:28:12,520 --> 00:28:14,120
and possibly a different inspection step

723
00:28:14,120 --> 00:28:15,560
before it can re-enter production.

724
00:28:15,560 --> 00:28:17,960
The schedule has to represent that changed condition

725
00:28:17,960 --> 00:28:20,120
instead of quietly assuming the original quantity

726
00:28:20,120 --> 00:28:21,800
will show up at the next operation.

727
00:28:21,800 --> 00:28:24,440
So re-planning starts with the constraints that remain true.

728
00:28:24,440 --> 00:28:26,120
Which resources can run this operation,

729
00:28:26,120 --> 00:28:27,880
which tools and fixtures they need,

730
00:28:27,880 --> 00:28:29,800
which operators can perform the work.

731
00:28:29,800 --> 00:28:31,480
Whether this batch must stay together,

732
00:28:31,480 --> 00:28:34,200
if a product can run after another without a cleaning step,

733
00:28:34,200 --> 00:28:36,600
and whether product rules prevent a simple resource swap.

734
00:28:36,600 --> 00:28:39,960
These details stop a schedule from becoming wishful thinking.

735
00:28:39,960 --> 00:28:42,600
An effective re-plan compares options

736
00:28:42,600 --> 00:28:45,240
before anyone releases another dispatch decision.

737
00:28:45,240 --> 00:28:47,880
One option might protect the highest priority customer order,

738
00:28:47,880 --> 00:28:50,040
but delay several lower priority orders.

739
00:28:50,040 --> 00:28:51,800
Another might keep the bottleneck loaded

740
00:28:51,800 --> 00:28:53,000
while waiting for material,

741
00:28:53,000 --> 00:28:55,160
but create a larger queue later.

742
00:28:55,160 --> 00:28:56,600
A third might require overtime

743
00:28:56,600 --> 00:28:58,200
and approved substitute material

744
00:28:58,200 --> 00:28:59,720
or a change to the customer promise.

745
00:28:59,720 --> 00:29:01,640
None of those are just scheduling choices.

746
00:29:01,640 --> 00:29:02,920
They are operational trade-offs.

747
00:29:02,920 --> 00:29:05,480
The system should show the effect in plain terms,

748
00:29:05,480 --> 00:29:07,160
which order now finishes late,

749
00:29:07,160 --> 00:29:09,720
which resource moves beyond its usable capacity,

750
00:29:09,720 --> 00:29:11,320
which work gets displaced,

751
00:29:11,320 --> 00:29:12,680
where recovery happens,

752
00:29:12,680 --> 00:29:14,520
and what assumption that recovery depends on.

753
00:29:14,520 --> 00:29:17,480
If the answer is, "We'll catch up later,"

754
00:29:17,480 --> 00:29:18,920
then it isn't an answer yet.

755
00:29:18,920 --> 00:29:21,800
This is where many spreadsheet-based revisions become fragile.

756
00:29:21,800 --> 00:29:23,800
A planner might create a sensible local sequence

757
00:29:23,800 --> 00:29:26,200
without realizing the downstream work center

758
00:29:26,200 --> 00:29:28,200
now carries more work than it can complete.

759
00:29:28,200 --> 00:29:30,200
Another person might change a priority

760
00:29:30,200 --> 00:29:32,120
without seeing the tool conflict it creates.

761
00:29:32,120 --> 00:29:34,920
The sheet can hold a revised order of work,

762
00:29:34,920 --> 00:29:36,680
but it can't reliably test every condition

763
00:29:36,680 --> 00:29:38,280
unless someone has built those conditions

764
00:29:38,280 --> 00:29:39,560
into the process around it.

765
00:29:39,560 --> 00:29:42,200
And even then, the human decision still matters.

766
00:29:42,200 --> 00:29:44,120
No system can decide by itself

767
00:29:44,120 --> 00:29:46,920
whether a strategic customer should take priority

768
00:29:46,920 --> 00:29:48,680
over a group of smaller orders.

769
00:29:48,680 --> 00:29:50,760
No algorithm owns the commercial impact

770
00:29:50,760 --> 00:29:52,360
of accepting a late delivery,

771
00:29:52,360 --> 00:29:54,120
the workforce impact of overtime

772
00:29:54,120 --> 00:29:56,440
or the process risk of a rushed recovery.

773
00:29:56,440 --> 00:29:59,560
Those decisions belong to people with the authority to make them.

774
00:29:59,560 --> 00:30:02,200
What technology can do is bring the facts, rules,

775
00:30:02,200 --> 00:30:03,640
and consequences together

776
00:30:03,640 --> 00:30:05,480
before that person approves a change.

777
00:30:05,480 --> 00:30:06,920
The planner should see the choices.

778
00:30:06,920 --> 00:30:10,040
The supervisor should confirm whether the proposed sequence can run.

779
00:30:10,040 --> 00:30:11,960
Quality and maintenance should provide the conditions

780
00:30:11,960 --> 00:30:13,640
that limit release or capacity.

781
00:30:13,640 --> 00:30:15,000
Then somebody owns the decision

782
00:30:15,000 --> 00:30:16,840
and releases a revised dispatch instruction

783
00:30:16,840 --> 00:30:18,120
that the floor can trust.

784
00:30:18,120 --> 00:30:20,120
That approval needs to leave a record,

785
00:30:20,120 --> 00:30:22,120
not as bureaucracy for its own sake,

786
00:30:22,120 --> 00:30:24,920
but because the next shift needs to know what changed,

787
00:30:24,920 --> 00:30:25,960
why it changed,

788
00:30:25,960 --> 00:30:29,160
and which assumptions now sit behind the new sequence.

789
00:30:29,160 --> 00:30:31,480
A replan is not a pretty aspred cheat.

790
00:30:31,480 --> 00:30:34,600
It is a controlled response to a changed factory state.

791
00:30:34,600 --> 00:30:35,800
And once you treat it that way,

792
00:30:35,800 --> 00:30:38,840
the five stages stop looking like separate activities.

793
00:30:38,840 --> 00:30:41,800
Planning starts the decision, execution tests it,

794
00:30:41,800 --> 00:30:43,400
measurement captures the result,

795
00:30:43,400 --> 00:30:45,000
learning updates the assumptions,

796
00:30:45,000 --> 00:30:46,920
and replanning turns the current state

797
00:30:46,920 --> 00:30:48,920
into the next workable instruction.

798
00:30:48,920 --> 00:30:50,920
The closed loop from plan to replan.

799
00:30:50,920 --> 00:30:54,440
Think of the loop as a continuous operating process,

800
00:30:54,440 --> 00:30:56,360
not a chain of software handoffs.

801
00:30:56,360 --> 00:30:57,560
It begins with planning,

802
00:30:57,560 --> 00:30:59,880
using the best facts available about demand,

803
00:30:59,880 --> 00:31:02,920
due dates, materials, routes, capacity,

804
00:31:02,920 --> 00:31:04,680
and the rules that constrain the work.

805
00:31:04,680 --> 00:31:06,680
That plan gives everyone a shared direction,

806
00:31:06,680 --> 00:31:08,600
but it should also state what it assumes

807
00:31:08,600 --> 00:31:11,080
about resource time, material readiness,

808
00:31:11,080 --> 00:31:12,840
and the order in which work can flow.

809
00:31:12,840 --> 00:31:15,800
Those assumptions need to stay visible,

810
00:31:15,800 --> 00:31:18,040
not buried in a document nobody reads.

811
00:31:18,040 --> 00:31:21,240
Then production executes against a clear dispatch decision.

812
00:31:21,240 --> 00:31:22,680
Not a vague priority list

813
00:31:22,680 --> 00:31:25,640
that leaves each shift to interpret what urgent means,

814
00:31:25,640 --> 00:31:27,240
but a practical instruction

815
00:31:27,240 --> 00:31:29,240
about what a resource should run next,

816
00:31:29,240 --> 00:31:31,080
with the conditions needed to run it.

817
00:31:31,080 --> 00:31:33,160
As work moves,

818
00:31:33,160 --> 00:31:35,640
the factory records events that change the decision.

819
00:31:35,640 --> 00:31:36,920
An operation starts,

820
00:31:36,920 --> 00:31:38,040
a job completes,

821
00:31:38,040 --> 00:31:39,000
a batch waits,

822
00:31:39,000 --> 00:31:40,760
a resource loses usable time,

823
00:31:40,760 --> 00:31:42,840
or an expected output doesn't arrive.

824
00:31:42,840 --> 00:31:45,880
The loop depends on the link between that event and the work.

825
00:31:45,880 --> 00:31:47,800
If an event cannot connect to an order,

826
00:31:47,800 --> 00:31:49,880
operation, resource, and point in time,

827
00:31:49,880 --> 00:31:52,200
it might still help maintenance or engineering,

828
00:31:52,200 --> 00:31:54,920
but it cannot reliably update a production decision.

829
00:31:54,920 --> 00:31:56,600
A schedule needs facts at the same grain

830
00:31:56,600 --> 00:31:58,520
as the choices it asks people to make.

831
00:31:58,520 --> 00:32:00,440
That means measuring at the level of order,

832
00:32:00,440 --> 00:32:02,200
operation, resource, and time,

833
00:32:02,200 --> 00:32:03,400
not at the department level.

834
00:32:03,400 --> 00:32:06,760
A daily total can tell you that a department finished less than planned,

835
00:32:06,760 --> 00:32:08,040
but it cannot tell a planner

836
00:32:08,040 --> 00:32:10,440
which operation consumed unexpected time,

837
00:32:10,440 --> 00:32:12,040
which order created a queue,

838
00:32:12,040 --> 00:32:13,800
or whether a resource can still support

839
00:32:13,800 --> 00:32:15,000
the next dispatch instruction.

840
00:32:15,000 --> 00:32:16,680
For the loop to work,

841
00:32:16,680 --> 00:32:18,920
the measurement follows the work through the factory

842
00:32:18,920 --> 00:32:20,280
at the level where capacity

843
00:32:20,280 --> 00:32:21,960
and sequence decisions actually happen.

844
00:32:22,280 --> 00:32:23,880
Those facts then support learning.

845
00:32:23,880 --> 00:32:26,040
But learning should not mean a weekly meeting

846
00:32:26,040 --> 00:32:28,680
where everyone explains why last week went badly.

847
00:32:28,680 --> 00:32:30,440
It means comparing what the plan expected

848
00:32:30,440 --> 00:32:32,520
with what execution repeatedly delivers,

849
00:32:32,520 --> 00:32:34,600
then tracing the conditions behind the gap.

850
00:32:34,600 --> 00:32:36,760
Maybe a route only works within its planned duration

851
00:32:36,760 --> 00:32:38,360
when a certain tool is ready.

852
00:32:38,360 --> 00:32:40,040
Maybe a resource loses capacity

853
00:32:40,040 --> 00:32:41,560
during a recurring handoff.

854
00:32:41,560 --> 00:32:43,160
Maybe material reaches the plant on time

855
00:32:43,160 --> 00:32:45,240
but becomes usable too late for the order sequence.

856
00:32:45,240 --> 00:32:46,760
The factory learns which conditions

857
00:32:46,760 --> 00:32:48,600
change its ability to make a promise,

858
00:32:48,600 --> 00:32:51,160
and that learning needs to update the next decision.

859
00:32:51,160 --> 00:32:52,760
Not live in a report nobody acts on.

860
00:32:52,760 --> 00:32:55,800
Replanning then starts from the current production state.

861
00:32:55,800 --> 00:32:57,640
It asks what work is complete,

862
00:32:57,640 --> 00:32:58,680
what work is blocked,

863
00:32:58,680 --> 00:33:00,200
which resources can actually run,

864
00:33:00,200 --> 00:33:01,560
which material is ready,

865
00:33:01,560 --> 00:33:03,160
and what constraints still apply.

866
00:33:03,160 --> 00:33:05,240
From there, it compares feasible choices

867
00:33:05,240 --> 00:33:07,960
and issues a new dispatch decision with a clear owner.

868
00:33:07,960 --> 00:33:09,720
The word feasible matters.

869
00:33:09,720 --> 00:33:11,320
A new schedule does not become useful

870
00:33:11,320 --> 00:33:13,640
because it improves a delivery date in a spreadsheet.

871
00:33:13,640 --> 00:33:15,400
It becomes useful when the people running the work

872
00:33:15,400 --> 00:33:18,440
can execute it without discovering hidden conflicts one hour later.

873
00:33:18,440 --> 00:33:20,200
That means the replan must respect

874
00:33:20,200 --> 00:33:22,520
the conditions learned from prior execution

875
00:33:22,520 --> 00:33:24,840
along with the conditions that exist right now.

876
00:33:24,840 --> 00:33:27,560
Over time, better assumptions feedback into planning.

877
00:33:27,560 --> 00:33:30,200
The next plan does not start from a fantasy version of capacity

878
00:33:30,200 --> 00:33:32,840
and it starts from a closer view of how the process performs

879
00:33:32,840 --> 00:33:34,360
where the recurring limits sit

880
00:33:34,360 --> 00:33:36,360
and which rules shape the order of work.

881
00:33:36,360 --> 00:33:37,960
You will still face disruptions

882
00:33:37,960 --> 00:33:40,280
and no data model can remove machine faults,

883
00:33:40,280 --> 00:33:42,440
late deliveries, or quality issues.

884
00:33:42,440 --> 00:33:43,800
But the plan gets less detached

885
00:33:43,800 --> 00:33:45,720
from the factory each time the loop closes.

886
00:33:45,720 --> 00:33:47,480
This also changes what success looks like.

887
00:33:47,480 --> 00:33:49,560
You are not trying to create one perfect schedule

888
00:33:49,560 --> 00:33:52,040
at the start of the week and defend it against reality.

889
00:33:52,040 --> 00:33:53,960
You are trying to run a controlled cycle

890
00:33:53,960 --> 00:33:56,280
where plans, facts, decisions,

891
00:33:56,280 --> 00:33:59,000
and revised instructions remain connected.

892
00:33:59,000 --> 00:34:01,640
Planning sets intent execution produces evidence,

893
00:34:01,640 --> 00:34:03,560
measurement gives that evidence a usable form,

894
00:34:03,560 --> 00:34:05,240
learning changes the assumptions

895
00:34:05,240 --> 00:34:07,960
and replanning turns the current state into action.

896
00:34:07,960 --> 00:34:10,600
That is the feedback loop most factories actually need.

897
00:34:10,600 --> 00:34:12,040
Yet many plans still operate

898
00:34:12,040 --> 00:34:13,720
through a much less connected pattern.

899
00:34:13,720 --> 00:34:16,920
The ERP releases a schedule, the floor adapts around it

900
00:34:16,920 --> 00:34:19,400
and the real decisions happen outside the systems

901
00:34:19,400 --> 00:34:21,080
meant to govern them.

902
00:34:21,080 --> 00:34:24,280
The broken loop ERP static schedule excel firefighting.

903
00:34:24,280 --> 00:34:28,520
Here is the problem most manufacturers don't talk about.

904
00:34:28,520 --> 00:34:30,520
ERP releases a schedule and that schedule

905
00:34:30,520 --> 00:34:33,240
takes on a life of its own the moment it leaves the system.

906
00:34:33,240 --> 00:34:35,880
Sometimes it gets printed and taped beside a work center.

907
00:34:35,880 --> 00:34:38,280
Sometimes someone exports it into a spreadsheet.

908
00:34:38,280 --> 00:34:41,480
Sometimes a planner emails a file to supervisors

909
00:34:41,480 --> 00:34:44,040
who turn it into a shift plan people can actually act on.

910
00:34:44,040 --> 00:34:46,280
The format changes but the pattern stays the same.

911
00:34:46,280 --> 00:34:48,040
A planning decision leaves the system

912
00:34:48,040 --> 00:34:50,280
and then the factory has to scramble to catch up.

913
00:34:50,280 --> 00:34:52,440
And the factory almost never follows it exactly.

914
00:34:52,440 --> 00:34:54,040
Material shows up later than planned.

915
00:34:54,040 --> 00:34:55,880
A machine takes longer to come back online,

916
00:34:55,880 --> 00:34:58,440
a job hits the floor with instructions nobody can follow.

917
00:34:58,440 --> 00:35:01,720
A priority customer calls so a supervisor sees a practical sequence

918
00:35:01,720 --> 00:35:05,000
that keeps work moving even though it doesn't match the released order.

919
00:35:05,000 --> 00:35:07,080
People adapt locally because they have to

920
00:35:07,080 --> 00:35:09,880
and that local adaptation is often what keeps the plant running.

921
00:35:09,880 --> 00:35:12,680
A supervisor might move an available job ahead of blocked work

922
00:35:12,680 --> 00:35:16,200
and operator switches to another task while waiting for a quality check.

923
00:35:16,200 --> 00:35:20,600
A planner protects a customer order by changing the sequence for one resource.

924
00:35:20,600 --> 00:35:22,520
Each choice makes sense in the moment,

925
00:35:22,520 --> 00:35:25,480
especially when the person making it understands current conditions better

926
00:35:25,480 --> 00:35:26,920
than the original schedule did.

927
00:35:26,920 --> 00:35:29,480
But the formal schedule doesn't always hear about the change.

928
00:35:29,480 --> 00:35:31,560
Instead the exception ends up in a spreadsheet.

929
00:35:31,560 --> 00:35:33,720
It starts as a few notes beside work orders,

930
00:35:33,720 --> 00:35:36,600
then grows into a color coded file that tracks what actually ran,

931
00:35:36,600 --> 00:35:38,920
what couldn't run, which orders need attention

932
00:35:38,920 --> 00:35:40,680
and which promises now look shaky.

933
00:35:40,680 --> 00:35:44,280
Before long the spreadsheet holds the most current version of production reality.

934
00:35:44,280 --> 00:35:46,440
Not the ERP schedule, the spreadsheet.

935
00:35:46,440 --> 00:35:48,680
That file might be maintained by a very capable person

936
00:35:48,680 --> 00:35:53,560
but it sits outside the systems that hold order, material, quality and capacity records.

937
00:35:53,560 --> 00:35:56,920
Its rules depend on formulas that only one or two people understand.

938
00:35:56,920 --> 00:35:59,000
Its latest version could live in a shared folder,

939
00:35:59,000 --> 00:36:01,160
an email thread, or on someone's desktop

940
00:36:01,160 --> 00:36:04,360
with a name like production plan final final Friday.

941
00:36:04,360 --> 00:36:07,000
Manufacturing has seen worse file names but not by much.

942
00:36:07,000 --> 00:36:09,240
The problem isn't that a spreadsheet exists.

943
00:36:09,240 --> 00:36:12,600
The problem is the production decision splits across different places.

944
00:36:12,600 --> 00:36:15,800
ERP carries the original intent, the spreadsheet carries the exceptions,

945
00:36:15,800 --> 00:36:17,640
the shop floor carries the immediate sequence.

946
00:36:17,640 --> 00:36:20,040
None of those views fully represent the other two.

947
00:36:20,040 --> 00:36:21,640
Then the revision start traveling.

948
00:36:21,640 --> 00:36:23,320
A planner emails an updated file,

949
00:36:23,320 --> 00:36:25,480
a supervisor calls another area for help.

950
00:36:25,480 --> 00:36:27,400
Someone messages about a material issue.

951
00:36:27,400 --> 00:36:30,680
At shift handover, an operator explains the next order shouldn't run yet

952
00:36:30,680 --> 00:36:32,440
even though it looks ready in the schedule.

953
00:36:32,440 --> 00:36:35,080
Another person remembers a tool needs attention

954
00:36:35,080 --> 00:36:37,560
before a certain product comes back to the machine.

955
00:36:37,560 --> 00:36:40,920
Critical scheduling knowledge becomes something people carry in their heads.

956
00:36:40,920 --> 00:36:43,400
That creates a time gap between the real factory

957
00:36:43,400 --> 00:36:45,560
and the view seen by people further from the work.

958
00:36:45,560 --> 00:36:48,840
Leadership opens a report showing late orders,

959
00:36:48,840 --> 00:36:51,480
output against plan and resource performance.

960
00:36:51,480 --> 00:36:54,600
That report may be accurate based on the latest approved data load

961
00:36:54,600 --> 00:36:57,800
but by the time it appears operators have changed the sequence twice.

962
00:36:57,800 --> 00:37:01,960
A report might show order 4-8-12 scheduled on a resource this afternoon

963
00:37:01,960 --> 00:37:03,960
while the supervisor knows the order moved,

964
00:37:03,960 --> 00:37:05,400
the material is still blocked,

965
00:37:05,400 --> 00:37:07,720
and another job now occupies that slot.

966
00:37:07,720 --> 00:37:10,200
Leadership sees a lagging picture of the plan.

967
00:37:10,200 --> 00:37:12,280
The floor deals with the live consequences.

968
00:37:12,280 --> 00:37:14,360
That leads to some frustrating meetings.

969
00:37:14,360 --> 00:37:16,360
One group asks why the schedule wasn't followed.

970
00:37:16,360 --> 00:37:19,240
Another group explains why following it would have stopped production.

971
00:37:19,240 --> 00:37:21,000
Both groups have reasonable information

972
00:37:21,000 --> 00:37:23,240
just from different moments and different systems.

973
00:37:23,240 --> 00:37:25,160
The argument sounds like a discipline issue

974
00:37:25,160 --> 00:37:27,800
but the deeper issue is that the decision path broke

975
00:37:27,800 --> 00:37:29,320
after the schedule left ERP.

976
00:37:29,320 --> 00:37:32,040
A printed schedule can't absorb a changed condition

977
00:37:32,040 --> 00:37:34,840
and exported schedule can't test whether each local adjustment

978
00:37:34,840 --> 00:37:36,680
still fits the wider production picture.

979
00:37:36,680 --> 00:37:39,880
Email notifies people but it doesn't create a shared current state

980
00:37:39,880 --> 00:37:43,240
and memory works surprisingly well right until the person holding it goes home

981
00:37:43,240 --> 00:37:46,600
so the broken loop doesn't fail because people refuse to use systems.

982
00:37:46,600 --> 00:37:48,760
It fails because the system of record

983
00:37:48,760 --> 00:37:51,880
and the system people use to react become two different things.

984
00:37:51,880 --> 00:37:53,880
The official schedule describes the plan.

985
00:37:53,880 --> 00:37:56,920
The actual production sequence emerges through exceptions, calls,

986
00:37:56,920 --> 00:37:59,400
local judgment and files trying to keep pace with the day.

987
00:37:59,400 --> 00:38:02,440
That's inefficient and sensible people keep using this pattern

988
00:38:02,440 --> 00:38:06,120
because it solves a real problem faster than the formal process does.

989
00:38:06,120 --> 00:38:07,720
Why Excel survives on the shop floor?

990
00:38:08,680 --> 00:38:10,680
So why does Excel stick around on the floor?

991
00:38:10,680 --> 00:38:12,520
Because it solves an immediate problem.

992
00:38:12,520 --> 00:38:14,520
A planner can change a sequence in seconds,

993
00:38:14,520 --> 00:38:16,200
add a note the local team understands

994
00:38:16,200 --> 00:38:19,320
and try a different order of work without waiting for an interface,

995
00:38:19,320 --> 00:38:22,120
a workflow or an overnight planning run.

996
00:38:22,120 --> 00:38:23,720
People already know how to use it

997
00:38:23,720 --> 00:38:25,400
and that matters when the shift has started.

998
00:38:25,400 --> 00:38:28,040
A customer is waiting and the person making the decision

999
00:38:28,040 --> 00:38:30,920
needs something more useful than a ticket in an IT backlog.

1000
00:38:30,920 --> 00:38:32,760
A spreadsheet handles exceptions well

1001
00:38:32,760 --> 00:38:34,520
because it doesn't ask many questions.

1002
00:38:34,520 --> 00:38:36,680
You can add a column for a tool issue, write a comment

1003
00:38:36,680 --> 00:38:40,120
about a material check, mark an order as only run after approval

1004
00:38:40,120 --> 00:38:40,760
and move on.

1005
00:38:40,760 --> 00:38:44,280
Formal systems tend to need a defined field,

1006
00:38:44,280 --> 00:38:46,840
a status, an owner and a process step

1007
00:38:46,840 --> 00:38:48,840
before they accept the same information.

1008
00:38:48,840 --> 00:38:50,200
That discipline helps later,

1009
00:38:50,200 --> 00:38:52,200
but it rarely helps when somebody needs to decide

1010
00:38:52,200 --> 00:38:53,400
what runs in the next hour.

1011
00:38:53,400 --> 00:38:55,160
There's another reason Excel hangs around.

1012
00:38:55,160 --> 00:38:57,240
It holds knowledge the formal model never captured.

1013
00:38:57,240 --> 00:38:59,880
A senior planner knows two products technically share a resource

1014
00:38:59,880 --> 00:39:01,960
but shouldn't follow each other without extra cleaning.

1015
00:39:01,960 --> 00:39:04,680
A supervisor knows one setup only runs smoothly

1016
00:39:04,680 --> 00:39:07,400
with a certain fixture and a person who has done it before.

1017
00:39:07,400 --> 00:39:10,360
Someone in the tool room knows a tool marked as available

1018
00:39:10,360 --> 00:39:12,920
is really available only after a quick inspection.

1019
00:39:12,920 --> 00:39:16,680
That knowledge lives in people, habits, notes and old spreadsheets.

1020
00:39:16,680 --> 00:39:17,640
It's very accurate,

1021
00:39:17,640 --> 00:39:20,040
but it isn't visible to everyone who needs it.

1022
00:39:20,040 --> 00:39:21,880
If a system takes too long to react,

1023
00:39:21,880 --> 00:39:23,880
people work around it and they should.

1024
00:39:23,880 --> 00:39:26,600
Production can't pause just because the approved workflow needs

1025
00:39:26,600 --> 00:39:28,520
three screens, two approvals and a sync job

1026
00:39:28,520 --> 00:39:30,200
that finishes after the shift ends.

1027
00:39:30,200 --> 00:39:32,520
The flow needs answers as fast as the work moves,

1028
00:39:32,520 --> 00:39:34,520
so I wouldn't frame Excel as the enemy.

1029
00:39:34,520 --> 00:39:35,560
That misses the point.

1030
00:39:35,560 --> 00:39:38,920
A spreadsheet often shows you where the factory has a real decision problem

1031
00:39:38,920 --> 00:39:40,600
with no formal way to handle it.

1032
00:39:40,600 --> 00:39:43,880
If planners keep maintaining a separate list of material ready orders,

1033
00:39:43,880 --> 00:39:46,040
the issue isn't that they enjoy typing into cells.

1034
00:39:46,040 --> 00:39:49,240
The issue is the official process doesn't give them material readiness

1035
00:39:49,240 --> 00:39:50,920
in a form they can trust for dispatch.

1036
00:39:50,920 --> 00:39:53,240
If supervisors maintain their own skill matrix,

1037
00:39:53,240 --> 00:39:55,080
system adoption isn't the problem.

1038
00:39:55,080 --> 00:39:59,080
The system may not describe who can run what under actual shift conditions.

1039
00:39:59,080 --> 00:40:01,480
When someone tracks setup families outside the schedule,

1040
00:40:01,480 --> 00:40:04,680
they're protecting capacity, the planning logic doesn't yet understand.

1041
00:40:04,680 --> 00:40:07,480
Excel is the symptom, missing knowledge is the condition,

1042
00:40:07,480 --> 00:40:09,000
but the risks are real.

1043
00:40:09,000 --> 00:40:12,200
Two planners can open the same file and create two different futures

1044
00:40:12,200 --> 00:40:13,400
for the same work center.

1045
00:40:13,400 --> 00:40:15,320
A supervisor acts on a printed copy

1046
00:40:15,320 --> 00:40:17,400
while another version sits in a shared folder.

1047
00:40:17,400 --> 00:40:19,480
A formula changes, a filter hides an order

1048
00:40:19,480 --> 00:40:23,240
and the person who built the logic is the only one who can explain why a job moved.

1049
00:40:23,240 --> 00:40:24,440
Then there's the audit problem.

1050
00:40:24,440 --> 00:40:25,880
When a delivery date slips,

1051
00:40:25,880 --> 00:40:27,800
you need to know who changed the sequence,

1052
00:40:27,800 --> 00:40:29,320
what information they used,

1053
00:40:29,320 --> 00:40:31,000
and what trade-off they accepted.

1054
00:40:31,000 --> 00:40:33,800
A spreadsheet can record that if the team has a disciplined process,

1055
00:40:33,800 --> 00:40:36,040
but in practice, many changes arrive through edits,

1056
00:40:36,040 --> 00:40:39,240
comments, calls, and handovers that leave no clear trail.

1057
00:40:39,240 --> 00:40:41,400
That becomes fragile when shifts change.

1058
00:40:41,400 --> 00:40:44,360
The day shift understands why an order moved behind another one.

1059
00:40:44,360 --> 00:40:47,320
The evening shift sees an order that looks late,

1060
00:40:47,320 --> 00:40:49,000
a schedule that looks outdated,

1061
00:40:49,000 --> 00:40:50,760
and maybe a note with no context.

1062
00:40:50,760 --> 00:40:53,960
People then spend time rebuilding the reasoning instead of running the work.

1063
00:40:53,960 --> 00:40:55,880
None of this means you should ban Excel.

1064
00:40:55,880 --> 00:40:58,840
Banging it usually drives the same behavior into private files,

1065
00:40:58,840 --> 00:41:01,480
paper notes and messages, and that's not progress.

1066
00:41:01,480 --> 00:41:04,120
It just makes the decision path harder to find.

1067
00:41:04,120 --> 00:41:06,920
The better goal is to absorb the knowledge Excel carries.

1068
00:41:06,920 --> 00:41:08,520
Find the recurring exceptions.

1069
00:41:08,520 --> 00:41:11,880
Listen to the planner who changes the same field every week.

1070
00:41:11,880 --> 00:41:14,760
Ask the supervisor why they ignore a sequence rule.

1071
00:41:14,760 --> 00:41:18,120
Look at the notes around the work that always creates confusion.

1072
00:41:18,120 --> 00:41:19,960
Those aren't annoying edge cases.

1073
00:41:19,960 --> 00:41:23,320
They're clues about constraints, states, and decision rules.

1074
00:41:23,320 --> 00:41:25,480
Your formal systems don't yet represent.

1075
00:41:25,480 --> 00:41:27,800
Some spreadsheet work will remain useful.

1076
00:41:27,800 --> 00:41:30,600
People need room to explore options and think through a problem.

1077
00:41:30,600 --> 00:41:33,000
But the facts behind a release production decision,

1078
00:41:33,000 --> 00:41:34,360
the rules that shape it,

1079
00:41:34,360 --> 00:41:38,520
and the reason for a change shouldn't depend on a file only one person can interpret.

1080
00:41:38,520 --> 00:41:41,480
That missing capability sits between collecting production data

1081
00:41:41,480 --> 00:41:43,160
and telling the floor what to run next.

1082
00:41:43,160 --> 00:41:45,800
The factory needs a shared way to turn current conditions

1083
00:41:45,800 --> 00:41:48,920
and practical knowledge into a decision people can execute.

1084
00:41:48,920 --> 00:41:51,800
The missing layer, production context, and decision logic.

1085
00:41:51,800 --> 00:41:54,120
So here's the real issue.

1086
00:41:54,120 --> 00:41:56,760
Excel holds all that knowledge the formal systems never catch,

1087
00:41:56,760 --> 00:41:59,320
which means you need a model that replaces that hidden layer.

1088
00:41:59,320 --> 00:42:02,760
I'm not talking about another dashboard or a bigger pile of machine data.

1089
00:42:02,760 --> 00:42:05,640
You need something that answers a direct production question.

1090
00:42:05,640 --> 00:42:09,320
What can run on which resource at what time and under what conditions?

1091
00:42:09,320 --> 00:42:12,440
That model connects the work to everything required to perform it.

1092
00:42:12,440 --> 00:42:14,760
Every production order has one or more operations

1093
00:42:14,760 --> 00:42:18,280
and each operation needs a process route, suitable machines,

1094
00:42:18,280 --> 00:42:20,120
material, a tool of fixture,

1095
00:42:20,120 --> 00:42:23,480
and sometimes a person with a specific skill or certification.

1096
00:42:23,480 --> 00:42:26,600
Those links have to exist before an exception ever happens.

1097
00:42:26,600 --> 00:42:29,560
If you try to build them after the fact, you're already behind.

1098
00:42:29,560 --> 00:42:32,040
Think about an order for a particular product variant.

1099
00:42:32,040 --> 00:42:34,600
It needs machining, then cleaning, then inspection.

1100
00:42:34,600 --> 00:42:36,840
The machining operation might run on two machines,

1101
00:42:36,840 --> 00:42:39,080
but only one has the right fixture installed right now.

1102
00:42:39,080 --> 00:42:40,520
The other machine can still run it,

1103
00:42:40,520 --> 00:42:44,040
but only after a tool change and only if a qualified operator is on shift,

1104
00:42:44,040 --> 00:42:45,400
that's production context.

1105
00:42:45,400 --> 00:42:48,280
Without it, your system sees two identical machines.

1106
00:42:48,280 --> 00:42:51,080
With it, you see one machine that can run the order now

1107
00:42:51,080 --> 00:42:53,880
and another that could run it later if several conditions come together.

1108
00:42:53,880 --> 00:42:55,800
Those are very different scheduling options.

1109
00:42:55,800 --> 00:42:57,720
Product, process, and resource relationships

1110
00:42:57,720 --> 00:42:59,960
give the factory a common language for all of this.

1111
00:42:59,960 --> 00:43:01,560
The product tells you what you're making

1112
00:43:01,560 --> 00:43:03,320
and the rules that follow that product.

1113
00:43:03,320 --> 00:43:06,040
The process gives you the route, the operation sequence,

1114
00:43:06,040 --> 00:43:08,920
and any conditions that control movement through the route.

1115
00:43:08,920 --> 00:43:11,320
And the resource tells you what can perform the work,

1116
00:43:11,320 --> 00:43:13,240
but also what that work depends on.

1117
00:43:13,240 --> 00:43:16,520
Machines, tools, fixtures, people, inspection stations,

1118
00:43:16,520 --> 00:43:17,800
and available time.

1119
00:43:17,800 --> 00:43:19,720
Now, a resource is rarely just a machine.

1120
00:43:19,720 --> 00:43:20,680
Take a furnace.

1121
00:43:20,680 --> 00:43:22,360
It may have open calendar time,

1122
00:43:22,360 --> 00:43:24,680
but it can't accept every job waiting outside.

1123
00:43:24,680 --> 00:43:26,680
Load size and temperature profiles matter.

1124
00:43:26,680 --> 00:43:28,360
Certain materials may not share load

1125
00:43:28,360 --> 00:43:31,320
and a batch may require traceability rules that prevent mixing.

1126
00:43:31,320 --> 00:43:32,520
The schedule needs those conditions.

1127
00:43:32,520 --> 00:43:34,920
If it wants to propose a load, people can actually run.

1128
00:43:34,920 --> 00:43:36,680
Without them, you're just guessing.

1129
00:43:36,680 --> 00:43:38,760
The same pattern shows up in food, chemicals,

1130
00:43:38,760 --> 00:43:41,560
medical devices, aerospace, and discrete assembly.

1131
00:43:41,560 --> 00:43:44,120
An allergen rule can force a cleaning sequence.

1132
00:43:44,120 --> 00:43:45,960
A certification rule can limit an operation

1133
00:43:45,960 --> 00:43:47,560
to certain people or resources.

1134
00:43:47,560 --> 00:43:50,520
A setup family can make one job a better next choice than another,

1135
00:43:50,520 --> 00:43:52,840
even when the second job has an earlier due date.

1136
00:43:52,840 --> 00:43:55,240
These aren't annoying exceptions around the schedule.

1137
00:43:55,240 --> 00:43:56,200
They are the schedule.

1138
00:43:56,200 --> 00:43:58,040
That changes how we think about constraints.

1139
00:43:58,040 --> 00:44:00,760
A constraint isn't just something that blocks an ideal plan.

1140
00:44:00,760 --> 00:44:02,200
It describes the real conditions

1141
00:44:02,200 --> 00:44:05,960
that make a production decision safe, legal, practical, and possible.

1142
00:44:05,960 --> 00:44:09,000
Some constraints come from the product, some from the process,

1143
00:44:09,000 --> 00:44:11,640
others from the resource, the workforce, quality rules,

1144
00:44:11,640 --> 00:44:13,800
or the order sequence already in progress.

1145
00:44:13,800 --> 00:44:17,080
The model doesn't need to make every rule mathematically complex.

1146
00:44:17,080 --> 00:44:18,840
It just needs to make them explicit enough

1147
00:44:18,840 --> 00:44:21,880
that people can see them, challenge them, and use them consistently.

1148
00:44:21,880 --> 00:44:23,640
There's another distinction that matters here.

1149
00:44:23,640 --> 00:44:25,400
Plan state versus current state.

1150
00:44:25,400 --> 00:44:28,120
Plan state describes what the schedule expects.

1151
00:44:28,120 --> 00:44:30,280
A machine to free up at a certain time,

1152
00:44:30,280 --> 00:44:33,880
a tool to be ready, an operator assignment, an available batch,

1153
00:44:33,880 --> 00:44:36,840
and a sequence of work that should fit the capacity model.

1154
00:44:36,840 --> 00:44:39,400
Current state describes what the factory can support now.

1155
00:44:39,400 --> 00:44:42,360
A machine might still be occupied, a tool might be an inspection.

1156
00:44:42,360 --> 00:44:44,120
A person may have moved to another area.

1157
00:44:44,120 --> 00:44:47,560
A batch may wait for release, the plan state can stay useful as intent,

1158
00:44:47,560 --> 00:44:49,800
but dispatch decisions need the current state.

1159
00:44:49,800 --> 00:44:53,240
Otherwise, the schedule keeps reserving capacity the plant can't use.

1160
00:44:53,240 --> 00:44:54,920
That brings us to decision logic.

1161
00:44:54,920 --> 00:44:56,760
Decision logic takes a change condition,

1162
00:44:56,760 --> 00:44:59,560
checks it against the product process and resource relationships,

1163
00:44:59,560 --> 00:45:02,040
then produces alternatives that respect the rules.

1164
00:45:02,040 --> 00:45:04,760
It can ask if this order cannot run now,

1165
00:45:04,760 --> 00:45:06,920
which other released order can run on this resource

1166
00:45:06,920 --> 00:45:10,200
without creating a material conflict, breaking a sequence rule,

1167
00:45:10,200 --> 00:45:13,320
or consuming capacity needed for a higher priority commitment.

1168
00:45:13,320 --> 00:45:16,200
That's a lot more useful than a red alert beside a machine name.

1169
00:45:16,200 --> 00:45:18,600
The result doesn't need to be one automatic answer.

1170
00:45:18,600 --> 00:45:21,800
In many cases, it should present a small number of feasible choices,

1171
00:45:21,800 --> 00:45:24,120
explain the conditions behind each and leave the trade-off

1172
00:45:24,120 --> 00:45:26,200
with the planner or supervisor who owns it.

1173
00:45:26,200 --> 00:45:29,720
You're not trying to model every thought in an experienced supervisor's head.

1174
00:45:29,720 --> 00:45:31,880
You're moving the recurring parts of that reasoning,

1175
00:45:31,880 --> 00:45:34,840
out of private memory, and into a shared decision process.

1176
00:45:34,840 --> 00:45:37,640
Once that model exists, one disruption can travel through the factory

1177
00:45:37,640 --> 00:45:38,600
in a much more useful way.

1178
00:45:38,600 --> 00:45:41,480
Let's make the whole idea concrete with a single machine stop.

1179
00:45:41,480 --> 00:45:43,720
A machine stops, and the schedule starts lying.

1180
00:45:43,720 --> 00:45:48,040
Picture a critical machine stopping halfway through an afternoon shift.

1181
00:45:48,520 --> 00:45:51,480
The operator reports a fault, maintenance starts diagnosis,

1182
00:45:51,480 --> 00:45:55,160
and the machine that carried a large share of a product family can no longer produce.

1183
00:45:55,160 --> 00:45:58,040
The first dashboard status might turn red. That part is easy.

1184
00:45:58,040 --> 00:46:00,920
The useful question begins one level later.

1185
00:46:00,920 --> 00:46:02,520
Which operations depend on that machine?

1186
00:46:02,520 --> 00:46:03,800
Which orders sit behind them?

1187
00:46:03,800 --> 00:46:07,880
And which customer commitments now depend on how long the machine stays unavailable?

1188
00:46:07,880 --> 00:46:10,360
Say the resource currently runs order 4.8.12,

1189
00:46:10,360 --> 00:46:14,440
which needs another two hours of machining before it can move to its next operation.

1190
00:46:14,440 --> 00:46:16,680
Behind it sits order 486 due tomorrow,

1191
00:46:16,680 --> 00:46:19,480
and then a smaller order for a customer shipment later in the week.

1192
00:46:19,480 --> 00:46:23,880
The schedule may show all three as planned and even continue to show expected finish times.

1193
00:46:23,880 --> 00:46:26,680
But those finish times now describe a factory that no longer exists.

1194
00:46:26,680 --> 00:46:28,840
A machine stop doesn't just remove capacity.

1195
00:46:28,840 --> 00:46:31,640
It changes the sequence of decisions around that capacity.

1196
00:46:31,640 --> 00:46:33,640
The order in progress may need to restart,

1197
00:46:33,640 --> 00:46:35,720
continue somewhere else, wait for repair,

1198
00:46:35,720 --> 00:46:39,640
or move into a recovery path that consumes time not included in the first plan.

1199
00:46:39,640 --> 00:46:40,840
And the effect moves quickly.

1200
00:46:40,840 --> 00:46:44,680
The plan needs to know whether other resources can perform the affected operation.

1201
00:46:44,680 --> 00:46:46,360
Can run the same product is not enough.

1202
00:46:46,360 --> 00:46:49,480
One alternative machine may process the product only with a certain fixture.

1203
00:46:49,480 --> 00:46:51,320
Another may need a tool currently in use,

1204
00:46:51,320 --> 00:46:53,160
and a third may have the technical capability

1205
00:46:53,160 --> 00:46:56,840
but lack an operator with the right qualification until the next shift.

1206
00:46:56,840 --> 00:47:01,240
So the list of alternative resources gets shorter as the actual conditions enter the decision.

1207
00:47:01,240 --> 00:47:04,680
Picture two machines that look interchangeable in a broad capacity model.

1208
00:47:04,680 --> 00:47:08,200
The first has the right tooling and an open window after its current job.

1209
00:47:08,200 --> 00:47:11,960
The second has a queue of work that feeds a different customer commitment.

1210
00:47:11,960 --> 00:47:15,240
Moving the failed machine order there pushes that queue into the next shift.

1211
00:47:15,240 --> 00:47:17,240
Both machines may look like capacity,

1212
00:47:17,240 --> 00:47:20,040
but only one may create a workable recovery option.

1213
00:47:20,040 --> 00:47:21,400
Even that option needs checking.

1214
00:47:21,400 --> 00:47:23,480
If the affected order moves to the first machine,

1215
00:47:23,480 --> 00:47:25,880
what happens to the job that machine was supposed to run?

1216
00:47:25,880 --> 00:47:27,000
Does it wait safely?

1217
00:47:27,000 --> 00:47:29,720
Does it need to reach a downstream process by a certain time?

1218
00:47:29,720 --> 00:47:33,960
Does postponing it create an empty period at another constrained resource later in the day?

1219
00:47:33,960 --> 00:47:38,440
This is why local rescheduling can create a second bottleneck without anyone noticing it first.

1220
00:47:38,440 --> 00:47:40,680
You solve the machine problem at one work center,

1221
00:47:40,680 --> 00:47:43,080
then move the pressure to inspection assembly,

1222
00:47:43,080 --> 00:47:46,520
coating, packing, or another process further along the route.

1223
00:47:46,520 --> 00:47:48,760
The order may finish the failed operation,

1224
00:47:48,760 --> 00:47:52,600
but it can still miss its customer date because the recovery decision created a queue

1225
00:47:52,600 --> 00:47:55,160
where the original schedule expected open capacity.

1226
00:47:55,160 --> 00:47:57,720
A red status icon cannot reason through that chain.

1227
00:47:57,720 --> 00:47:59,400
It can tell you the resource stopped,

1228
00:47:59,400 --> 00:48:01,880
but not which order should move, which should wait,

1229
00:48:01,880 --> 00:48:03,960
or which customer promise takes the least damage

1230
00:48:03,960 --> 00:48:05,720
that needs a set of feasible options.

1231
00:48:05,720 --> 00:48:07,720
One option might keep the order on the failed machine

1232
00:48:07,720 --> 00:48:11,080
and protect its queue position if maintenance expects a short recovery.

1233
00:48:11,080 --> 00:48:13,720
Another might move the order to an alternative resource,

1234
00:48:13,720 --> 00:48:15,480
accepting a slip on a different order.

1235
00:48:15,480 --> 00:48:19,560
A third might change the sequence across several resources to protect a customer date,

1236
00:48:19,560 --> 00:48:21,560
while clearly showing the displaced work.

1237
00:48:21,560 --> 00:48:24,280
None of those choices is automatically right.

1238
00:48:24,280 --> 00:48:27,080
The right one depends on the current queue, actual capability,

1239
00:48:27,080 --> 00:48:28,920
available tools, qualified people,

1240
00:48:28,920 --> 00:48:31,320
and the commitments the business has decided to protect.

1241
00:48:31,320 --> 00:48:33,960
That's the level where a planner and supervisor need support,

1242
00:48:33,960 --> 00:48:37,240
not a system that claims it can optimize the factory with a button press.

1243
00:48:37,240 --> 00:48:39,400
Factories tend to object to that sort of confidence.

1244
00:48:39,400 --> 00:48:41,640
They need a system that shows the available choices,

1245
00:48:41,640 --> 00:48:43,080
the constraints behind them,

1246
00:48:43,080 --> 00:48:47,080
and the consequence of each choice before someone releases a new instruction.

1247
00:48:47,080 --> 00:48:50,360
The machine fault then becomes a controlled scheduling question

1248
00:48:50,360 --> 00:48:52,280
rather than a chain of calls and guesses.

1249
00:48:52,280 --> 00:48:54,200
Material disruptions follow the same pattern.

1250
00:48:54,200 --> 00:48:56,280
A component can look available in a purchasing view,

1251
00:48:56,280 --> 00:48:58,520
yet still block the work that needs to run next.

1252
00:48:58,520 --> 00:49:01,160
Material availability is a scheduling constraint,

1253
00:49:01,160 --> 00:49:02,280
not a purchasing report.

1254
00:49:02,280 --> 00:49:05,240
Material sets the same trap.

1255
00:49:05,240 --> 00:49:07,560
An ERP status can look perfectly fine,

1256
00:49:07,560 --> 00:49:11,080
but that doesn't mean the material can actually support the next production operation.

1257
00:49:11,080 --> 00:49:13,080
Now here's where the disconnect shows up.

1258
00:49:13,080 --> 00:49:14,920
A buyer sees a purchase order arrived,

1259
00:49:14,920 --> 00:49:17,560
or an expected receipt falls inside the delivery window.

1260
00:49:17,560 --> 00:49:21,800
The ERP shows the quantity as on hand, allocated, or due today.

1261
00:49:21,800 --> 00:49:24,280
But none of that guarantees an operator can walk over,

1262
00:49:24,280 --> 00:49:25,800
pick the material and start the job.

1263
00:49:25,800 --> 00:49:28,520
The material has to be physically whether work happens

1264
00:49:28,520 --> 00:49:30,760
with the right quantity and quality status,

1265
00:49:30,760 --> 00:49:34,200
and it has to arrive at the point of use before the operation needs it.

1266
00:49:34,200 --> 00:49:35,160
That seems basic,

1267
00:49:35,160 --> 00:49:37,880
but the single word available hides all those conditions.

1268
00:49:37,880 --> 00:49:39,720
Take a typical scenario,

1269
00:49:39,720 --> 00:49:41,560
a work order due to start after lunch.

1270
00:49:41,560 --> 00:49:44,120
Purchasing reports the component arrived first thing in the morning,

1271
00:49:44,120 --> 00:49:45,720
so the planner sees a receipt.

1272
00:49:45,720 --> 00:49:48,040
But part of the delivery sits on the receiving dock,

1273
00:49:48,040 --> 00:49:49,640
part waits for incoming inspection,

1274
00:49:49,640 --> 00:49:52,840
and the usable quantity at the line doesn't cover the full order.

1275
00:49:52,840 --> 00:49:54,120
The component arrived,

1276
00:49:54,120 --> 00:49:56,040
but the production material didn't.

1277
00:49:56,040 --> 00:49:59,000
Partial deliveries create the same problem in a different form.

1278
00:49:59,000 --> 00:50:01,000
You have enough material to begin the order,

1279
00:50:01,000 --> 00:50:03,960
but not enough to complete the batch or feed the next operation.

1280
00:50:03,960 --> 00:50:06,680
Starting anyway, burnt setup time creates work in progress

1281
00:50:06,680 --> 00:50:08,840
and leaves half-finished product waiting for the balance.

1282
00:50:08,840 --> 00:50:10,200
Sometimes that's the right call.

1283
00:50:10,200 --> 00:50:11,160
Often it isn't.

1284
00:50:11,160 --> 00:50:13,560
A substitute material can look like an easy fix,

1285
00:50:13,560 --> 00:50:15,960
until you ask whether it has approval for the product,

1286
00:50:15,960 --> 00:50:17,400
needs a different process setting,

1287
00:50:17,400 --> 00:50:19,560
or requires separate batch trays.

1288
00:50:19,560 --> 00:50:21,960
A material planner sees an inventory alternative.

1289
00:50:21,960 --> 00:50:24,600
Quality, engineering, or the customer agreement

1290
00:50:24,600 --> 00:50:27,160
sees a restriction that stops the substitution cold.

1291
00:50:27,160 --> 00:50:28,600
Then you run into batch rules.

1292
00:50:28,600 --> 00:50:31,160
In many factories, material can't move through production

1293
00:50:31,160 --> 00:50:32,840
as one anonymous quantity.

1294
00:50:32,840 --> 00:50:35,240
A batch may need full traceability through a route,

1295
00:50:35,240 --> 00:50:37,800
a lot may be locked to a specific customer order,

1296
00:50:37,800 --> 00:50:41,000
and material can sit blocked because inspection hasn't released it,

1297
00:50:41,000 --> 00:50:44,040
shelf life changed, or a deviation needs review.

1298
00:50:44,040 --> 00:50:45,960
The schedule has to know the difference.

1299
00:50:45,960 --> 00:50:49,080
For a work order release material readiness is more than a green status

1300
00:50:49,080 --> 00:50:50,600
against a bill of material line.

1301
00:50:50,600 --> 00:50:53,160
The system or planner has to check the quantity needed now,

1302
00:50:53,160 --> 00:50:54,600
the quantity needed to finish the work,

1303
00:50:54,600 --> 00:50:57,160
the storage location, the quality release state,

1304
00:50:57,160 --> 00:50:59,880
and the timing of when the operation will consume it.

1305
00:50:59,880 --> 00:51:02,360
Consumption timing matters more than most people realize.

1306
00:51:02,360 --> 00:51:04,520
A job might need one component at the start,

1307
00:51:04,520 --> 00:51:07,400
another part way through, and packaging only near completion.

1308
00:51:07,400 --> 00:51:11,000
If the schedule treats all material as a single yes or no condition,

1309
00:51:11,000 --> 00:51:12,840
it may hold a job unnecessarily,

1310
00:51:12,840 --> 00:51:15,240
or start work that hits a preventable stop later,

1311
00:51:15,240 --> 00:51:18,040
that changes how you replan after a material problem.

1312
00:51:18,040 --> 00:51:21,400
The easy move is to pull any order forward that appears ready,

1313
00:51:21,400 --> 00:51:23,080
but that creates future starvation.

1314
00:51:23,080 --> 00:51:26,360
You could use the last available material on a lower priority order

1315
00:51:26,360 --> 00:51:29,400
and then discover the order tied to a firm customer commitment

1316
00:51:29,400 --> 00:51:32,760
can no longer run when its resource becomes free.

1317
00:51:32,760 --> 00:51:35,880
That's why the decision has to look beyond the next open slot,

1318
00:51:35,880 --> 00:51:39,400
which orders can really run with the material physically released today,

1319
00:51:39,400 --> 00:51:41,720
which material allocation protects the customer commitments

1320
00:51:41,720 --> 00:51:43,160
the business cares about most.

1321
00:51:43,160 --> 00:51:44,440
If you pull one order forward,

1322
00:51:44,440 --> 00:51:47,480
does the remaining material still support the next constraint in the route,

1323
00:51:47,480 --> 00:51:50,040
or do you simply shift the shortage to a later shift?

1324
00:51:50,040 --> 00:51:53,240
A material aware schedule can answer those questions in a practical way.

1325
00:51:53,240 --> 00:51:55,960
It doesn't need to predict every delivery perfectly.

1326
00:51:55,960 --> 00:51:58,120
It needs to distinguish confirmed physical readiness

1327
00:51:58,120 --> 00:51:59,400
from a purchasing promise,

1328
00:51:59,400 --> 00:52:02,200
and then show the consequence of each sequencing choice.

1329
00:52:02,200 --> 00:52:05,160
That gives the planner a more honest view of the customer day too.

1330
00:52:05,160 --> 00:52:07,880
A due date in ERP is a commercial commitment.

1331
00:52:07,880 --> 00:52:10,760
The factory can only meet it if the work receives usable material

1332
00:52:10,760 --> 00:52:12,760
at the right time, moves through the route,

1333
00:52:12,760 --> 00:52:15,720
and finds capacity at each required operation.

1334
00:52:15,720 --> 00:52:17,240
When material status changes,

1335
00:52:17,240 --> 00:52:18,840
the commitment risk changes with it.

1336
00:52:18,840 --> 00:52:20,760
That connection should travel both ways.

1337
00:52:20,760 --> 00:52:24,360
Production needs to see which orders carry the greatest customer exposure,

1338
00:52:24,360 --> 00:52:25,880
and commercial teams need to see

1339
00:52:25,880 --> 00:52:30,120
when a material restriction has changed what the factory can honestly promise.

1340
00:52:30,120 --> 00:52:31,960
Material isn't just a purchasing detail.

1341
00:52:31,960 --> 00:52:33,480
It shapes what work can happen next.

1342
00:52:33,480 --> 00:52:36,040
Once material can block, delay, or redirect work,

1343
00:52:36,040 --> 00:52:38,280
quality needs to enter the same feedback loop.

1344
00:52:38,280 --> 00:52:41,320
Quality events change more than yield.

1345
00:52:41,320 --> 00:52:44,920
Quality events change the schedule long before you see them in a yield number

1346
00:52:44,920 --> 00:52:46,120
at the end of a shift.

1347
00:52:46,120 --> 00:52:48,120
Take an inspection that finds a defect in a batch

1348
00:52:48,120 --> 00:52:49,720
after its first production step.

1349
00:52:49,720 --> 00:52:52,840
The affected quantity sits on hold while quality investigates.

1350
00:52:52,840 --> 00:52:54,680
That work can't move to the next operation,

1351
00:52:54,680 --> 00:52:56,360
can't count as available output,

1352
00:52:56,360 --> 00:52:59,240
and may prevent the next resource from receiving the load it expected.

1353
00:52:59,240 --> 00:53:02,760
A non-conformance changes the state of the work in a real, measurable way.

1354
00:53:02,760 --> 00:53:04,600
The batch needs a disposition decision.

1355
00:53:04,600 --> 00:53:07,320
It could be scrapped, accepted under a defined deviation,

1356
00:53:07,320 --> 00:53:10,200
sent for rework, or split into good and suspect quantity.

1357
00:53:10,200 --> 00:53:11,960
Each path changes the schedule differently

1358
00:53:11,960 --> 00:53:14,440
because each path changes both the available inventory

1359
00:53:14,440 --> 00:53:15,880
and the capacity still required.

1360
00:53:15,880 --> 00:53:19,400
Re-work is where the first schedule often loses contact with the factory.

1361
00:53:19,400 --> 00:53:22,680
A routing assumes the product moves from operation one to operation two

1362
00:53:22,680 --> 00:53:24,120
then into finished goods.

1363
00:53:24,120 --> 00:53:26,520
But a quality issue can send it back to an earlier step

1364
00:53:26,520 --> 00:53:29,240
add cleaning add inspection require a different machine

1365
00:53:29,240 --> 00:53:32,120
or hold it until an engineer approves a path forward.

1366
00:53:32,120 --> 00:53:33,400
The work returns to the factory,

1367
00:53:33,400 --> 00:53:36,200
but it returns with a new demand for time and capacity

1368
00:53:36,200 --> 00:53:38,200
and that demand competes with planned work.

1369
00:53:38,200 --> 00:53:40,280
If rework needs the same constrained resource

1370
00:53:40,280 --> 00:53:42,040
that already carries a full queue,

1371
00:53:42,040 --> 00:53:43,720
somebody has to decide what moves.

1372
00:53:43,720 --> 00:53:45,720
You can't simply add the rework order to the end

1373
00:53:45,720 --> 00:53:47,640
and assume the customer date stays intact.

1374
00:53:47,640 --> 00:53:49,720
You also can't treat the quantity as available

1375
00:53:49,720 --> 00:53:51,640
just because the first operation finished.

1376
00:53:51,640 --> 00:53:53,640
Until the batch meets its release condition,

1377
00:53:53,640 --> 00:53:56,680
it remains uncertain that release condition needs a clear definition

1378
00:53:56,680 --> 00:53:58,760
it might be an inspector approving the result,

1379
00:53:58,760 --> 00:54:00,520
quality signing off on a deviation

1380
00:54:00,520 --> 00:54:02,440
or a rework operation finishing

1381
00:54:02,440 --> 00:54:04,680
and a second inspection clearing the batch.

1382
00:54:04,680 --> 00:54:07,400
Different products and processes need different conditions

1383
00:54:07,400 --> 00:54:09,080
and planning needs access to them.

1384
00:54:09,080 --> 00:54:10,680
Otherwise the schedule sees output

1385
00:54:10,680 --> 00:54:12,280
where the factory sees blocked work.

1386
00:54:12,280 --> 00:54:14,280
Inspection lead time matters too.

1387
00:54:14,280 --> 00:54:16,360
A batch might finish an operation at 10 in the morning

1388
00:54:16,360 --> 00:54:18,600
but if inspection can't review it until the afternoon

1389
00:54:18,600 --> 00:54:20,520
the next step can't start until then

1390
00:54:20,520 --> 00:54:22,760
that waiting time becomes a hidden part of the route

1391
00:54:22,760 --> 00:54:24,600
especially where inspection capacity

1392
00:54:24,600 --> 00:54:26,520
or specialist approval limits flow.

1393
00:54:26,520 --> 00:54:29,560
You don't have to turn every quality check into a scheduling crisis.

1394
00:54:29,560 --> 00:54:31,640
Many checks happen quickly and predictably

1395
00:54:31,640 --> 00:54:34,120
so they can sit inside a normal routing assumption.

1396
00:54:34,120 --> 00:54:37,000
The problem begins when the plan treats variable release time

1397
00:54:37,000 --> 00:54:38,360
as if it were instant.

1398
00:54:38,360 --> 00:54:40,600
Then the scheduled borrows capacity from a future

1399
00:54:40,600 --> 00:54:42,120
that may never arrive.

1400
00:54:42,120 --> 00:54:43,480
The same goes for disposition.

1401
00:54:43,480 --> 00:54:45,240
If a batch sits on hold for two days

1402
00:54:45,240 --> 00:54:47,320
because the decision path is unclear,

1403
00:54:47,320 --> 00:54:49,320
that delay isn't just a quality concern.

1404
00:54:49,320 --> 00:54:50,600
It affects work in progress,

1405
00:54:50,600 --> 00:54:52,920
available stock, downstream machine loading

1406
00:54:52,920 --> 00:54:55,880
and any order that expected the batch to feed the next stage.

1407
00:54:55,880 --> 00:54:58,920
Quality status needs to travel with the production state.

1408
00:54:58,920 --> 00:55:00,360
For each relevant event

1409
00:55:00,360 --> 00:55:02,760
capture enough detail to support a real decision.

1410
00:55:02,760 --> 00:55:03,720
What failed?

1411
00:55:03,720 --> 00:55:05,480
Which product and batch it affects?

1412
00:55:05,480 --> 00:55:06,760
What quantity is blocked?

1413
00:55:06,760 --> 00:55:08,200
The planned disposition?

1414
00:55:08,200 --> 00:55:10,040
The route if rework is required?

1415
00:55:10,040 --> 00:55:12,360
The expected inspection or approval lead time

1416
00:55:12,360 --> 00:55:14,200
and the condition that releases the work

1417
00:55:14,200 --> 00:55:15,480
back into the schedule.

1418
00:55:15,480 --> 00:55:17,080
That gives the plan something usable.

1419
00:55:17,080 --> 00:55:19,720
Instead of a vague note saying quality issue

1420
00:55:19,720 --> 00:55:22,200
they can see whether the order needs another operation,

1421
00:55:22,200 --> 00:55:24,200
whether it can continue after approval,

1422
00:55:24,200 --> 00:55:25,960
whether a substitute batch exists,

1423
00:55:25,960 --> 00:55:29,080
and what extra capacity the recovery path consumes.

1424
00:55:29,080 --> 00:55:31,400
Quality still owns the technical decision.

1425
00:55:31,400 --> 00:55:34,200
Planning can then model its operational consequence.

1426
00:55:34,200 --> 00:55:37,400
Over time, those records also reveal repeated disruption.

1427
00:55:37,400 --> 00:55:39,080
You might see a certain product variant

1428
00:55:39,080 --> 00:55:41,160
repeatedly waiting for a specific inspection,

1429
00:55:41,160 --> 00:55:44,040
a tool producing defects after certain run length

1430
00:55:44,040 --> 00:55:45,400
or one process step,

1431
00:55:45,400 --> 00:55:49,320
sending work into rework far more often than the routing assumes.

1432
00:55:49,320 --> 00:55:51,640
You don't need to assign blame to see the pattern.

1433
00:55:51,640 --> 00:55:53,400
You just need to connect the quality condition

1434
00:55:53,400 --> 00:55:55,720
to the product process and resource involved.

1435
00:55:55,720 --> 00:55:57,400
That learning can change the next plan.

1436
00:55:57,400 --> 00:55:58,760
It may lead to a better process

1437
00:55:58,760 --> 00:56:00,520
standard, a different inspection assumption,

1438
00:56:00,520 --> 00:56:01,560
a maintenance action,

1439
00:56:01,560 --> 00:56:04,440
or a more realistic route for work that carries no risk.

1440
00:56:04,440 --> 00:56:06,680
The idea isn't to schedule failure as normal.

1441
00:56:06,680 --> 00:56:09,240
It's to stop pretending that repeat quality conditions

1442
00:56:09,240 --> 00:56:11,320
have no effect on capacity and dates.

1443
00:56:11,320 --> 00:56:13,240
A customer commitment only becomes credible

1444
00:56:13,240 --> 00:56:15,080
when the plan includes the conditions required

1445
00:56:15,080 --> 00:56:16,040
to release the product.

1446
00:56:16,040 --> 00:56:17,480
Material needs to be ready, yes.

1447
00:56:17,480 --> 00:56:19,000
Capacity needs to exist, yes.

1448
00:56:19,000 --> 00:56:21,240
But the work must also meet the quality conditions

1449
00:56:21,240 --> 00:56:22,920
that allow it to move through the route

1450
00:56:22,920 --> 00:56:24,520
and become usable output.

1451
00:56:24,520 --> 00:56:26,600
People and tooling create the same kind of limits

1452
00:56:26,600 --> 00:56:29,160
even when the machine calendar looks wide open.

1453
00:56:29,160 --> 00:56:32,360
Finite capacity means real resources, not calendar hours.

1454
00:56:32,360 --> 00:56:37,000
People talk about capacity like it lives inside a calendar.

1455
00:56:37,000 --> 00:56:38,600
A machine has eight hours in a shift,

1456
00:56:38,600 --> 00:56:39,880
maybe three shifts a day,

1457
00:56:39,880 --> 00:56:42,680
and the schedule treats those hours as available production time.

1458
00:56:42,680 --> 00:56:43,880
That sounds fine on paper.

1459
00:56:43,880 --> 00:56:45,400
But the factory knows better.

1460
00:56:45,400 --> 00:56:48,200
Capacity comes from a working combination of resources.

1461
00:56:48,200 --> 00:56:49,480
You need the machine, sure.

1462
00:56:49,480 --> 00:56:52,280
But you also need the right operator, the correct fixture,

1463
00:56:52,280 --> 00:56:55,640
a tool that passed inspection, material ready at the point of use,

1464
00:56:55,640 --> 00:56:59,080
and support from quality, maintenance, or internal logistics.

1465
00:56:59,080 --> 00:57:00,760
If any one of those pieces is missing,

1466
00:57:00,760 --> 00:57:02,600
the machine calendar may look wide open

1467
00:57:02,600 --> 00:57:04,840
while the operation simply can't run.

1468
00:57:04,840 --> 00:57:06,680
A work center doesn't produce on its own.

1469
00:57:06,680 --> 00:57:07,880
Let's zoom out for a second.

1470
00:57:07,880 --> 00:57:10,840
Picture a machining cell with six hours of unassigned time.

1471
00:57:10,840 --> 00:57:12,760
On paper that looks like free capacity.

1472
00:57:12,760 --> 00:57:14,680
But the job waiting for that cell needs a fixture

1473
00:57:14,680 --> 00:57:17,480
that stuck on another machine until late afternoon.

1474
00:57:17,480 --> 00:57:19,640
The operator on the next shift can run the cell,

1475
00:57:19,640 --> 00:57:21,240
but they're not certified for this product.

1476
00:57:21,240 --> 00:57:23,640
Meanwhile, the first piece inspection needs a quality technician

1477
00:57:23,640 --> 00:57:25,400
whose shift ends before setup finishes.

1478
00:57:25,400 --> 00:57:28,120
Those six hours are not usable capacity for that job,

1479
00:57:28,120 --> 00:57:29,240
not even close.

1480
00:57:29,240 --> 00:57:32,200
That's where nameplate capacity and usable capacity part company.

1481
00:57:32,200 --> 00:57:34,840
Nameplate capacity describes what a resource could produce

1482
00:57:34,840 --> 00:57:36,200
under ideal conditions.

1483
00:57:36,200 --> 00:57:38,520
Usable capacity is what the factory can actually commit

1484
00:57:38,520 --> 00:57:40,440
after you account for planned maintenance,

1485
00:57:40,440 --> 00:57:43,080
shift patterns, tool availability, staffing,

1486
00:57:43,080 --> 00:57:44,680
product rules, setup time,

1487
00:57:44,680 --> 00:57:47,800
and the ordinary operating limits that come with real production.

1488
00:57:47,800 --> 00:57:50,520
Both numbers matter, but they answer different questions.

1489
00:57:50,520 --> 00:57:52,680
Nameplate capacity helps with long-term investment

1490
00:57:52,680 --> 00:57:54,040
and broad demand planning.

1491
00:57:54,040 --> 00:57:56,840
Usable capacity needs to drive the daily production decision.

1492
00:57:56,840 --> 00:58:00,040
If the schedule fills every calendar hour based on nameplate time,

1493
00:58:00,040 --> 00:58:02,680
you start the week with no room for changeovers, checks,

1494
00:58:02,680 --> 00:58:04,280
planned maintenance, handovers,

1495
00:58:04,280 --> 00:58:06,280
or the smaller delays that come with actual work.

1496
00:58:06,280 --> 00:58:09,320
Then every disturbance looks like poor execution

1497
00:58:09,320 --> 00:58:11,320
even when the plan left no realistic space

1498
00:58:11,320 --> 00:58:12,840
for execution in the first place.

1499
00:58:12,840 --> 00:58:14,760
Maintenance windows make this really clear.

1500
00:58:14,760 --> 00:58:16,760
A machine may show is available every weekday,

1501
00:58:16,760 --> 00:58:19,800
but the maintenance team needs regular access to keep it running well.

1502
00:58:19,800 --> 00:58:22,600
That work protects the machine's ability to run later.

1503
00:58:22,600 --> 00:58:25,480
If you put it off the plan to build a prettier schedule for one day,

1504
00:58:25,480 --> 00:58:28,280
you end up with a less reliable resource for the next month.

1505
00:58:28,280 --> 00:58:30,840
Maintenance time isn't lost production by default.

1506
00:58:30,840 --> 00:58:32,680
It's part of the capacity model.

1507
00:58:32,680 --> 00:58:35,480
People bring another limit that enterprise calendars tend to flatten.

1508
00:58:35,480 --> 00:58:38,120
A person can be present, trained, and assigned to a department,

1509
00:58:38,120 --> 00:58:40,280
but that doesn't mean they can release every operation.

1510
00:58:40,280 --> 00:58:41,880
Some work needs formal certification.

1511
00:58:41,880 --> 00:58:45,000
Some needs experience with a specific process or product family.

1512
00:58:45,000 --> 00:58:47,640
Some needs two people working together for safety reasons.

1513
00:58:47,640 --> 00:58:50,520
A schedule that ignores those rules creates false choices.

1514
00:58:50,520 --> 00:58:53,000
You can assign an order to a machine that has free time,

1515
00:58:53,000 --> 00:58:55,880
then discover a dispatch that no qualified person can run it.

1516
00:58:55,880 --> 00:58:58,120
The machine stays empty, the order stays late,

1517
00:58:58,120 --> 00:59:01,000
and the schedule reports capacity that only existed in a database.

1518
00:59:01,000 --> 00:59:03,160
That's why WorkForce Data needs more than a headcount.

1519
00:59:03,160 --> 00:59:05,880
It has to describe who can perform which work,

1520
00:59:05,880 --> 00:59:08,360
under which conditions, and during which shift.

1521
00:59:08,360 --> 00:59:10,360
High Mix production adds another complication.

1522
00:59:10,360 --> 00:59:12,040
The sequence changes capacity.

1523
00:59:12,040 --> 00:59:13,800
If ten jobs need the same resource,

1524
00:59:13,800 --> 00:59:16,760
the order you run them in can change the total time required.

1525
00:59:16,760 --> 00:59:19,160
Two products might share tooling, setup settings,

1526
00:59:19,160 --> 00:59:21,240
cleaning conditions, or inspection steps.

1527
00:59:21,240 --> 00:59:24,120
Running them together can cut, change over time significantly.

1528
00:59:24,120 --> 00:59:28,040
Separating them can eat hours that a simple rooting standard never captures.

1529
00:59:28,040 --> 00:59:29,400
The opposite can happen too.

1530
00:59:29,400 --> 00:59:32,680
Two jobs might look similar, but require a full clean down between them.

1531
00:59:32,680 --> 00:59:35,400
Slap them back to back because their due dates are close together,

1532
00:59:35,400 --> 00:59:39,240
and you turn an apparently open day into one dominated by changeovers.

1533
00:59:39,240 --> 00:59:41,000
The sequence isn't just a preference,

1534
00:59:41,000 --> 00:59:44,360
it changes the capacity available to everything else on the schedule.

1535
00:59:44,360 --> 00:59:47,880
That doesn't mean every due date should bow to the most efficient setup pattern.

1536
00:59:47,880 --> 00:59:49,480
Customer commitments matter,

1537
00:59:49,480 --> 00:59:53,000
and sometimes a less efficient sequence protects the right order.

1538
00:59:53,000 --> 00:59:54,680
But the tradeoff needs to be visible.

1539
00:59:54,680 --> 00:59:56,680
You should know when you're choosing a due date priority

1540
00:59:56,680 --> 00:59:58,680
that consumes more usable capacity

1541
00:59:58,680 --> 01:00:00,760
and which other work that choice displaces.

1542
01:00:00,760 --> 01:00:02,360
Feasibility comes first.

1543
01:00:02,360 --> 01:00:05,000
You can rank orders by due date, customer priority,

1544
01:00:05,000 --> 01:00:06,920
margin, or any business rule you like,

1545
01:00:06,920 --> 01:00:08,920
but it still has to pass the physical test.

1546
01:00:08,920 --> 01:00:13,320
Can the factory actually run this work with the people, machines, tools,

1547
01:00:13,320 --> 01:00:16,840
fixtures, support steps, and time that's really available?

1548
01:00:16,840 --> 01:00:20,040
If it can't, the due date remains a request, not a production promise.

1549
01:00:20,040 --> 01:00:22,440
That distinction leads into another common mistake.

1550
01:00:22,440 --> 01:00:26,040
Many manufacturers can see the gap between plan and actual in a report.

1551
01:00:26,040 --> 01:00:29,320
Spotting the gap helps, but it doesn't tell the planner which move to make next

1552
01:00:29,320 --> 01:00:30,760
with the capacity that's left.

1553
01:00:30,760 --> 01:00:34,520
Dashboards explain yesterday, the loop must guide the next move.

1554
01:00:34,520 --> 01:00:38,280
A Power BI report can tell you a lot about production.

1555
01:00:38,280 --> 01:00:41,640
It can show plan versus actual output cues that keep growing,

1556
01:00:41,640 --> 01:00:44,760
orders at risk of being late, and overall equipment effectiveness

1557
01:00:44,760 --> 01:00:46,440
across the resources that matter.

1558
01:00:46,440 --> 01:00:50,280
That shared view helps team stop arguing over whose number is right

1559
01:00:50,280 --> 01:00:52,360
and start asking why the number changed.

1560
01:00:52,360 --> 01:00:53,320
But here's the catch.

1561
01:00:53,320 --> 01:00:56,200
A dashboard usually answers a backward-looking question.

1562
01:00:56,200 --> 01:00:58,040
What changed? Where did output fall behind?

1563
01:00:58,040 --> 01:00:59,880
Which work centre missed its planned hours?

1564
01:00:59,880 --> 01:01:01,400
Which orders now look exposed?

1565
01:01:01,400 --> 01:01:05,320
These are useful questions and plenty of factories still struggle to answer them with confidence,

1566
01:01:05,320 --> 01:01:07,800
but they don't automatically answer the next one.

1567
01:01:07,800 --> 01:01:09,240
Which sequence can still run?

1568
01:01:09,240 --> 01:01:14,360
Consider a planner staring at a report that shows a cue building in front of a bottleneck resource.

1569
01:01:14,360 --> 01:01:18,040
The same report reveals a drop in OEE, a rise in actual cycle time,

1570
01:01:18,040 --> 01:01:20,280
and several orders moving toward their due dates.

1571
01:01:20,280 --> 01:01:22,440
That tells the planner where the pressure sits.

1572
01:01:22,440 --> 01:01:25,880
It does not tell them which job should run next without breaking a material rule

1573
01:01:25,880 --> 01:01:29,320
consuming a needed tool or pushing another customer commitment into danger.

1574
01:01:29,320 --> 01:01:31,880
A dashboard describes the current condition.

1575
01:01:31,880 --> 01:01:34,760
Decision support needs to test a choice against that condition.

1576
01:01:34,760 --> 01:01:36,280
That means it needs constraints.

1577
01:01:36,280 --> 01:01:38,120
The rules that limit what can happen.

1578
01:01:38,120 --> 01:01:42,440
It needs objectives because not every order carries the same business consequence.

1579
01:01:42,440 --> 01:01:46,760
And it needs a way to compare scenarios before someone tells the floor to change direction.

1580
01:01:46,760 --> 01:01:49,560
Say you have two realistic options after a delay.

1581
01:01:49,560 --> 01:01:51,880
One sequence protects the earliest customer date,

1582
01:01:51,880 --> 01:01:53,480
but creates more setup time.

1583
01:01:53,480 --> 01:01:56,360
The other groups similar work reduces the setup burden

1584
01:01:56,360 --> 01:01:59,320
and still leaves one lower priority order late.

1585
01:01:59,320 --> 01:02:02,040
A report can show the late risk around both orders,

1586
01:02:02,040 --> 01:02:05,720
but decision support should show the consequences of choosing either path.

1587
01:02:05,720 --> 01:02:09,560
This distinction matters because manufacturers often respond to a visibility gap

1588
01:02:09,560 --> 01:02:11,080
by building another dashboard.

1589
01:02:11,080 --> 01:02:13,640
Then another appears for maintenance, another for quality,

1590
01:02:13,640 --> 01:02:17,000
another for material shortages, soon there's a report for every pain point

1591
01:02:17,000 --> 01:02:20,120
and the planner still opens Excel when the shift needs a decision.

1592
01:02:20,120 --> 01:02:21,720
More reports don't close the loop.

1593
01:02:21,720 --> 01:02:24,600
Reports should help the factory build trust in the facts.

1594
01:02:24,600 --> 01:02:27,800
They can expose where the plan and actual work diverge.

1595
01:02:27,800 --> 01:02:30,200
They can reveal where event timing is missing,

1596
01:02:30,200 --> 01:02:32,440
where resource names don't match across systems,

1597
01:02:32,440 --> 01:02:34,920
or where a status means different things to different teams.

1598
01:02:34,920 --> 01:02:36,040
That is useful work.

1599
01:02:36,040 --> 01:02:37,480
Now here's where it gets interesting.

1600
01:02:37,480 --> 01:02:41,240
If a Power BI report shows that a queue grows every time a certain product

1601
01:02:41,240 --> 01:02:44,280
family reaches inspection, the report has done its job.

1602
01:02:44,280 --> 01:02:46,440
It pointed to a condition worth investigating.

1603
01:02:46,440 --> 01:02:48,280
The next part requires a decision process

1604
01:02:48,280 --> 01:02:50,200
that can account for inspection capacity,

1605
01:02:50,200 --> 01:02:51,480
current work, priorities,

1606
01:02:51,480 --> 01:02:53,480
and the choices available to the planner.

1607
01:02:53,480 --> 01:02:56,920
Think of reporting as the place where the factory can agree on the question.

1608
01:02:56,920 --> 01:03:01,080
It should not pretend to be the place where every answer appears automatically.

1609
01:03:01,080 --> 01:03:03,080
A chart may identify the constraint,

1610
01:03:03,080 --> 01:03:05,960
but it won't necessarily decide whether to move work,

1611
01:03:05,960 --> 01:03:08,040
wait for capacity, approve overtime,

1612
01:03:08,040 --> 01:03:09,720
or change a customer promise.

1613
01:03:09,720 --> 01:03:12,600
Those actions need context, rules, and accountable people.

1614
01:03:12,600 --> 01:03:15,480
That also keeps the conversation around dashboards more honest.

1615
01:03:15,480 --> 01:03:17,880
Power BI can give production teams real-time visibility

1616
01:03:17,880 --> 01:03:20,280
when the data arrives with enough speed and meaning.

1617
01:03:20,280 --> 01:03:21,800
It can support daily management,

1618
01:03:21,800 --> 01:03:24,440
exception reviews, and plan versus actual analysis.

1619
01:03:24,440 --> 01:03:26,040
It can make hidden gaps visible.

1620
01:03:26,040 --> 01:03:28,120
Still, a dashboard doesn't become a scheduling engine

1621
01:03:28,120 --> 01:03:29,800
just because it uses live data.

1622
01:03:29,800 --> 01:03:31,160
The factory needs both.

1623
01:03:31,160 --> 01:03:32,920
It needs reporting that people trust,

1624
01:03:32,920 --> 01:03:34,920
and it needs a path from reported conditions

1625
01:03:34,920 --> 01:03:37,000
to test it, approve production decisions.

1626
01:03:37,000 --> 01:03:40,200
Without that path, the dashboard becomes a very polished way

1627
01:03:40,200 --> 01:03:42,520
to watch the same firefighting happen again and again.

1628
01:03:42,520 --> 01:03:44,200
So now let's connect this to Microsoft,

1629
01:03:44,200 --> 01:03:45,720
but keep the roles clear.

1630
01:03:45,720 --> 01:03:47,640
The data foundation can bring ERP,

1631
01:03:47,640 --> 01:03:51,480
MES, quality, maintenance, and operational data into a shared place.

1632
01:03:51,480 --> 01:03:52,520
That helps the loop.

1633
01:03:52,520 --> 01:03:54,920
It doesn't replace the factory knowledge and decision logic

1634
01:03:54,920 --> 01:03:57,640
that turn data into the next executable instruction.

1635
01:03:57,640 --> 01:03:59,800
Microsoft fabric as the data foundation,

1636
01:03:59,800 --> 01:04:01,800
not the factory brain.

1637
01:04:01,800 --> 01:04:04,680
Microsoft fabric gives you that shared data foundation.

1638
01:04:04,680 --> 01:04:07,480
It pulls together data from ERP, MES,

1639
01:04:07,480 --> 01:04:09,320
quality systems, maintenance records,

1640
01:04:09,320 --> 01:04:12,200
and IoT sources into govern data products.

1641
01:04:12,200 --> 01:04:14,520
So your teams work from the same set of facts

1642
01:04:14,520 --> 01:04:16,680
instead of building a new extract for every question.

1643
01:04:16,680 --> 01:04:18,600
Here's why that matters.

1644
01:04:18,600 --> 01:04:20,280
The loop crosses system boundaries.

1645
01:04:20,280 --> 01:04:21,800
A plan might start in ERP

1646
01:04:21,800 --> 01:04:23,800
while the actual work comes from MES

1647
01:04:23,800 --> 01:04:26,040
and a quality hold might sit in the quality system

1648
01:04:26,040 --> 01:04:27,880
while a maintenance event sits somewhere else

1649
01:04:27,880 --> 01:04:30,120
as a machine signal arrives through IoT.

1650
01:04:30,120 --> 01:04:33,720
If each team only sees their own slides,

1651
01:04:33,720 --> 01:04:35,480
nobody can trace a change condition

1652
01:04:35,480 --> 01:04:37,080
all the way to a production decision.

1653
01:04:37,080 --> 01:04:39,640
Fabric gives those sources a common place to meet.

1654
01:04:39,640 --> 01:04:41,800
Getting the data in one place is only half the job.

1655
01:04:41,800 --> 01:04:43,720
The records need common identifiers.

1656
01:04:43,720 --> 01:04:45,240
So the production order in ERP

1657
01:04:45,240 --> 01:04:47,320
actually links to the order reported by MES,

1658
01:04:47,320 --> 01:04:49,000
the machine identifier in maintenance

1659
01:04:49,000 --> 01:04:50,840
connects to the resource used in planning

1660
01:04:50,840 --> 01:04:53,080
and material lots, work centers,

1661
01:04:53,080 --> 01:04:54,920
product codes, and shift definitions

1662
01:04:54,920 --> 01:04:56,760
all need rules people can trust.

1663
01:04:56,760 --> 01:04:57,800
Without those links,

1664
01:04:57,800 --> 01:04:59,320
you just have more data in one place

1665
01:04:59,320 --> 01:05:00,760
and the same old disagreement.

1666
01:05:00,760 --> 01:05:02,600
Time logic needs the same attention.

1667
01:05:02,600 --> 01:05:05,240
An event time, a system posting time,

1668
01:05:05,240 --> 01:05:08,040
and the time a person enters the status can all differ,

1669
01:05:08,040 --> 01:05:09,880
but that doesn't mean any of them is wrong.

1670
01:05:09,880 --> 01:05:11,640
They describe different moments.

1671
01:05:11,640 --> 01:05:13,960
When you compare plans start against actual start,

1672
01:05:13,960 --> 01:05:16,360
you have to define which time counts as actual start

1673
01:05:16,360 --> 01:05:18,280
and apply that definition consistently

1674
01:05:18,280 --> 01:05:20,520
because otherwise plan versus actual analysis

1675
01:05:20,520 --> 01:05:22,680
becomes an argument about timestamps.

1676
01:05:22,680 --> 01:05:24,040
Lineage matters too.

1677
01:05:24,040 --> 01:05:27,320
If a planner asks why an order now shows a delay risk,

1678
01:05:27,320 --> 01:05:29,800
the answer should trace back through the data.

1679
01:05:29,800 --> 01:05:33,000
A confirmed machine stop, a revised material receipt,

1680
01:05:33,000 --> 01:05:36,120
an MES event that marked an operation incomplete

1681
01:05:36,120 --> 01:05:38,760
or a manually entered quality hold.

1682
01:05:38,760 --> 01:05:40,680
People need to know where the fact came from,

1683
01:05:40,680 --> 01:05:43,560
when it arrived, and whether it changed after the first load.

1684
01:05:43,560 --> 01:05:45,400
Share business definitions help with words

1685
01:05:45,400 --> 01:05:48,440
that sound simple until several systems use them.

1686
01:05:48,440 --> 01:05:51,080
Complete might mean a machine finished processing

1687
01:05:51,080 --> 01:05:53,080
or it might mean quality released the batch

1688
01:05:53,080 --> 01:05:55,240
or that the next operation accepted it,

1689
01:05:55,240 --> 01:05:58,440
and available might mean material exists in stock

1690
01:05:58,440 --> 01:06:01,000
or that production can consume it now.

1691
01:06:01,000 --> 01:06:02,840
Fabric can support governed definitions

1692
01:06:02,840 --> 01:06:05,000
so the same term doesn't quietly mean different things

1693
01:06:05,000 --> 01:06:06,440
in different reports and decisions.

1694
01:06:06,440 --> 01:06:07,960
Once those foundations exist,

1695
01:06:07,960 --> 01:06:10,200
you can build a history of plan versus actual

1696
01:06:10,200 --> 01:06:11,800
that goes beyond a report showing

1697
01:06:11,800 --> 01:06:13,800
where the last month met plan does.

1698
01:06:13,800 --> 01:06:16,120
It becomes a record of what the schedule expected,

1699
01:06:16,120 --> 01:06:17,720
what execution produced,

1700
01:06:17,720 --> 01:06:19,080
which conditions changed,

1701
01:06:19,080 --> 01:06:20,920
and how the decision changed in response.

1702
01:06:21,880 --> 01:06:24,280
That history gives learning somewhere solid to live.

1703
01:06:24,280 --> 01:06:26,520
You can examine repeated gaps between planned

1704
01:06:26,520 --> 01:06:27,880
and actual cycle time,

1705
01:06:27,880 --> 01:06:30,440
trace where orders waited longer than expected,

1706
01:06:30,440 --> 01:06:32,520
compare the original dispatch sequence

1707
01:06:32,520 --> 01:06:33,880
with what actually ran,

1708
01:06:33,880 --> 01:06:36,280
and look for patterns around material release,

1709
01:06:36,280 --> 01:06:38,520
quality status, resource loss,

1710
01:06:38,520 --> 01:06:40,200
or recurring replans.

1711
01:06:40,200 --> 01:06:43,000
The point isn't to build a data lake full of factory regret,

1712
01:06:43,000 --> 01:06:45,160
it's to give planners, production teams,

1713
01:06:45,160 --> 01:06:47,000
and engineers a dependable history

1714
01:06:47,000 --> 01:06:48,680
they can use to test assumptions.

1715
01:06:48,680 --> 01:06:50,200
If a routing time keeps missing

1716
01:06:50,200 --> 01:06:51,800
under a known set of conditions,

1717
01:06:51,800 --> 01:06:53,000
that should become visible.

1718
01:06:53,000 --> 01:06:55,080
If a schedule repeatedly depends on workarounds

1719
01:06:55,080 --> 01:06:56,600
that never reached the formal record,

1720
01:06:56,600 --> 01:06:58,440
the data should expose that gap too.

1721
01:06:58,440 --> 01:07:00,760
Still, I'd keep the role of fabric very clear.

1722
01:07:00,760 --> 01:07:02,280
Fabric supports data access,

1723
01:07:02,280 --> 01:07:04,440
governance analysis, and shared data products,

1724
01:07:04,440 --> 01:07:06,440
but it does not automatically understand a factory

1725
01:07:06,440 --> 01:07:08,520
just because you loaded ERP tables,

1726
01:07:08,520 --> 01:07:11,880
MES events, and sensor readings into the same platform.

1727
01:07:11,880 --> 01:07:13,880
They can't infer every capability rule,

1728
01:07:13,880 --> 01:07:15,960
sequence restriction, tool requirement,

1729
01:07:15,960 --> 01:07:18,120
or approval condition that experienced people use

1730
01:07:18,120 --> 01:07:20,120
when they decide what should run next.

1731
01:07:20,120 --> 01:07:21,960
Those relationships need explicit modeling.

1732
01:07:21,960 --> 01:07:24,600
You need to describe which product operations can run,

1733
01:07:24,600 --> 01:07:26,920
on which resources, which conditions affect release,

1734
01:07:26,920 --> 01:07:29,160
and which rules apply when the state changes.

1735
01:07:29,160 --> 01:07:31,800
Some of that information may already exist in source systems,

1736
01:07:31,800 --> 01:07:33,320
while some lives in documents,

1737
01:07:33,320 --> 01:07:34,840
work instructions, spreadsheets,

1738
01:07:34,840 --> 01:07:37,240
or the heads of people who know the process well.

1739
01:07:37,240 --> 01:07:39,000
Fabric can hold and connect the data,

1740
01:07:39,000 --> 01:07:41,240
but your manufacturing model gives it meaning.

1741
01:07:41,240 --> 01:07:43,720
That distinction protects you from a common mistake.

1742
01:07:43,720 --> 01:07:46,120
A well-governed data platform can give you a trusted view

1743
01:07:46,120 --> 01:07:48,120
of the factory, but it does not replace

1744
01:07:48,120 --> 01:07:51,160
MES execution, ERP ownership of commercial demand,

1745
01:07:51,160 --> 01:07:54,440
or the decision logic needed to create a feasible revised schedule.

1746
01:07:54,440 --> 01:07:56,040
Each part has a job.

1747
01:07:56,040 --> 01:07:58,840
With that in mind, let's follow the data from operational technology

1748
01:07:58,840 --> 01:08:00,520
through the decision loop and keep the boundary

1749
01:08:00,520 --> 01:08:04,520
between machine control and production intelligence clear.

1750
01:08:04,520 --> 01:08:08,520
Azure and OT move data without pretending the cloud runs the machine.

1751
01:08:08,520 --> 01:08:11,960
The data foundation needs facts from the factory,

1752
01:08:11,960 --> 01:08:14,680
but that does not mean the cloud should take over machine control.

1753
01:08:14,680 --> 01:08:18,600
Programmable logic controllers, or PLCs, remain close to the equipment

1754
01:08:18,600 --> 01:08:20,920
because they handle control tasks where timing,

1755
01:08:20,920 --> 01:08:23,160
safety, and predictable behavior matter.

1756
01:08:23,160 --> 01:08:25,160
And an edge system may collect signals,

1757
01:08:25,160 --> 01:08:27,560
normalize protocols, buffer events,

1758
01:08:27,560 --> 01:08:30,040
or connect older equipment to newer systems

1759
01:08:30,040 --> 01:08:32,920
that work belongs near the machine for good reasons.

1760
01:08:32,920 --> 01:08:35,640
A cloud service should not sit in the path of a safety stop.

1761
01:08:35,640 --> 01:08:38,840
If a network link drops, the machine still needs to stop safely,

1762
01:08:38,840 --> 01:08:41,720
run its local control logic and keep operators protected,

1763
01:08:41,720 --> 01:08:44,440
and if a cloud service takes longer than expected to respond,

1764
01:08:44,440 --> 01:08:47,320
the production cell cannot simply wait and hope the connection recovers.

1765
01:08:47,320 --> 01:08:51,000
Physical production has different failure modes than a web application,

1766
01:08:51,000 --> 01:08:53,400
and that boundary needs to stay clear from the beginning.

1767
01:08:53,400 --> 01:08:56,760
Now here's how factory data actually moves into the wider feedback loop.

1768
01:08:56,760 --> 01:08:59,080
It happens through approved OT pathways,

1769
01:08:59,080 --> 01:09:01,240
with agreement from the people who own the equipment

1770
01:09:01,240 --> 01:09:02,760
and the people who own the data.

1771
01:09:02,760 --> 01:09:06,040
And depending on the plant, that may mean an industrial gateway,

1772
01:09:06,040 --> 01:09:08,440
an edge platform, an MS integration,

1773
01:09:08,440 --> 01:09:11,480
or a controlled interface from an existing historian.

1774
01:09:11,480 --> 01:09:13,160
The route matters as much as the event.

1775
01:09:13,160 --> 01:09:15,560
OT teams need to know what connects to the network,

1776
01:09:15,560 --> 01:09:17,160
which systems can read data,

1777
01:09:17,160 --> 01:09:18,680
who can change a configuration,

1778
01:09:18,680 --> 01:09:20,200
and what happens during a fault,

1779
01:09:20,200 --> 01:09:23,080
while IT teams need identity, access control,

1780
01:09:23,080 --> 01:09:26,920
audit records, and a way to move trusted data into the data platform.

1781
01:09:26,920 --> 01:09:29,320
Neither side can treat the other as an obstacle.

1782
01:09:29,320 --> 01:09:32,040
They are protecting different parts of the same production system.

1783
01:09:32,040 --> 01:09:35,160
A useful design assumes connectivity will fail sometimes.

1784
01:09:35,160 --> 01:09:37,400
A plant may have weak coverage in one area,

1785
01:09:37,400 --> 01:09:40,360
a machine network may sit behind strict security controls

1786
01:09:40,360 --> 01:09:44,040
or an edge device may need to store events locally until it can reconnect.

1787
01:09:44,040 --> 01:09:45,160
Clocks can drift,

1788
01:09:45,160 --> 01:09:47,480
and a message can arrive late, arrive twice,

1789
01:09:47,480 --> 01:09:50,040
or arrive after the production state has changed again.

1790
01:09:50,040 --> 01:09:51,880
Those are normal engineering conditions.

1791
01:09:51,880 --> 01:09:54,280
The system needs to record enough time and source detail

1792
01:09:54,280 --> 01:09:55,640
to deal with these conditions,

1793
01:09:55,640 --> 01:09:57,720
and it also needs to distinguish a live signal

1794
01:09:57,720 --> 01:09:59,720
from a confirmed production event.

1795
01:09:59,720 --> 01:10:02,600
A machine may indicate that it is running, for example,

1796
01:10:02,600 --> 01:10:04,360
but that does not automatically prove

1797
01:10:04,360 --> 01:10:08,200
a specific order produced acceptable output during that time.

1798
01:10:08,200 --> 01:10:11,080
The production context still matters.

1799
01:10:11,080 --> 01:10:13,160
Security boundaries deserve the same care.

1800
01:10:13,160 --> 01:10:15,400
Reading a machine status and writing a machine command

1801
01:10:15,400 --> 01:10:16,840
are completely different permissions,

1802
01:10:16,840 --> 01:10:20,520
so a service that supports planning should not gain the right to alter a PLC

1803
01:10:20,520 --> 01:10:23,080
just because both systems connect through Azure.

1804
01:10:23,080 --> 01:10:24,920
Keep control actions local, controlled,

1805
01:10:24,920 --> 01:10:27,640
and owned by the operational teams responsible for the process

1806
01:10:27,640 --> 01:10:30,520
and cloud connectivity should support awareness and decisions

1807
01:10:30,520 --> 01:10:32,360
not bypass plant governance.

1808
01:10:32,360 --> 01:10:34,760
This is also where people collect far too much data

1809
01:10:34,760 --> 01:10:36,520
without a clear question behind it.

1810
01:10:36,520 --> 01:10:39,720
A machine can produce a huge stream of tags, counters, states,

1811
01:10:39,720 --> 01:10:41,880
temperatures, alarms, and diagnostic values,

1812
01:10:41,880 --> 01:10:43,880
some of those signals matter for maintenance,

1813
01:10:43,880 --> 01:10:45,240
others matter for quality,

1814
01:10:45,240 --> 01:10:47,880
and a smaller set may affect a production dispatch decision.

1815
01:10:47,880 --> 01:10:51,480
Send the event with the meaning needed upstream,

1816
01:10:51,480 --> 01:10:54,120
rather than pushing every raw change into a central platform

1817
01:10:54,120 --> 01:10:56,040
and asking someone to find a use later,

1818
01:10:56,040 --> 01:10:58,520
define the events the feedback loop actually needs.

1819
01:10:58,520 --> 01:11:00,760
A resource entered an unavailable state,

1820
01:11:00,760 --> 01:11:03,720
an operation began, a count changed a tool,

1821
01:11:03,720 --> 01:11:05,640
reached a condition that blocks use,

1822
01:11:05,640 --> 01:11:08,520
or a batch completed a process step by the way it's released.

1823
01:11:08,520 --> 01:11:11,960
Each event should carry the right context where possible,

1824
01:11:11,960 --> 01:11:13,320
which resource produced it,

1825
01:11:13,320 --> 01:11:15,480
which order or operation it relates to,

1826
01:11:15,480 --> 01:11:17,560
when it occurred, which source reported it,

1827
01:11:17,560 --> 01:11:19,480
and whether it's a machine observation,

1828
01:11:19,480 --> 01:11:21,160
an MES confirmed event,

1829
01:11:21,160 --> 01:11:23,560
or a manual status entered by an operator.

1830
01:11:23,560 --> 01:11:25,560
Those details let the rest of the architecture

1831
01:11:25,560 --> 01:11:27,880
treat the event as evidence instead of noise.

1832
01:11:27,880 --> 01:11:31,240
That doesn't mean every source system needs to become perfect before you start.

1833
01:11:31,240 --> 01:11:33,640
It means you should know the limits of each signal

1834
01:11:33,640 --> 01:11:37,240
and avoid turning a technical data feed into a false production claim.

1835
01:11:37,240 --> 01:11:41,480
OT and IT convergence works when both sides contribute to a shared decision process

1836
01:11:41,480 --> 01:11:43,240
while keeping their own responsibilities.

1837
01:11:43,240 --> 01:11:45,720
OT protects safe, reliable operation of equipment.

1838
01:11:45,720 --> 01:11:48,280
IT provides governed data, identity, integration,

1839
01:11:48,280 --> 01:11:50,760
and access across business systems and production

1840
01:11:50,760 --> 01:11:54,280
owns the decision about what work should run under the current conditions.

1841
01:11:54,280 --> 01:11:56,360
No single team owns the whole loop alone,

1842
01:11:56,360 --> 01:11:59,960
as you can provide services for ingesting, processing, securing,

1843
01:11:59,960 --> 01:12:02,360
and connecting operational data with enterprise data.

1844
01:12:02,360 --> 01:12:04,760
But the design still needs plant-approved pathways,

1845
01:12:04,760 --> 01:12:08,920
clear ownership, and a firm line between observing a machine and controlling it.

1846
01:12:08,920 --> 01:12:11,960
Identity can tell you that several systems refer to the same machine,

1847
01:12:11,960 --> 01:12:14,120
order, or material lot, which is helpful,

1848
01:12:14,120 --> 01:12:16,920
but relationships tell you why a change in one of those things

1849
01:12:16,920 --> 01:12:18,440
should alter a production decision.

1850
01:12:18,440 --> 01:12:22,680
Digital Twin and Knowledge Graph two ways to model factory meaning.

1851
01:12:22,680 --> 01:12:26,280
Once you can match the same order, machine, material,

1852
01:12:26,280 --> 01:12:28,120
lot, and operation across different systems,

1853
01:12:28,120 --> 01:12:30,200
you can start building meaning out of those pieces.

1854
01:12:31,000 --> 01:12:33,480
Two concepts come up a lot in that conversation.

1855
01:12:33,480 --> 01:12:37,560
Digital Twin and Knowledge Graph people sometimes throw them around like they're interchangeable.

1856
01:12:37,560 --> 01:12:41,800
They're not. They can work together, but they answer completely different questions about production.

1857
01:12:41,800 --> 01:12:45,560
A digital twin is a live digital snapshot of something physical on the floor,

1858
01:12:45,560 --> 01:12:50,600
a machine, a cell, a line, a tool, or even a batch moving through a process.

1859
01:12:50,600 --> 01:12:51,720
Take a machining center.

1860
01:12:51,720 --> 01:12:54,680
Its twin holds the state that matters for production right now,

1861
01:12:54,680 --> 01:13:00,120
available, running, stopped, under maintenance, waiting for a tool or blocked by something else.

1862
01:13:00,120 --> 01:13:02,920
It can also connect that state to things like capability,

1863
01:13:02,920 --> 01:13:06,440
current setup, maintenance status, and whatever work is assigned to it at the moment.

1864
01:13:06,440 --> 01:13:09,640
The entire focus is the state of the asset right now.

1865
01:13:09,640 --> 01:13:13,560
That makes a digital twin useful when you need to monitor something physical

1866
01:13:13,560 --> 01:13:16,600
and understand its current condition in operational terms.

1867
01:13:16,600 --> 01:13:19,000
Is the resource ready for dispatch? Is it occupied?

1868
01:13:19,000 --> 01:13:21,720
Did a maintenance event cut into its usable time?

1869
01:13:21,720 --> 01:13:24,440
Is the line producing, waiting, or stopped?

1870
01:13:24,440 --> 01:13:28,360
A twin gives that status structured home instead of leaving it buried across machine tags,

1871
01:13:28,360 --> 01:13:31,560
MES statuses, maintenance logs, and sticky notes.

1872
01:13:31,560 --> 01:13:33,960
A knowledge graph starts from a completely different angle.

1873
01:13:33,960 --> 01:13:35,720
It models relationships between things.

1874
01:13:35,720 --> 01:13:38,280
For production planning, those things include products,

1875
01:13:38,280 --> 01:13:41,720
routes, operations, work orders, machines, tools, people,

1876
01:13:41,720 --> 01:13:44,360
material lots, quality conditions, and business rules.

1877
01:13:44,360 --> 01:13:46,120
The graph records how they connect.

1878
01:13:46,120 --> 01:13:47,640
An order needs an operation.

1879
01:13:47,640 --> 01:13:49,720
That operation can run on certain resources,

1880
01:13:49,720 --> 01:13:51,880
and each resource might need a specific tool.

1881
01:13:51,880 --> 01:13:53,960
A tool may only support one product family.

1882
01:13:53,960 --> 01:13:56,600
A person needs a certification to run a certain process.

1883
01:13:56,600 --> 01:13:59,640
A quality hold blocks a batch from moving to the next operation.

1884
01:13:59,640 --> 01:14:03,080
That network of relationships is what a knowledge graph captures.

1885
01:14:03,080 --> 01:14:05,880
It becomes useful when your question involves dependencies.

1886
01:14:05,880 --> 01:14:07,640
Imagine a machine goes down unexpectedly.

1887
01:14:07,640 --> 01:14:10,680
With a knowledge graph, you can trace the work affected by that change.

1888
01:14:10,680 --> 01:14:12,360
Which current orders depend on it?

1889
01:14:12,360 --> 01:14:15,080
Which jobs can move to an alternative resource?

1890
01:14:15,080 --> 01:14:18,120
What if that alternative also needs a tool already in use?

1891
01:14:18,120 --> 01:14:20,120
Which downstream commitments are at risk?

1892
01:14:20,120 --> 01:14:21,240
If nothing changes.

1893
01:14:21,240 --> 01:14:23,160
You aren't just looking at the machine anymore.

1894
01:14:23,160 --> 01:14:24,680
You're following the links around it.

1895
01:14:24,680 --> 01:14:26,760
That's where the two models complement each other.

1896
01:14:26,760 --> 01:14:28,600
The digital twin holds a current state.

1897
01:14:28,600 --> 01:14:30,520
Say, a machine becomes unavailable.

1898
01:14:30,520 --> 01:14:32,920
The knowledge graph helps you trace what that state affects

1899
01:14:32,920 --> 01:14:35,320
across the product, process, and resource model.

1900
01:14:35,320 --> 01:14:37,240
But I'd be careful not to treat either term

1901
01:14:37,240 --> 01:14:39,640
as a replacement for the systems you already use.

1902
01:14:39,640 --> 01:14:42,680
An MES still owns and records execution activity.

1903
01:14:42,680 --> 01:14:45,160
ERP still carries demand, commercial commitments,

1904
01:14:45,160 --> 01:14:46,600
and core business records.

1905
01:14:46,600 --> 01:14:49,400
A scheduling tool, whether it sits inside a broader platform

1906
01:14:49,400 --> 01:14:51,240
or comes from a specialist provider,

1907
01:14:51,240 --> 01:14:53,800
still needs to calculate and manage the sequence of work.

1908
01:14:53,800 --> 01:14:56,120
A digital twin doesn't replace an MES

1909
01:14:56,120 --> 01:14:58,040
because it only describes current state.

1910
01:14:58,040 --> 01:15:00,440
A knowledge graph doesn't replace scheduling software

1911
01:15:00,440 --> 01:15:03,080
because it maps relationships instead of sequencing work.

1912
01:15:03,080 --> 01:15:05,240
Both can give those systems better context

1913
01:15:05,240 --> 01:15:07,880
and both can support the decisions that happen between them.

1914
01:15:07,880 --> 01:15:11,160
The model you choose should follow the problem you're trying to solve.

1915
01:15:11,160 --> 01:15:13,640
If you need a clear picture of operational state,

1916
01:15:13,640 --> 01:15:15,160
start with a digital twin approach.

1917
01:15:15,160 --> 01:15:17,800
You want to know whether a bottleneck resource can take work,

1918
01:15:17,800 --> 01:15:21,000
what state it's in, and how that state changes through the shift.

1919
01:15:21,000 --> 01:15:23,560
If you need to trace dependencies, investigate impact,

1920
01:15:23,560 --> 01:15:25,560
or reason across rules and relationships,

1921
01:15:25,560 --> 01:15:27,240
a knowledge graph might fit better.

1922
01:15:27,240 --> 01:15:28,840
It can answer questions that look simple

1923
01:15:28,840 --> 01:15:30,760
until the factory enters the conversation.

1924
01:15:30,760 --> 01:15:33,080
Like, what else changes if we move this order?

1925
01:15:33,080 --> 01:15:35,880
You don't need to model the entire enterprise to begin.

1926
01:15:35,880 --> 01:15:39,240
Pick a bottleneck where planners already spend time chasing updates

1927
01:15:39,240 --> 01:15:41,880
or choose one product family with complex resource rules,

1928
01:15:41,880 --> 01:15:44,520
tooling dependencies, and recurring schedule changes.

1929
01:15:44,520 --> 01:15:46,520
Model the resource's operations, orders,

1930
01:15:46,520 --> 01:15:48,920
and conditions around that small area first

1931
01:15:48,920 --> 01:15:50,360
that gives you a practical test.

1932
01:15:50,360 --> 01:15:52,600
Can the model explain why a job cannot run?

1933
01:15:52,600 --> 01:15:55,400
Can it trace the impact of a change resource state?

1934
01:15:55,400 --> 01:15:56,840
Can it show feasible alternatives

1935
01:15:56,840 --> 01:15:59,720
that match how experience production people think about the work?

1936
01:15:59,720 --> 01:16:02,600
If it can't, adding every other factory asset won't help.

1937
01:16:02,600 --> 01:16:05,560
Once the feedback loop has current facts and explicit constraints,

1938
01:16:05,560 --> 01:16:08,440
AI becomes a much more grounded discussion.

1939
01:16:08,440 --> 01:16:11,240
Before that point, it mostly has fragments of the factory

1940
01:16:11,240 --> 01:16:13,720
and a confident way of arranging words around them.

1941
01:16:13,720 --> 01:16:17,080
Industrial AI, where it helps, and where it should not decide.

1942
01:16:18,360 --> 01:16:21,480
Industrial AI becomes useful when it works on a specific production question

1943
01:16:21,480 --> 01:16:24,360
with facts you can trace and constraints your team agrees on.

1944
01:16:24,360 --> 01:16:26,120
Without that, it can sound convincing

1945
01:16:26,120 --> 01:16:28,600
while recommending work that cannot actually run.

1946
01:16:28,600 --> 01:16:29,960
Take duration forecasting.

1947
01:16:29,960 --> 01:16:32,600
If an operation has taken longer than its routing standard

1948
01:16:32,600 --> 01:16:33,880
under certain conditions,

1949
01:16:33,880 --> 01:16:38,040
a model can estimate a more realistic finish time for work in progress.

1950
01:16:38,040 --> 01:16:40,200
It can account for patents in actual cycle time,

1951
01:16:40,200 --> 01:16:42,040
product type, resource state,

1952
01:16:42,040 --> 01:16:43,960
and maybe a known setup condition.

1953
01:16:43,960 --> 01:16:45,480
If those facts exist in the data,

1954
01:16:45,480 --> 01:16:47,800
that doesn't replace the standard time in the route.

1955
01:16:47,800 --> 01:16:49,320
It gives the planner a better warning

1956
01:16:49,320 --> 01:16:51,400
when current work is likely to miss it.

1957
01:16:51,400 --> 01:16:55,240
AI can also flag schedule risk before the order becomes officially late.

1958
01:16:55,240 --> 01:16:56,920
Suppose the remaining work needs more time

1959
01:16:56,920 --> 01:16:59,080
than the open capacity left in its route,

1960
01:16:59,080 --> 01:17:00,920
or the cycle time on a bottleneck resource

1961
01:17:00,920 --> 01:17:03,720
starts drifting above the normal range for that product family.

1962
01:17:03,720 --> 01:17:05,480
The system can point to the risk early

1963
01:17:05,480 --> 01:17:07,160
while the team still has choices.

1964
01:17:07,160 --> 01:17:09,480
Detecting abnormal patterns works the same way.

1965
01:17:09,480 --> 01:17:10,760
A machine might keep running,

1966
01:17:10,760 --> 01:17:13,000
so a simple availability status stays green,

1967
01:17:13,000 --> 01:17:15,000
but its cycle time begins to stretch.

1968
01:17:15,000 --> 01:17:17,640
A process might produce more small stops than usual.

1969
01:17:17,640 --> 01:17:20,520
An operation might enter inspection later than similar orders.

1970
01:17:20,520 --> 01:17:22,760
None of those events alone proves a problem,

1971
01:17:22,760 --> 01:17:24,280
but the pattern may deserve attention

1972
01:17:24,280 --> 01:17:27,320
before the schedule absorbs the delay as though nothing changed.

1973
01:17:27,320 --> 01:17:29,240
That's a sensible use of industrial AI.

1974
01:17:29,240 --> 01:17:30,840
It watches for conditions people miss

1975
01:17:30,840 --> 01:17:32,360
when they are busy running production.

1976
01:17:32,360 --> 01:17:34,120
Generative AI has a different role.

1977
01:17:34,120 --> 01:17:36,360
It can help people ask questions in normal language

1978
01:17:36,360 --> 01:17:38,600
and retrieve governed operational context.

1979
01:17:38,600 --> 01:17:41,560
A planner might ask why a customer order now faces risk,

1980
01:17:41,560 --> 01:17:44,200
and the response can bring together the linked facts.

1981
01:17:44,200 --> 01:17:46,200
An operation ran longer than expected,

1982
01:17:46,200 --> 01:17:48,760
an alternative resource lacks the required tool,

1983
01:17:48,760 --> 01:17:51,800
and the remaining route includes a queue at a constraint process.

1984
01:17:51,800 --> 01:17:53,320
The answer still needs sources.

1985
01:17:53,320 --> 01:17:56,120
A useful response should tell the planner which event,

1986
01:17:56,120 --> 01:17:59,400
record, rule and time stamp, support the explanation.

1987
01:17:59,400 --> 01:18:02,280
If it cannot do that, it may be a well-written guess.

1988
01:18:02,280 --> 01:18:04,360
Production teams don't need more guesses,

1989
01:18:04,360 --> 01:18:06,360
even if the guess uses perfect grammar.

1990
01:18:06,360 --> 01:18:09,480
Generative AI can also help explain the effect of a proposed change.

1991
01:18:09,480 --> 01:18:13,240
If someone asks what happens when order 4812 moves ahead of another order,

1992
01:18:13,240 --> 01:18:15,320
the assistant can retrieve the affected route,

1993
01:18:15,320 --> 01:18:18,440
current resource conditions, and stated business priorities,

1994
01:18:18,440 --> 01:18:21,320
then explain the likely impact in plain language.

1995
01:18:21,320 --> 01:18:23,800
But explanation and decision are different jobs.

1996
01:18:23,800 --> 01:18:26,040
Constraint sequencing needs a different kind of engine.

1997
01:18:26,040 --> 01:18:29,560
An optimization engine works through defined choices, rules and objectives.

1998
01:18:29,560 --> 01:18:32,040
It can test sequences against finite capacity,

1999
01:18:32,040 --> 01:18:35,000
material conditions, tool limits, sequence restrictions,

2000
01:18:35,000 --> 01:18:36,280
and due date priorities.

2001
01:18:36,280 --> 01:18:38,360
It can compare one feasible schedule against another

2002
01:18:38,360 --> 01:18:39,480
and show the trade-offs.

2003
01:18:39,480 --> 01:18:43,160
Generative AI doesn't replace that work simply because it can describe a schedule.

2004
01:18:43,160 --> 01:18:44,200
Think of it this way.

2005
01:18:44,200 --> 01:18:46,920
One system may explain why a recommendation exists.

2006
01:18:46,920 --> 01:18:49,640
Another may calculate which recommendations remain feasible

2007
01:18:49,640 --> 01:18:51,480
when the factory has hard constraints.

2008
01:18:51,480 --> 01:18:53,240
In some designs, they work together.

2009
01:18:53,240 --> 01:18:55,640
The optimization engine produces tested options.

2010
01:18:55,640 --> 01:18:58,520
While generative AI helps a planner understand the reasoning

2011
01:18:58,520 --> 01:19:00,920
and ask follow-up questions, that split is healthy.

2012
01:19:00,920 --> 01:19:03,960
A production recommendation should show more than a final answer.

2013
01:19:03,960 --> 01:19:05,160
It needs traceability.

2014
01:19:05,160 --> 01:19:06,760
Which current facts did it use?

2015
01:19:06,760 --> 01:19:08,280
Which rules removed other options?

2016
01:19:08,280 --> 01:19:10,680
Which assumption shaped the projected finish time?

2017
01:19:10,680 --> 01:19:12,040
When did the data last update?

2018
01:19:12,040 --> 01:19:13,640
And if uncertainty exists?

2019
01:19:13,640 --> 01:19:16,200
How confident is the system in its forecast?

2020
01:19:16,200 --> 01:19:20,040
Confidence does not mean the system gets to hide doubt behind a percentage.

2021
01:19:20,040 --> 01:19:23,400
It means the planner can see when the recommendation rests on thin evidence

2022
01:19:23,400 --> 01:19:25,880
missing data or a condition that changed recently.

2023
01:19:25,880 --> 01:19:27,640
Human review remains part of the design.

2024
01:19:27,640 --> 01:19:31,080
A system may identify the sequence that protects the most due dates,

2025
01:19:31,080 --> 01:19:33,080
but a supervisor may know a practical condition

2026
01:19:33,080 --> 01:19:34,680
that hasn't reached the model yet.

2027
01:19:34,680 --> 01:19:38,200
A planner may accept one late order to protect a customer relationship.

2028
01:19:38,200 --> 01:19:41,080
Quality may block a route that looked available an hour ago.

2029
01:19:41,080 --> 01:19:42,600
Those are operational choices,

2030
01:19:42,600 --> 01:19:44,120
not defects in the system.

2031
01:19:44,120 --> 01:19:47,480
So before anyone gives AI authority over dispatch decisions,

2032
01:19:47,480 --> 01:19:48,760
ask a blunt question.

2033
01:19:48,760 --> 01:19:49,880
What does it actually know?

2034
01:19:49,880 --> 01:19:52,680
Does it know the current order state or only yesterday's plan?

2035
01:19:52,680 --> 01:19:54,440
Does it know the product and process rules?

2036
01:19:54,440 --> 01:19:56,760
Does it know which resources can perform the operation?

2037
01:19:56,760 --> 01:19:59,640
Which tools are available and which conditions prevent release?

2038
01:19:59,640 --> 01:20:02,040
Does it know the business priority behind the due date?

2039
01:20:02,040 --> 01:20:03,320
If the answer is unclear,

2040
01:20:03,320 --> 01:20:05,400
the AI should advise, not decide.

2041
01:20:05,400 --> 01:20:06,680
The technology matters,

2042
01:20:06,680 --> 01:20:08,600
but the next problem sits with people,

2043
01:20:08,600 --> 01:20:09,480
authority,

2044
01:20:09,480 --> 01:20:12,200
and how the factory acts when the recommendation arrives.

2045
01:20:12,200 --> 01:20:14,600
The planner's role changes, it doesn't disappear.

2046
01:20:14,600 --> 01:20:17,560
Here's the question I keep hearing.

2047
01:20:17,560 --> 01:20:21,160
With all this buzz about better scheduling, data models, and AI in planning,

2048
01:20:21,160 --> 01:20:22,760
does the planner's job go away?

2049
01:20:22,760 --> 01:20:24,200
That's the wrong question to ask.

2050
01:20:24,200 --> 01:20:27,240
A working feedback loop gets rid of the worst part of planning.

2051
01:20:27,240 --> 01:20:29,640
The hours spent hunting for facts across ERP,

2052
01:20:29,640 --> 01:20:31,880
MES, emails, phone calls,

2053
01:20:31,880 --> 01:20:34,760
and that spreadsheet someone updated during the night shift.

2054
01:20:34,760 --> 01:20:36,840
Instead, the planner's focus moves to the decisions

2055
01:20:36,840 --> 01:20:38,760
that actually need human judgment.

2056
01:20:38,760 --> 01:20:40,760
You stop manually piecing together

2057
01:20:40,760 --> 01:20:43,240
the current state of the factory from fragments

2058
01:20:43,240 --> 01:20:45,800
and start dealing with exceptions and trade-offs.

2059
01:20:45,800 --> 01:20:47,640
Which customer date to protect,

2060
01:20:47,640 --> 01:20:49,400
whether overtime makes sense

2061
01:20:49,400 --> 01:20:52,040
if a recovery option introduces too much risk,

2062
01:20:52,040 --> 01:20:55,480
or if the data is complete enough to release a new sequence?

2063
01:20:55,480 --> 01:20:56,440
That's not less planning,

2064
01:20:56,440 --> 01:20:58,200
that it's a smarter use of planning skill.

2065
01:20:58,200 --> 01:21:01,240
A good planner knows a schedule is full of conflicting promises.

2066
01:21:01,240 --> 01:21:02,680
The customer wants delivery,

2067
01:21:02,680 --> 01:21:03,880
production needs a sequence,

2068
01:21:03,880 --> 01:21:05,160
people can actually run,

2069
01:21:05,160 --> 01:21:07,880
maintenance needs time to keep equipment dependable,

2070
01:21:07,880 --> 01:21:10,200
quality needs the right release conditions,

2071
01:21:10,200 --> 01:21:13,240
and finance cares about inventory and expediting costs.

2072
01:21:13,240 --> 01:21:15,240
Software can test the rules it knows,

2073
01:21:15,240 --> 01:21:17,720
but the planner decides which trade-off the business

2074
01:21:17,720 --> 01:21:19,240
is actually willing to accept.

2075
01:21:19,240 --> 01:21:20,280
And the supervisor,

2076
01:21:20,280 --> 01:21:21,800
there's still very much in the loop.

2077
01:21:21,800 --> 01:21:23,560
A supervisor sees conditions

2078
01:21:23,560 --> 01:21:25,640
that might be real, but not yet formal.

2079
01:21:25,640 --> 01:21:29,080
A changeover team stuck longer than expected.

2080
01:21:29,080 --> 01:21:30,520
An operator who knows a job

2081
01:21:30,520 --> 01:21:31,560
will need extra attention

2082
01:21:31,560 --> 01:21:33,640
because the last batch caused trouble,

2083
01:21:33,640 --> 01:21:36,760
a machine that's physically blocked by material handling.

2084
01:21:36,760 --> 01:21:38,200
That knowledge should not stay

2085
01:21:38,200 --> 01:21:40,760
as a private correction after the schedule reaches the floor.

2086
01:21:40,760 --> 01:21:42,360
The supervisor needs a practical way

2087
01:21:42,360 --> 01:21:44,360
to feed that condition into the decision

2088
01:21:44,360 --> 01:21:47,000
and then confirm whether the revised dispatch instruction

2089
01:21:47,000 --> 01:21:48,200
can actually run.

2090
01:21:48,200 --> 01:21:49,320
That confirmation matters

2091
01:21:49,320 --> 01:21:51,880
because a schedule can be technically feasible in a system

2092
01:21:51,880 --> 01:21:53,560
but still fail on a busy shift

2093
01:21:53,560 --> 01:21:55,960
if it ignores something visible on the floor.

2094
01:21:55,960 --> 01:21:59,480
Maintenance and quality play similar roles from different angles.

2095
01:21:59,480 --> 01:22:02,360
Maintenance changes capacity when a resource becomes uncertain,

2096
01:22:02,360 --> 01:22:04,200
when a planned intervention needs protection

2097
01:22:04,200 --> 01:22:06,200
or when a repair estimate changes.

2098
01:22:06,200 --> 01:22:08,200
Quality changes release when material,

2099
01:22:08,200 --> 01:22:10,280
work in progress or finished output,

2100
01:22:10,280 --> 01:22:13,160
enters a hold, needs rework or waits for approval.

2101
01:22:13,160 --> 01:22:15,320
Those teams don't need to become schedulers.

2102
01:22:15,320 --> 01:22:16,840
They just need to provide the conditions

2103
01:22:16,840 --> 01:22:18,280
that change the schedule clearly enough

2104
01:22:18,280 --> 01:22:19,640
so planning can act on them.

2105
01:22:19,640 --> 01:22:21,400
This brings us to clear authority.

2106
01:22:21,400 --> 01:22:23,640
Who decides that a machine loss requires

2107
01:22:23,640 --> 01:22:25,640
re-planning rather than local recovery?

2108
01:22:25,640 --> 01:22:28,440
Who decides when a material issue changes the dispatch order

2109
01:22:28,440 --> 01:22:31,080
who can accept the impact of moving a lower priority order

2110
01:22:31,080 --> 01:22:32,360
behind a more urgent one?

2111
01:22:32,360 --> 01:22:33,880
If nobody owns those calls,

2112
01:22:33,880 --> 01:22:35,560
the plant still replants.

2113
01:22:35,560 --> 01:22:37,880
But through side conversations and personal influence,

2114
01:22:37,880 --> 01:22:39,480
the loop needs name decision rights.

2115
01:22:39,480 --> 01:22:42,360
For example, a supervisor might handle minor sequence swaps

2116
01:22:42,360 --> 01:22:43,640
within a defined window,

2117
01:22:43,640 --> 01:22:47,080
a planner owns changes affecting the next shift or customer exposure.

2118
01:22:47,080 --> 01:22:49,640
Quality controls whether held work can re-enter,

2119
01:22:49,640 --> 01:22:52,280
maintenance provides resource status and repair outlook

2120
01:22:52,280 --> 01:22:55,880
and production decides how the capacity loss changes the plan.

2121
01:22:55,880 --> 01:22:57,320
The exact split depends on the plant

2122
01:22:57,320 --> 01:22:58,840
but ambiguity doesn't help.

2123
01:22:58,840 --> 01:23:01,080
Trust also determines whether people use the system

2124
01:23:01,080 --> 01:23:02,440
or find ways around it.

2125
01:23:02,440 --> 01:23:05,640
A recommendation that says move order 4-8-12 to machine-b

2126
01:23:05,640 --> 01:23:08,200
invites resistance because people immediately ask questions

2127
01:23:08,200 --> 01:23:10,520
the system hasn't answered why this order,

2128
01:23:10,520 --> 01:23:12,440
why that machine, which work moves out,

2129
01:23:12,440 --> 01:23:14,920
what assumption did it make about the tool operator material

2130
01:23:14,920 --> 01:23:15,960
and repair time?

2131
01:23:15,960 --> 01:23:18,760
A recommendation earns trust when it explains itself,

2132
01:23:18,760 --> 01:23:20,040
showing the facts it used,

2133
01:23:20,040 --> 01:23:21,560
the rules that shaped the option

2134
01:23:21,560 --> 01:23:23,320
and the consequence of approving it.

2135
01:23:23,320 --> 01:23:25,240
And it should state uncertainty directly.

2136
01:23:25,240 --> 01:23:27,320
If the repair estimate is still unconfirmed,

2137
01:23:27,320 --> 01:23:29,160
the recommendation should say so.

2138
01:23:29,160 --> 01:23:31,640
Pretending certainty around incomplete shop floor facts

2139
01:23:31,640 --> 01:23:34,520
is one of the fastest ways to send everyone back to Excel

2140
01:23:34,520 --> 01:23:37,480
that same explanation gives experienced people a way to correct the model.

2141
01:23:37,480 --> 01:23:40,600
If a supervisor rejects a proposal because a fixture is unavailable,

2142
01:23:40,600 --> 01:23:42,120
that becomes a useful question.

2143
01:23:42,120 --> 01:23:43,640
Was the fixture state missing?

2144
01:23:43,640 --> 01:23:44,600
Was the rule absent?

2145
01:23:44,600 --> 01:23:47,240
Or did the condition change after the data arrived?

2146
01:23:47,240 --> 01:23:48,680
Over time, the system improves

2147
01:23:48,680 --> 01:23:52,040
because the people doing the work can see where its reasoning falls short.

2148
01:23:52,040 --> 01:23:54,040
Which leads to the next practical question,

2149
01:23:54,040 --> 01:23:58,040
how often should this loop act without turning every small event into a new schedule?

2150
01:23:58,040 --> 01:24:00,440
Event driven re-planning without constant schedule?

2151
01:24:00,440 --> 01:24:01,400
Chaos.

2152
01:24:01,400 --> 01:24:06,120
A working loop doesn't mean the schedule changes every time a sensor sends a message.

2153
01:24:06,120 --> 01:24:09,080
If it did, the floor would stop trusting the schedule altogether.

2154
01:24:09,080 --> 01:24:10,760
Machines produce normal noise,

2155
01:24:10,760 --> 01:24:12,520
a short stop that clears in minutes,

2156
01:24:12,520 --> 01:24:14,440
an operator correcting a minor issue,

2157
01:24:14,440 --> 01:24:15,960
a cycle running slightly longer,

2158
01:24:15,960 --> 01:24:17,880
without affecting any customer commitment

2159
01:24:17,880 --> 01:24:19,320
or blocking the next operation.

2160
01:24:19,320 --> 01:24:22,040
Re-planning leads a reason, not just an event.

2161
01:24:22,040 --> 01:24:25,240
The plant needs trigger rules to decide when an operational change

2162
01:24:25,240 --> 01:24:26,760
becomes a scheduling review.

2163
01:24:26,760 --> 01:24:31,000
A machine stop might trigger that review only after it passes a defined duration

2164
01:24:31,000 --> 01:24:35,720
or once maintenance confirms recovery will extend beyond the available slack in the current plan.

2165
01:24:35,720 --> 01:24:39,720
A material event might trigger it when a required component enters a quality block

2166
01:24:39,720 --> 01:24:42,760
when the usable quantity falls below what an operation needs

2167
01:24:42,760 --> 01:24:46,760
or when a confirmed delivery moves past the time when work must start.

2168
01:24:46,760 --> 01:24:48,200
Quality needs its own triggers,

2169
01:24:48,200 --> 01:24:50,840
a small inspection delay stays inside the normal flow

2170
01:24:50,840 --> 01:24:54,360
but a hole that blocks a batch feeding a constrained resource needs action immediately

2171
01:24:54,360 --> 01:24:56,600
because the effect reaches multiple orders.

2172
01:24:56,600 --> 01:24:58,440
Demand can also trigger a review,

2173
01:24:58,440 --> 01:25:01,400
a customer escalation, a change shipment priority

2174
01:25:01,400 --> 01:25:03,880
or a commercial decision to protect one order

2175
01:25:03,880 --> 01:25:05,160
can alter the sequence.

2176
01:25:05,160 --> 01:25:08,520
That should not happen through an informal phone call and a quiet change in Excel.

2177
01:25:08,520 --> 01:25:10,680
The change needs a visible decision path.

2178
01:25:10,680 --> 01:25:12,440
Trigger rules don't remove judgment.

2179
01:25:12,440 --> 01:25:13,880
They give it a starting point.

2180
01:25:13,880 --> 01:25:16,120
You might define a threshold for a resource loss,

2181
01:25:16,120 --> 01:25:19,960
then let the planner and supervisor decide whether the current queue can absorb it.

2182
01:25:19,960 --> 01:25:22,040
You might flag a material shortage automatically,

2183
01:25:22,040 --> 01:25:25,960
then check whether an approved substitute or another released batch removes the problem.

2184
01:25:25,960 --> 01:25:28,040
The rule says stop and review this.

2185
01:25:28,040 --> 01:25:31,080
It does not claim the answer is already known.

2186
01:25:31,080 --> 01:25:34,120
That distinction protects the factory from two bad extremes.

2187
01:25:34,120 --> 01:25:38,760
One plant waits too long because nobody knows when a disturbance becomes serious enough to raise,

2188
01:25:38,760 --> 01:25:41,800
another rebuilds the entire schedule for every small change.

2189
01:25:41,800 --> 01:25:45,720
So people stop following the latest version because a newer one is always coming.

2190
01:25:45,720 --> 01:25:47,640
Neither approach gives production control.

2191
01:25:47,640 --> 01:25:49,160
Freeze Windows help create stability.

2192
01:25:49,160 --> 01:25:52,760
Close to execution, the schedule should only change for real operational reasons,

2193
01:25:52,760 --> 01:25:55,400
not because a model found a slightly tidier sequence.

2194
01:25:55,400 --> 01:25:57,160
Think about the next few hours.

2195
01:25:57,160 --> 01:26:00,280
Operators have stage material, a tool sits at the machine,

2196
01:26:00,280 --> 01:26:03,480
a changeover is underway, someone has started first piece checks.

2197
01:26:03,480 --> 01:26:07,960
Moving that work around to gain a small improvement on paper can create more disruption than it removes.

2198
01:26:07,960 --> 01:26:09,480
Near term work needs protection.

2199
01:26:09,480 --> 01:26:11,800
The size of the freeze window depends on the process.

2200
01:26:11,800 --> 01:26:15,480
A high volume line with fastest batch decisions might use a shorter window,

2201
01:26:15,480 --> 01:26:18,360
a complex process with long setups, batch rules,

2202
01:26:18,360 --> 01:26:22,360
or strict material handling needs more stability around the work already prepared.

2203
01:26:22,360 --> 01:26:24,040
The point is not to freeze the factory.

2204
01:26:24,040 --> 01:26:27,080
It's to stop the schedule from behaving like a nervous spreadsheet.

2205
01:26:27,080 --> 01:26:29,800
Outside that protected period the replant needs a horizon.

2206
01:26:29,800 --> 01:26:31,640
If you only look at the next open slot,

2207
01:26:31,640 --> 01:26:34,920
you might recover one delayed order and create pressure later in the route.

2208
01:26:34,920 --> 01:26:38,360
If you try to rebuild every order for the next month after a small stop,

2209
01:26:38,360 --> 01:26:40,760
you create work without improving the decision.

2210
01:26:40,760 --> 01:26:43,080
The horizon should match the consequence of the event.

2211
01:26:43,080 --> 01:26:47,960
A brief issue on a non-constrained resource might only need a local review through the end of the shift.

2212
01:26:47,960 --> 01:26:50,920
Lots of a bottleneck machine, a blocked material batch,

2213
01:26:50,920 --> 01:26:54,680
or a changed customer priority may require a broader look across the next few days

2214
01:26:54,680 --> 01:26:56,680
and through the affected downstream operations.

2215
01:26:56,680 --> 01:27:00,200
That's how the factory balances local recovery with the wider production flow.

2216
01:27:00,200 --> 01:27:02,360
You don't need a global answer for every event.

2217
01:27:02,360 --> 01:27:06,520
You need enough view to avoid solving today's problem by quietly creating tomorrow's.

2218
01:27:06,520 --> 01:27:08,600
Every approved change should leave a record.

2219
01:27:08,600 --> 01:27:10,040
Which schedule version changed?

2220
01:27:10,040 --> 01:27:11,720
What event triggered the review?

2221
01:27:11,720 --> 01:27:13,240
Which option did the team select?

2222
01:27:13,240 --> 01:27:14,040
Who approved it?

2223
01:27:14,040 --> 01:27:16,120
What assumptions shaped the decision?

2224
01:27:16,120 --> 01:27:19,400
Like an expected repair time or a planned material receipt?

2225
01:27:19,400 --> 01:27:21,320
This isn't paperwork for its own sake.

2226
01:27:21,320 --> 01:27:24,600
When the next shift asks why its dispatch list changed,

2227
01:27:24,600 --> 01:27:27,880
the answer should not depend on finding the person who took the phone call.

2228
01:27:27,880 --> 01:27:29,480
The record also supports learning later.

2229
01:27:29,480 --> 01:27:32,680
If the factory repeatedly changes the sequence for the same reason,

2230
01:27:32,680 --> 01:27:34,440
that pattern becomes visible.

2231
01:27:34,440 --> 01:27:37,720
If a decision worked poorly because a repair estimate changed

2232
01:27:37,720 --> 01:27:39,400
or a material promise failed,

2233
01:27:39,400 --> 01:27:42,920
people can examine the assumption rather than argue from memory.

2234
01:27:42,920 --> 01:27:45,640
A scheduled version is a production decision with a history

2235
01:27:45,640 --> 01:27:49,560
and you don't need to build this across every plant line, product and order at once.

2236
01:27:49,560 --> 01:27:50,920
Start with a smaller scope,

2237
01:27:50,920 --> 01:27:52,840
where replanting already hurts enough

2238
01:27:52,840 --> 01:27:54,920
that people feel the gap every day.

2239
01:27:54,920 --> 01:27:57,160
Start with one decision that hurts today.

2240
01:27:57,160 --> 01:28:01,240
Here's the problem most manufacturers don't talk about.

2241
01:28:01,240 --> 01:28:05,800
Don't kick off with a factory wide program called closed loop scheduling.

2242
01:28:05,800 --> 01:28:08,120
That name sounds great in a steering meeting,

2243
01:28:08,120 --> 01:28:10,600
but when the bottleneck sequence shifts at 10 in the morning,

2244
01:28:10,600 --> 01:28:13,160
it doesn't give anyone on the shop floor a better answer.

2245
01:28:13,160 --> 01:28:15,400
Start with one decision that already causes friction.

2246
01:28:15,400 --> 01:28:17,400
Maybe it's the sequence at the bottleneck resource.

2247
01:28:17,400 --> 01:28:19,960
Every day, somebody decides which job runs next,

2248
01:28:19,960 --> 01:28:23,080
often with incomplete facts and a lot of pressure from due dates.

2249
01:28:23,080 --> 01:28:26,040
Or maybe the real pain sits in late order recovery,

2250
01:28:26,040 --> 01:28:29,160
where planners spend hours working out whether an order can still ship

2251
01:28:29,160 --> 01:28:31,560
without creating a worse problem somewhere else.

2252
01:28:31,560 --> 01:28:33,880
Material ready dispatch is another place to start.

2253
01:28:33,880 --> 01:28:37,720
Some plants lose time because work reaches the resource before

2254
01:28:37,720 --> 01:28:40,280
all required material has been released and staged.

2255
01:28:40,280 --> 01:28:43,560
People reshuffle jobs locally, usually for sensible reasons,

2256
01:28:43,560 --> 01:28:46,200
but the formal schedule never learns from the change.

2257
01:28:46,200 --> 01:28:48,600
Choose the decision people already argue about.

2258
01:28:48,600 --> 01:28:51,000
Look for one with a repeat pattern, a clear owner,

2259
01:28:51,000 --> 01:28:53,000
and a consequence everyone recognizes.

2260
01:28:53,000 --> 01:28:55,480
If every shift handles it differently, that's useful.

2261
01:28:55,480 --> 01:28:57,560
That means there's probably knowledge worth capturing,

2262
01:28:57,560 --> 01:29:00,840
even if it currently lives in calls, whiteboards, and files called

2263
01:29:00,840 --> 01:29:02,760
final schedule, really final v7.

2264
01:29:02,760 --> 01:29:05,960
Excel does not survive by accident.

2265
01:29:05,960 --> 01:29:09,400
So once you have the decision, map how it happens today.

2266
01:29:09,400 --> 01:29:11,480
Not the process shown in a procedure document.

2267
01:29:11,480 --> 01:29:15,160
Follow the real path from the first trigger to the dispatch instruction

2268
01:29:15,160 --> 01:29:17,480
that reaches the floor, who notices the problem first.