In this episode of the M365.fm podcast, we explore how organizations can maximize the value of Microsoft Copilot across Microsoft 365, Power Platform, Security, and Azure environments. The discussion focuses on practical, real-world use cases that help businesses automate repetitive work, improve collaboration, accelerate reporting, and enhance decision-making using AI.

The episode explains how Copilot integrates with tools like Word, Excel, Outlook, Teams, Power BI, and Power Automate to simplify everyday tasks such as content creation, meeting summaries, data analysis, workflow automation, and reporting. It also covers how Microsoft Graph enables Copilot to provide context-aware assistance by connecting emails, files, meetings, and business data.

Beyond productivity, the episode dives into advanced scenarios including Copilot extensibility, custom plugins, AI agents, governance, security operations, and compliance automation. Special attention is given to the importance of data governance, user training, prompt engineering, and change management when deploying Copilot at scale.

Listeners will learn practical strategies for improving Copilot adoption, avoiding common implementation mistakes, and building a secure and scalable AI foundation inside Microsoft 365. Whether you are an IT professional, architect, business leader, or Microsoft partner, this episode provides actionable insights for turning Microsoft Copilot into a real productivity and business transformation tool.

You’ve probably seen Microsoft Copilot in demos, but it does much more in real work. It helps you speed up tasks by 12% and saves about 26 minutes daily, according to studies with thousands of users. Copilot fits right into Microsoft 365 apps you use every day, like Word and Teams, making your workflow smoother. It also shines a light on security and governance by controlling data access and keeping your information safe. When you focus on maximizing microsoft copilot, you unlock real productivity and protect your data at the same time.

Key security features include:
Role-based access control
Data encryption during transit and rest
Compliance with GDPR and HIPAA
Activity monitoring and auditing

Key Takeaways

Microsoft Copilot can boost your productivity by 12% and save you about 26 minutes each day.
Use the voice-to-text feature in Copilot to transcribe meetings in real time, allowing you to focus on discussions without missing key points.
Copilot's AI-powered proofreading tools help you catch grammar mistakes and improve your writing, making communication clearer and more professional.
Automate repetitive tasks with Copilot to streamline your workflow, allowing you to focus on more important projects.
Leverage Copilot's summarization feature to quickly digest long reports and emails, saving you time and reducing overwhelm.
Utilize Copilot for effective scheduling, ensuring you never miss important meetings or deadlines while keeping your calendar organized.
Stay updated with new features and training resources to maximize your use of Copilot and enhance your team's efficiency.
Implement strong governance and security measures with Copilot to protect your data and ensure compliance with regulations.

Maximizing Microsoft Copilot Features

Voice-to-Text Tools

Real-Time Transcription

You can turn spoken words into text instantly with the voice-to-text feature in Copilot. This tool works right inside Teams and Word, so you don’t need to switch apps or copy notes. When you join a meeting, Copilot listens and transcribes everything in real time. You get a written record that’s accurate and easy to search. This voice-to-text feature uses advanced AI to recognize context, industry terms, and even proper names. It’s much more than basic dictation. In fact, top speech-to-text tools now reach over 50% higher accuracy than standard dictation, letting you replace up to 95% of your typing with your voice.

Here’s how Copilot’s voice-to-text feature compares to other solutions:

Feature	Willow	Dragon Pro	Google Docs	Word Voice Input	Apple Dictation
Context Awareness	✅	Basic	No	No	No
Real-time Processing	✅	✅	✅	✅	Basic
Custom Vocabularies	✅	✅	No	No	No
Multi-language	✅	Limited	✅	✅	✅

You’ll notice that Copilot stands out for context awareness and real-time processing. This means you can trust your transcripts to capture the details that matter.

Meeting Notes

Taking notes during meetings can slow you down. With Copilot’s voice-to-text feature, you can focus on the conversation while Copilot creates structured meeting notes for you. In Teams, Copilot automatically organizes key points, decisions, and action items. You don’t have to worry about missing anything. Over 70% of Teams users rely on Copilot for meeting recaps, which shows how valuable this feature is for maximizing microsoft copilot in your daily workflow.

Tip: After your meeting, you can ask Copilot to summarize the discussion or highlight follow-up tasks. This keeps everyone on the same page and boosts productivity.

Proofreading & Editing

Grammar Suggestions

You want your writing to look professional and clear. Copilot’s AI-powered proofreading tools help you catch grammar mistakes and awkward sentences before you hit send. Whether you’re drafting an email in Outlook or writing a report in Word, Copilot scans your text and offers instant suggestions. This efficient proofreading process saves you time and helps you avoid embarrassing errors. In fact, 81% of Word users use Copilot to rewrite or simplify language, making editing tasks much easier.

Tone Adjustment

Sometimes, you need to adjust your message for the right audience. Maybe you want your email to sound friendly, or your report to feel formal. Copilot’s AI can analyze your writing and suggest changes to match your desired tone. You can switch from casual to professional with just a click. This efficient proofreading feature ensures your communication always fits the situation. You’ll find this especially helpful when you need to rewrite paragraphs or adapt your brand voice.

Content Reformatting

Document Conversion

Switching between formats can be a hassle. Copilot makes it simple to convert documents from one style to another. You can turn meeting notes into a polished report, or change a Word document into a PowerPoint presentation. Copilot handles the formatting, so you don’t waste time on manual edits. This feature helps you keep your workflow smooth and consistent.

Summarization

Long reports and emails can overwhelm you. Copilot’s AI can summarize lengthy documents in seconds. You get the main points without reading every word. This voice-to-text feature also extracts action items and highlights for quick review. Here are some ways Copilot’s content reformatting boosts your efficiency:

Generate first drafts from short prompts or notes.
Summarize long reports for executive review.
Extract action items from multi-page documents.
Format and structure proposals with a professional touch.

You’ll see a real impact on productivity when you start maximizing microsoft copilot for these tasks. For example, in the financial sector, a global bank used Copilot to reduce quarterly reporting times by 40%. In healthcare, providers saved time on clinical documentation, giving them more time for patient care.

Note: Maximizing microsoft copilot means letting AI handle routine editing tasks, so you can focus on what matters most.

Task Automation

When you think about getting more done in less time, task automation stands out as a game changer. With copilot, you can hand off repetitive chores and focus on work that matters. Let’s look at how you can use copilot to automate your daily routines and keep your projects moving forward.

Workflow Integration

You don’t have to jump between apps or copy information from one place to another. Copilot connects your favorite Microsoft 365 tools, like Outlook, Teams, and Excel, so your tasks flow smoothly. For example, you can ask copilot to pull action items from a Teams meeting and add them to your Planner board. It can even update your Excel project tracker when someone completes a task in Teams.

Here’s what you can do with workflow integration:

Build project plans from a simple prompt. Just tell copilot what you want, and it creates a structure for you.
Spot overloaded team members. Copilot reviews your project and highlights who has too much on their plate.
Track progress automatically. Copilot monitors deadlines and sends reminders, so nothing slips through the cracks.

Avanade’s project team used copilot to automate post-meeting documentation. They cut down their documentation time by 40%. That’s about two hours saved each week for every team member. Imagine what you could do with that extra time.

Tip: Try using copilot to summarize meeting notes and send updates to your team. You’ll keep everyone in sync without extra effort.

Scheduling

Keeping track of meetings, deadlines, and appointments can get overwhelming. Copilot makes scheduling easy. You can ask it to find open times on your calendar, set up meetings, or even reschedule appointments if something changes.

Here’s how copilot helps with scheduling:

Suggests meeting times that work for everyone, even across time zones.
Books rooms and sends invites in Outlook with just a quick prompt.
Reminds you of upcoming deadlines and helps you prioritize your day.

You don’t have to worry about double-booking or missing important events. Copilot handles the details, so you can focus on your work.

Note: When you let copilot manage your scheduling, you free up mental space and reduce stress. It’s like having a personal assistant who never forgets.

With copilot, you turn routine tasks into automated processes. You save time, avoid mistakes, and keep your workflow running smoothly.

Daily Copilot Applications

Communication Management

Email Drafting

You probably spend a lot of time in your inbox. With copilot, you can draft emails faster and respond to messages without starting from scratch. The Draft tool gives you context-based suggestions, so you don’t have to worry about missing important details. If you get a long email thread, just hit the Summarize button. Copilot gives you a quick overview, so you catch up in seconds. You can also set up inbox rules with ai to organize your messages and highlight what matters most. This keeps your inbox clean and your mind clear.

Did you know? Most employees get over 120 emails every day. Copilot helps you manage this overload by suggesting replies and prioritizing messages. You stay on top of your communications and keep your workflow moving.

Chat Summaries

Chat apps like Teams can fill up fast. Copilot steps in by generating summaries of your chats, so you don’t have to scroll through endless messages. You get the main points and action items right away. This feature is perfect for busy days when you need to catch up quickly. Automating meeting notes and generating summaries with ai means you never miss a detail.

Collaboration Tools

Shared Editing

Working on documents with your team gets easier with copilot. You can co-edit files in real time, and ai helps by suggesting edits, fixing grammar, and adjusting tone. This makes your documents look polished and professional. Copilot also tracks changes and keeps everyone on the same page.

Project Support

Managing projects can feel overwhelming. Copilot helps by automating repetitive tasks, like drafting budgets or timelines. It also summarizes meetings and tracks tasks, so nothing slips through the cracks. Teams using copilot have saved hundreds of hours each month. That’s real efficiency.

Scenario	How Copilot Helps
Crafting a business plan	Generates a detailed document with market analysis and projections from a prompt.
Managing emails	Summarizes threads and drafts replies efficiently.
Designing presentations	Turns rough notes into polished PowerPoint slides.
Automating routine tasks	Schedules meetings and organizes customer data.

Data Insights

Reporting

You don’t need to be a data scientist to get insights. Copilot acts as your analyst agent, simplifying complex datasets with ai. You can ask questions in plain language and get formulas, summaries, or even visualizations. This saves you hours on data gathering and analysis.

Visualization

Seeing your data makes decision-making easier. Copilot uses ai to create charts, graphs, and forecasts from your numbers. You get narrative summaries and observations, so you understand the story behind the data. Integrating ai into your daily routine means you make smarter choices, faster.

Tip: Use copilot to synthesize information from multiple sources. You’ll boost productivity and free up time for creative problem-solving.

Personal Productivity

Time Management

Managing your time can feel like a juggling act. With copilot, you get a digital assistant that helps you stay on track. You can ask copilot to suggest meeting times that fit everyone’s schedule. It even drafts invites with pre-filled agendas and sends reminders, so you never miss a beat. After a meeting, copilot understands the context and helps you schedule next steps without opening your calendar.

Here’s how copilot makes time management easier:

Suggests the best meeting times based on everyone’s availability.
Drafts and sends invites with all the details included.
Sets reminders for important deadlines and events.
Helps you quickly schedule follow-up tasks after meetings.

Tip: Let copilot handle your scheduling. You’ll spend less time planning and more time doing.

Task Prioritization

It’s easy to feel overwhelmed by a long to-do list. Copilot steps in to help you focus on what matters most. You get a prioritized action list that highlights overdue deliverables and urgent assignments. This means you always know which tasks need your attention first.

Copilot keeps your task list up to date. It automatically incorporates changes, so you don’t have to sort tasks by hand. You also get clear visibility into your team’s assignments. Copilot offers actionable suggestions for managing priorities, which helps you spot bottlenecks and avoid confusion.

Here’s what you can expect when you use copilot for task prioritization:

Automatic sorting of tasks by urgency and importance.
Regular prompts to review and adjust your priorities.
Clear suggestions for tackling high-impact work.
Updates that reflect shifting project needs, so you stay aligned with your goals.

Note: When you let copilot guide your priorities, you reduce stress and boost your efficiency. You’ll always know where to focus your energy.

With copilot, you turn time management and task prioritization into strengths. You get more done, feel less stressed, and keep your projects moving forward.

Prompting Copilot for Best Results

Getting the most out of ai starts with how you talk to it. When you use copilot, the way you ask for help shapes the answers you get. Let’s break down how you can prompt ai for the best results and boost your efficiency every day.

Effective Prompts

Clarity

Clear prompts lead to better answers. If you want ai to write an executive summary, say exactly that. Don’t just ask for “a summary.” Instead, try, “Write a three-sentence executive summary for this report.” This direct approach helps ai understand your goal and deliver what you need.

Set a clear goal for each request.
Tell ai what you want the output to look like.
Use simple language and avoid vague words.

Tip: The clearer your instructions, the more precise and useful the ai’s response will be.

Context

Context gives ai the background it needs to tailor its answers. If you’re asking for a summary, explain who will read it and why. For example, “Summarize this project update for the leadership team. Focus on key milestones and next steps.” This helps ai pick the right details and tone.

Share why you need the output.
Add any important background information.
Mention how you plan to use the results.

When you combine clarity and context, you get responses that fit your needs perfectly.

Iterative Requests

Refinement

Sometimes, you won’t get the perfect answer on the first try. That’s normal with ai. Start with a general prompt, then ask for changes. For example, if the first draft is too long, say, “Make it shorter and more formal.” This step-by-step method helps ai refine its output.

Begin with a well-structured prompt.
Review the response and ask for tweaks.
Break big tasks into smaller steps.

Feedback

Giving feedback is key. If ai misses the mark, tell it what to fix. You might say, “Add more data from last quarter,” or “Use a friendlier tone.” Each round of feedback helps ai learn your preferences and improve future results.

Note: Iterative requests make complex tasks easier and help ai deliver exactly what you want.

Templates & Customization

Pre-Built Prompts

You don’t have to start from scratch every time. Copilot Studio offers templates for common tasks, like drafting emails or creating reports. These templates save time and help you get consistent results.

Feature	Customization Level
Copilot Studio templates	Moderate effort
Custom build	Higher effort

Custom Strategies

If you need something unique, you can customize ai agents to fit your workflow. Tuning these agents lets you match outputs to your company’s standards or your personal style. This flexibility supports everything from personal productivity to enterprise-wide solutions.

Adjust templates to fit your needs.
Build custom agents for special projects.
Fine-tune outputs for different teams or departments.

With the right prompts and a bit of customization, you can unlock the full power of ai and make your daily work smoother.

Copilot Governance & Security

When you bring AI into your daily work, you want to know your data stays safe and your team follows the rules. Copilot makes this easier by building on Microsoft 365’s strong security and governance features. Let’s break down what you need to know.

Data Governance

Permissions

You control who can see and use sensitive information. Copilot uses Microsoft 365’s Role-Based Access Control (RBAC) to make sure only the right people get access. You can set up permissions so team members see only what they need. This helps you avoid accidental data leaks and keeps your business safe.

Review your Microsoft Purview settings for data classification.
Use sensitivity labels to mark important files.
Set up role-based access controls to block unauthorized users.

Compliance

If you work in a regulated industry, you need to meet strict rules. Copilot supports compliance with standards like GDPR, HIPAA, and ISO certifications. Microsoft offers tools for data residency, risk management, and audit trails. For healthcare, you get business associate agreements and special data boundaries. Financial services teams benefit from extra controls and monitoring for sensitive information. No matter where you work, you can trust that copilot helps you stay compliant.

Note: Microsoft keeps up with new laws, like the EU AI Act, so your AI tools always meet the latest standards.

Security Challenges

Data Handling

You want to know how your data moves and who can see it. Copilot encrypts data at rest and in transit. It also limits data access to only what’s needed for each task. Microsoft 365 keeps logs of all activities, so you can audit and track what happens.

Confidentiality

Security threats can pop up in any system. Here’s a quick look at some challenges and how Microsoft addresses them:

Security Challenge	What It Means	How It’s Handled
Overpermissioning	Too many users have access to sensitive data	Strict user permissions
Model Inversion Attacks	Attackers try to extract private info from AI models	Continuous monitoring
Integration Vulnerabilities	Weak spots in connected apps	Advanced security tools
Prompt Injection Attacks	Bad actors trick AI into sharing data	Monitoring and security tools

You can feel confident knowing these risks get managed with strong controls and constant oversight.

Responsible Use

Human Oversight

AI can do a lot, but you still need to keep an eye on it. Set up review gates for high-risk content. Always have a person check decisions that could have a big impact. Try using copilot for low-risk tasks first, so your team gets comfortable with prompting and checking results.

Verification Habits

Get into the habit of double-checking AI outputs. Train your team to review and validate important information before sharing it. Use pilot programs to measure efficiency, error reduction, and user satisfaction. This helps you build trust in your AI tools and keeps your business running smoothly.

Tip: Responsible use means combining smart technology with human judgment. You get the best results when people and AI work together.

Microsoft Copilot Studio

Microsoft Copilot Studio gives you a powerful way to build and customize AI tools for your business. You don’t need to be a coding expert to get started. Let’s look at how you can use Copilot Studio to create solutions that fit your needs.

Low-Code Solutions

Citizen Development

You can become a citizen developer with Copilot Studio. The platform makes it easy for anyone to build AI assistants for different jobs. You can automate tasks like data entry or scheduling without writing much code. This means you save time and reduce mistakes.

You get a user-friendly interface that guides you step by step.
You can create tools that help your team work faster.
You don’t need deep coding skills to make a big impact.
Many organizations use Copilot Studio to boost efficiency and productivity.

When you use low-code tools, you help your company move faster. You also make AI development more open to everyone, not just IT experts.

Pro-Code Solutions

Enterprise Applications

If you have advanced skills, you can build complex enterprise applications with Copilot Studio. Large companies use the platform to solve big challenges.

ABN AMRO Bank improved customer service by building smarter chatbots.
Cineplex used Copilot Studio to automate business processes. They saved over 30,000 hours of manual work each year.

You can connect Copilot Studio with other Microsoft tools. This lets you create solutions that fit your company’s needs. You can handle large projects and support many users at once.

Custom AI Integration

Tailored Automation

You can take automation to the next level with custom AI integration. Many technology leaders plan to spend more on automation because they see its value.

Benefit	Description
Faster processes	AI agents handle routine work, so you can focus on bigger goals.
Better productivity	Teams spend less time searching for information. Some companies cut search time by 70%.
Business growth	AI tools help you deliver better results for your customers.

Some organizations use Copilot Studio to build AI workbenches with multiple agents. These agents help teams deliver projects faster and improve business outcomes.

Tip: Start small with one process, then expand as your team gets comfortable. You’ll see how custom AI can transform your daily work.

Staying Ahead with Microsoft Copilot

To get the most from Microsoft Copilot, you need to keep building your skills and stay up to date with new features. Let’s look at how you can keep learning and help your team grow.

Skill Building

Training

You have many ways to learn how to use Copilot better. Hands-on, live virtual learning sessions let you practice real tasks. Workshops give you a chance to apply what you learn right away. Microsoft offers a learning hub with courses for all skill levels, from beginners to advanced users. You can join the Microsoft 365 Champions program or explore the AI Skills Navigator to boost your knowledge. These programs help you and your team roll out Copilot smoothly and confidently.

Live virtual learning sessions
Workshops for hands-on practice
Microsoft 365 Champions program
AI Skills Navigator
Learning hub with Copilot courses
Advanced resources for IT pros, admins, and leaders

Tip: Try a hands-on session to see how Copilot fits into your daily work. You’ll pick up new tricks and feel more comfortable using AI.

Community Resources

You don’t have to learn alone. The Microsoft Tech Community and Developer Community are great places to ask questions and share ideas. You can join community events or find a dedicated support hub for Copilot users. Early adopters often become champions who help others and share best practices. Community experts pin important resources and answer questions, so you always have help when you need it.

Join a learning and support hub
Ask questions and get answers from peers
Find early adopters and champions for guidance
Access pinned resources and expert advice
Use a community agent for quick answers

Continuous Improvement

Updates

Microsoft Copilot gets frequent updates with new tools and features. These updates make your experience better and keep you ahead of the curve. To get the latest improvements, turn on auto-updates. This way, you always have access to the newest capabilities and enhancements.

Note: Staying updated means you can try new features as soon as they launch. You’ll keep your workflow fresh and efficient.

Adoption Strategies

Rolling out Copilot across your organization works best with a clear plan. Many teams use a 30-60-90 day roadmap. Start by setting a baseline, then drive activation, and finally scale up usage. Teach your team how to craft effective prompts. Use digital adoption platforms for in-context guidance and self-service support. Monitor user engagement and feature usage to spot knowledge gaps. Regular surveys and feedback sessions help you understand what’s working and where to improve.

Use a 30-60-90 day adoption plan
Teach prompt crafting for better results
Provide just-in-time training
Use digital adoption platforms for support
Track engagement and gather feedback

Callout: When you invest in training and community, you help everyone get more from Copilot. Continuous learning and feedback keep your team ahead.

You have a chance to unlock more value with copilot every day. Start by exploring new features and integrations. To get the most out of your journey, try these steps:

Enable Copilot Chat and prepare your team.
Define your core unit of work.
Prioritize and scope agentic projects.
Test ideas with a pilot.
Assess and scale what works.

Strong governance, security, and custom development help you stay safe, meet standards, and build solutions that fit your needs. Keep learning and experimenting to drive real innovation.

FAQ

How do I start using Microsoft Copilot in Microsoft 365?

You just need to open Word, Excel, Outlook, or Teams. Look for the Copilot icon or prompt bar. Click it and type your request. Copilot will guide you step by step.

Can I customize Copilot for my team’s needs?

Yes! You can use Copilot Studio to build custom workflows or prompts. You don’t need to know how to code. Just follow the guided steps and test your changes.

Is my data safe when I use Copilot?

Absolutely. Microsoft Copilot uses strong encryption and role-based access. Only people with permission can see your data. Microsoft also follows strict compliance standards.

What apps work with Microsoft Copilot?

You can use Copilot in Word, Excel, PowerPoint, Outlook, Teams, and SharePoint. Copilot also connects with other Microsoft 365 tools for a smooth workflow.

How does Copilot help with meetings?

Copilot can transcribe meetings, create summaries, and highlight action items. You can ask it to send follow-up emails or set reminders. This keeps everyone on track.

Do I need special training to use Copilot?

No special training is needed. Copilot is easy to use. You can learn by trying out features or joining a Microsoft learning session. The interface is user-friendly.

Can Copilot help me manage my time?

Yes! Copilot suggests meeting times, sends reminders, and helps you prioritize tasks. You can ask it to organize your day or week.

Where can I get help if I have questions about Copilot?

You can visit the Microsoft Tech Community or use the in-app help feature. There are also learning hubs and support forums where you can ask questions and get answer

🚀 Want to be part of m365.fm?

Then stop just listening… and start showing up.

👉 Connect with me on LinkedIn and let’s make something happen:

🎙️ Be a podcast guest and share your story
🎧 Host your own episode (yes, seriously)
💡 Pitch topics the community actually wants to hear
🌍 Build your personal brand in the Microsoft 365 space

This isn’t just a podcast — it’s a platform for people who take action.

🔥 Most people wait. The best ones don’t.

👉 Connect with me on LinkedIn and send me a message:
"I want in"

Let’s build something awesome 👊

1
00:00:00,000 --> 00:00:04,480
Welcome to another edition of the MC65 podcast.

2
00:00:04,480 --> 00:00:07,040
Today we are driving deep into the real world,

3
00:00:07,040 --> 00:00:09,120
state of Microsoft co-pilot, AI,

4
00:00:09,120 --> 00:00:11,360
adoption governance, prompting automation

5
00:00:11,360 --> 00:00:16,480
and what's actually takes to get value behind the flashy demos.

6
00:00:16,480 --> 00:00:19,440
Joining me today is Ryan Rivas, MVP,

7
00:00:19,440 --> 00:00:22,000
also known as the co-pilot Junkie,

8
00:00:22,000 --> 00:00:25,920
a Microsoft 365 architect and governance expert

9
00:00:25,920 --> 00:00:28,720
with a deep experience in cross-power platform,

10
00:00:28,720 --> 00:00:30,320
SharePoint administration,

11
00:00:30,320 --> 00:00:34,080
especially co-pilot studio and AI driven workflows.

12
00:00:34,080 --> 00:00:38,000
In this episode, we talk about where Microsoft co-pilot

13
00:00:38,000 --> 00:00:42,160
stand today, the biggest mistake organizations are making,

14
00:00:42,160 --> 00:00:45,520
how prompting and agents are changing, what are work,

15
00:00:45,520 --> 00:00:49,760
and what companies should focus on, is they want real business

16
00:00:49,760 --> 00:00:51,600
outcomes from AI.

17
00:00:51,600 --> 00:00:55,920
If you're trying to separate AI hype from practical reality

18
00:00:55,920 --> 00:00:58,960
in Microsoft 365, this is for you.

19
00:00:58,960 --> 00:01:02,800
So sit back, relax and let us go in.

20
00:01:02,800 --> 00:01:05,200
So, Ryan, welcome.

21
00:01:05,200 --> 00:01:07,200
Yes, oh, finally, Junkie.

22
00:01:07,200 --> 00:01:08,480
Welcome to the show.

23
00:01:08,480 --> 00:01:09,520
Thank you. Hold on a second.

24
00:01:09,520 --> 00:01:12,080
I have to get my fix here in just a moment.

25
00:01:12,080 --> 00:01:13,520
And there we go. I got my fix.

26
00:01:13,520 --> 00:01:14,640
I'm ready to continue.

27
00:01:14,640 --> 00:01:19,360
Can you, for the people they don't know,

28
00:01:19,360 --> 00:01:21,280
can you introduce yourself?

29
00:01:21,280 --> 00:01:21,680
Sure.

30
00:01:21,680 --> 00:01:26,160
This little, a little bit, a bunch of people.

31
00:01:26,160 --> 00:01:27,280
So, what's your problem?

32
00:01:27,280 --> 00:01:29,280
This, in Chicago is where I'm at.

33
00:01:29,280 --> 00:01:31,680
In Chicago, also known for some folks that might know the city

34
00:01:31,680 --> 00:01:33,680
as the home of second city,

35
00:01:33,680 --> 00:01:36,880
where a lot of the comedy that the America come from,

36
00:01:36,880 --> 00:01:41,520
from America tends to have come from for a lot of the last years

37
00:01:41,520 --> 00:01:43,040
that we know.

38
00:01:43,040 --> 00:01:46,640
It's maybe not a surprise that I might have that particular way

39
00:01:46,640 --> 00:01:49,280
of talking and presenting everything else.

40
00:01:49,280 --> 00:01:52,240
It helps. Let's just say in very tense situations.

41
00:01:52,240 --> 00:01:56,800
And to be sure, AI is not yet caught up with my demeanor.

42
00:01:56,800 --> 00:01:59,200
So, I think we're okay for now.

43
00:01:59,200 --> 00:02:03,360
Is that I'm still working the user groups

44
00:02:03,360 --> 00:02:04,960
and everything else here is the thing.

45
00:02:04,960 --> 00:02:07,200
Even though I do this as they say, the day job

46
00:02:07,200 --> 00:02:09,520
and much has grown up my stakeholders at home,

47
00:02:09,520 --> 00:02:12,560
I'm doing this for the community, hence the MVP stuff.

48
00:02:12,560 --> 00:02:14,000
Thank you very much, Microsoft.

49
00:02:14,000 --> 00:02:17,360
But mostly, like for shows like this,

50
00:02:17,360 --> 00:02:20,560
who has a shared information to love,

51
00:02:20,560 --> 00:02:22,720
because the more we know, the more all of us know,

52
00:02:22,720 --> 00:02:25,360
the better. And that's what I'm about, I hope, to say.

53
00:02:25,360 --> 00:02:30,800
Awesome. Where did your journey into Microsoft say 65?

54
00:02:30,800 --> 00:02:32,080
Acro-system begin?

55
00:02:32,080 --> 00:02:37,200
So, it began actually pretty much the way a lot of folks,

56
00:02:37,200 --> 00:02:42,880
at least for at my level back in the turn of the century over here,

57
00:02:42,880 --> 00:02:45,760
when they said, "Hey, there's this program."

58
00:02:45,760 --> 00:02:47,600
It's called SharePoint. Could you do it?

59
00:02:47,600 --> 00:02:50,320
And the person said SharePoint, and then a moment you said SharePoint?

60
00:02:50,320 --> 00:02:52,320
Yes, okay, good. You were doing it.

61
00:02:52,320 --> 00:02:54,960
I think it was SharePoint that got me into a lot of those things.

62
00:02:54,960 --> 00:03:00,480
But I've been doing the Microsoft world for quite a while before, and SharePoint,

63
00:03:00,480 --> 00:03:03,840
to me, was just the culmination of where things in Excel and everything else

64
00:03:03,840 --> 00:03:10,480
were going. The office essentially sweet, finding a home to live, hence SharePoint.

65
00:03:10,480 --> 00:03:13,280
Right? But through that, of course, it expanded.

66
00:03:13,280 --> 00:03:17,840
And it eventually became its own SaaS product.

67
00:03:17,840 --> 00:03:23,200
As when it became a SaaS product, I thought it was a great idea, because even before that,

68
00:03:23,200 --> 00:03:26,800
I was already trying to experiment with my other buddy and my other podcast.

69
00:03:26,800 --> 00:03:31,440
I'll quickly put a shout out for that for the CloudShow podcast.

70
00:03:31,440 --> 00:03:37,200
The Skylary Smith-Mire, we were like an Azure building SharePoint servers in there,

71
00:03:37,200 --> 00:03:42,960
so that we could get stuff going because we couldn't wait for infrastructure to actually do some work.

72
00:03:42,960 --> 00:03:47,920
So here's the SaaS, we're doing the SaaS, we jump into it, we're doing this work just fine.

73
00:03:47,920 --> 00:03:51,600
Next thing you know, it becomes something called M365 at some point.

74
00:03:51,600 --> 00:03:58,160
And by that time, I was already deeply immersed. So everything else that followed was where I thought

75
00:03:58,160 --> 00:04:03,600
maybe I'd be a SharePoint MVP, but no, instead I became a platform MVP, which of course,

76
00:04:03,600 --> 00:04:08,400
called Cloud Studio, M365, called Cloud, all of that stuff follows from over there.

77
00:04:08,400 --> 00:04:10,400
So that's the story of how I got there.

78
00:04:10,960 --> 00:04:12,720
Well, a short version that is.

79
00:04:12,720 --> 00:04:19,120
And also all the links in the show notes for all the people.

80
00:04:19,120 --> 00:04:20,880
Absolutely.

81
00:04:20,880 --> 00:04:23,840
Sorry.

82
00:04:23,840 --> 00:04:34,240
How important is architecture and governance when organization starts adopting AI tools like

83
00:04:34,240 --> 00:04:37,600
Copilot, Copilot Studio or something else?

84
00:04:38,800 --> 00:04:44,480
It's a very good question. So when I perhaps neglected from this story over here is if you realize

85
00:04:44,480 --> 00:04:50,240
what I did, I basically defied the company in my particular case. So we just went online and we

86
00:04:50,240 --> 00:04:55,520
just simply put our SharePoint servers and started working. We essentially defied what would have been

87
00:04:55,520 --> 00:05:00,640
governance to getting through. Since then, of course, realizing that not only was I perhaps the

88
00:05:00,640 --> 00:05:05,600
main perpetrator, but the one that said, you know, I probably don't like if other people do this to us,

89
00:05:05,600 --> 00:05:09,920
and we have to help companies get under control. Governance is the key over here.

90
00:05:09,920 --> 00:05:15,040
All right. So from the very beginning, I think we would always look at things like power platform

91
00:05:15,040 --> 00:05:20,480
or any kind of applications or it didn't have to be power platformed really. Just governance in general,

92
00:05:20,480 --> 00:05:26,880
the concept of what shadow IT was all about was something that just had to be under control. As I got more,

93
00:05:26,880 --> 00:05:33,600
I guess, more experienced and moved up in the level as they say, my focus was not so much building

94
00:05:33,600 --> 00:05:40,720
things, but making sure things were under control and assuring IT that it was okay to let this happen,

95
00:05:40,720 --> 00:05:44,960
but how could they let it happen if they're if they couldn't see it, they couldn't monitor it and do

96
00:05:44,960 --> 00:05:51,520
any of those stuff. So here was the key. The irony is the tools have been there all this time, you're

97
00:05:51,520 --> 00:05:57,600
just not necessarily applied or not necessarily put together. I always look at governance first. So

98
00:05:57,600 --> 00:06:02,720
the one thing that at least I was impressed about on Microsoft of late is there at least they think

99
00:06:02,720 --> 00:06:07,680
about governance ahead of time. So when we talk about governance, say for like the AI tools and

100
00:06:07,680 --> 00:06:12,480
everything else, they're already thinking about it. They're already working on it. However, maybe

101
00:06:12,480 --> 00:06:17,360
they're not just as fast to come up with it. So I think a lot of companies were afraid just to do AI

102
00:06:17,360 --> 00:06:21,040
in general. So oh my gosh, we don't know if our data is safe and everything else is like, well,

103
00:06:21,040 --> 00:06:25,840
they tried to assure you, but it was just not enough. Not until agent 365. And by the way,

104
00:06:25,840 --> 00:06:34,800
that's the thing that I'm very excited about of late agent 365's introduction. I think at the start,

105
00:06:34,800 --> 00:06:43,040
Microsoft brings out co-pilot and a lot of people are of end users. I'm not happy with co-pilot.

106
00:06:43,040 --> 00:06:49,600
They're not with this shadow IT topic. I think a lot of people also on Microsoft side,

107
00:06:49,600 --> 00:07:00,080
I heard that they use chatGPT instead of the co-pilot model. Yes. So what was the reason for this and

108
00:07:00,080 --> 00:07:09,200
show they give a try to use co-pilot now? So this was interesting. I noticed this of Microsoft of late,

109
00:07:09,200 --> 00:07:13,920
where basically something that's not exactly ready at their starting to put out much sooner

110
00:07:13,920 --> 00:07:19,040
than perhaps it should mature. Now this is a very interesting thing because some companies wait

111
00:07:19,040 --> 00:07:23,680
until they feel very, very mature on a product before it comes out. And I'm going to say that those

112
00:07:23,680 --> 00:07:30,400
companies that wait too long are not in the market anymore. And this happened going to say with

113
00:07:30,400 --> 00:07:36,080
with me and another company called, another product called Lotus123, this is going back, I'm

114
00:07:36,080 --> 00:07:40,400
dating myself away back in the day. When I remember when I used to work for Reuters and we asked them,

115
00:07:40,400 --> 00:07:46,320
look, they have this thing on the Microsoft side that did this capability, it was called DDE,

116
00:07:46,320 --> 00:07:51,120
the ability to bring live data into spreadsheets. And they said, well, we're working on it. We want to

117
00:07:51,120 --> 00:07:56,240
make it perfect. We don't want to just put something out. And they have, let's just say they gave a

118
00:07:56,240 --> 00:08:03,520
numerous amount of reasons why they were going to not have it right away. Well, the market passed them,

119
00:08:03,520 --> 00:08:08,080
surpassed them. And other aspects of course of Lotus have been surpassed. Maybe there was other

120
00:08:08,080 --> 00:08:13,120
reasons for that not to happen, but that was the one thing that I took as a lesson that said,

121
00:08:13,120 --> 00:08:18,000
if I see Microsoft doing something like this coming out with something a little sooner, I'll app, as

122
00:08:18,000 --> 00:08:23,680
they say, I want to give it whatever support I can, but I will try to give as much, oh, feedback,

123
00:08:23,680 --> 00:08:29,120
positive or negative as much as I can. Microsoft was coming out with things faster because they

124
00:08:29,120 --> 00:08:33,360
wanted to try to, I think what they're trying to do is just get the, the, a bit not so much market share,

125
00:08:33,360 --> 00:08:40,000
but they get the idea, look, we are, we are there. We're in the game. All right, teams, I give,

126
00:08:40,000 --> 00:08:44,880
teams, I think is the biggest example teams came out and everyone was not very happy about that. Now,

127
00:08:44,880 --> 00:08:49,680
it pretty much it's so ubiquitous everywhere for good or ill, what are people love it or not?

128
00:08:49,680 --> 00:08:57,040
That, that thing is solid and it's got its, it's place. Copilot, when it came out, it was week compared

129
00:08:57,040 --> 00:09:02,080
to say the, the, the other products that were out there, this is sure, but all of a sudden we turn

130
00:09:02,080 --> 00:09:06,640
around and like, whoa, look what copilot's doing. Because now I'm seeing what some of these other

131
00:09:06,640 --> 00:09:11,760
products are still doing and they feel beyond, to me, those products now feel behind. And then I find

132
00:09:11,760 --> 00:09:18,160
out the reason for that is how they waited to implement themselves into the, at least our market,

133
00:09:18,160 --> 00:09:22,160
such that when they showed up, they seem to be always a little bit more behind in Microsoft because

134
00:09:22,160 --> 00:09:27,360
even watching Microsoft evolve. So why did they do that? I really believe it was just to get a hold

135
00:09:27,360 --> 00:09:33,840
of us ahead of time. What's the word for it? The gateway drug. You get in. It's free. Oh, yes. And now I

136
00:09:33,840 --> 00:09:40,160
must have some more. Hence my term, the, this copilot junkie comes into play once again. Wow,

137
00:09:40,160 --> 00:09:46,400
big circles. Thanks for thinking that one, that question up. Yeah. Let's look at the, at the,

138
00:09:46,400 --> 00:09:52,080
at the big picture. Yes. Where is Microsoft copilot today in your opinion?

139
00:09:52,080 --> 00:09:59,120
Copilot today sits in the top tiers with the other, with the other systems. And the,

140
00:09:59,120 --> 00:10:07,280
the main reason for that is how it is, I believe, integrated into the M365 ecosystem. Basically,

141
00:10:07,280 --> 00:10:12,000
everyone is using Outlook, everyone is using Excel, everyone is using SharePoint in those products.

142
00:10:12,000 --> 00:10:19,360
It's so built in, right? That if anyone else wants to do something there, they have to figure out how

143
00:10:19,360 --> 00:10:24,880
to work with said vendor on everything else. Now, not that said vendor in this particular case,

144
00:10:24,880 --> 00:10:29,600
Microsoft is going to make it hard for them or prevent them or everything else. No, everyone

145
00:10:29,600 --> 00:10:35,040
likes to be friends when in this particular case. But if there's anyone that's probably going to know

146
00:10:35,040 --> 00:10:40,480
their system best or at least be responsible for when something goes wrong, it'll probably be

147
00:10:40,480 --> 00:10:46,400
Microsoft. So as far as where they sit, I'm going to say it's the easiest access. It's the easiest

148
00:10:46,400 --> 00:10:52,560
waypoint, the fastest way to get towards a lot of the AI. Now, that's pretty much again for

149
00:10:52,560 --> 00:10:58,880
personal productivity and to a great deal, a lot of the enterprise stuff. But I suppose there will

150
00:10:58,880 --> 00:11:04,960
still be room for everyone else when it comes to getting the, the precise solutions that you actually

151
00:11:04,960 --> 00:11:14,320
need for whatever the problem is. Yeah, I think last month's most MVP summit at Rutland. That's right.

152
00:11:16,000 --> 00:11:24,640
That was March, right? Yes. Yeah. And yeah, I see there was a lot, I see a lot of posts on LinkedIn

153
00:11:24,640 --> 00:11:35,120
forbidden or not, not a lot of photograph you need. So can you as of, yeah, I hope it's not forbidden.

154
00:11:35,120 --> 00:11:40,800
I can you say what Microsoft plans co-pellet in the next one? What can I tell is a secret before

155
00:11:40,800 --> 00:11:47,680
going? That was a good secret. Good thing you blocked that out. No, just kidding. No, seriously, as

156
00:11:47,680 --> 00:11:52,400
far as what they're coming, here's the thing. Microsoft, even though they have the the the NDAs for things,

157
00:11:52,400 --> 00:11:59,440
lately, NDAs for our level, our level are more like the very specifics of how something is going to

158
00:11:59,440 --> 00:12:06,320
work, not necessarily what they're going to do. Microsoft is telegraphing pretty loudly and publicly,

159
00:12:06,320 --> 00:12:12,480
this is where we're going. This is what we like to do. For those who are administrators out there,

160
00:12:12,480 --> 00:12:17,840
if you maybe maybe it's too much information, but try to funnel and filter through the information

161
00:12:17,840 --> 00:12:24,160
in your M365 message center and see that they actually tell you months ahead what's coming and

162
00:12:24,160 --> 00:12:30,320
sometimes those little devil in the details are things that are game changers for what's what's

163
00:12:30,320 --> 00:12:36,160
going on. So they're actually telegraphing it. The specific details, maybe the exact date for something,

164
00:12:36,160 --> 00:12:41,600
or anything else, those are the NDA pieces over here. So it might be like, I'll maybe give you an

165
00:12:41,600 --> 00:12:48,160
example of something that came out not too long ago. The concept of Gen pages, Gen pages is the

166
00:12:48,160 --> 00:12:53,120
in model driven applications, you could basically use vibe coding to create beautiful pages and

167
00:12:53,120 --> 00:12:58,720
everything else, which have been out when it was coming out. And NDA, it was like, oh my gosh,

168
00:12:58,720 --> 00:13:03,760
this is going to be a very good. Maybe we should plan for this now and start talking about it.

169
00:13:03,760 --> 00:13:08,320
They said, well, we're not sure it's not cooked yet, maybe not really talk about it, but that didn't

170
00:13:08,320 --> 00:13:14,240
last long. As far as they were, they cooked it up fast mainly because folks like partners,

171
00:13:14,240 --> 00:13:19,920
their customers that they're working with and ourselves or the MPPs working with them

172
00:13:19,920 --> 00:13:25,520
helped at least get that feedback, they get things out there, they get things going. So it's,

173
00:13:25,520 --> 00:13:29,760
a lot of the things are in that particular nature. There's a small piece of a product that's going

174
00:13:29,760 --> 00:13:35,760
to change that might seem like it's going to change the way things work. It's really hard. A lot of

175
00:13:35,760 --> 00:13:43,520
these have to be looked at in general. Maybe there's one huge thing that was NDA-ish. Actually,

176
00:13:43,520 --> 00:13:47,520
before the summit that I found out during the summit, no, they're actually telling everybody it's

177
00:13:47,520 --> 00:13:53,440
out there is co-work. My current, if there's a junky tool that I'm right now completely addicted to,

178
00:13:54,000 --> 00:14:00,320
the co-work, all of a sudden I don't have to use GitHub, Copilot Pro to do any major stuff. I can actually

179
00:14:00,320 --> 00:14:06,000
just do it in my frontier co-work. That's available. Everyone knows it's there. There's no secret

180
00:14:06,000 --> 00:14:11,280
itself there. It's frontier. What will it look like when it's finally done? Well, maybe that could be

181
00:14:11,280 --> 00:14:17,200
maybe not something you can talk about, but it doesn't matter. It's like the idea of it is already

182
00:14:17,200 --> 00:14:20,160
out there. So I can talk about that to my heart's content, fortunately.

183
00:14:22,240 --> 00:14:31,360
I think Microsoft starts at the start, sell Copilot as productivity tool. For me,

184
00:14:31,360 --> 00:14:41,680
it was not so productivity. It doesn't help me so much. It's nothing for me. No slides, no

185
00:14:41,680 --> 00:14:53,440
excel, yes, yes, yes, which copilot capabilities are useful now and production?

186
00:14:53,440 --> 00:15:00,880
So production wise, the bigger one, especially through copilot studio, there are certain things that

187
00:15:00,880 --> 00:15:05,440
perhaps used to be in the realm of you had to be an Azure, you had to add a web app, you had to

188
00:15:05,440 --> 00:15:11,520
have the bot framework and all of those kind of systems working. All of a sudden now in copilot

189
00:15:11,520 --> 00:15:18,000
studio, through itself, through its multi orchestration capabilities and some of the other elements,

190
00:15:18,000 --> 00:15:26,960
through the MCP tooling that is out there is just ridiculously powerful. They seem to be coming out

191
00:15:26,960 --> 00:15:31,520
with an MCP every time I look around. There's a new MCP and for that matter, anyone can make one of

192
00:15:31,520 --> 00:15:35,600
those because this is just open source type information for putting one of those things together.

193
00:15:35,600 --> 00:15:42,800
That suddenly makes a tool that was built to help, how do we say, the business users, the load coders,

194
00:15:42,800 --> 00:15:50,480
the makers. That same tool is what people can use now to build enterprise level capability.

195
00:15:50,480 --> 00:15:58,400
You're now supporting much larger scales and you're doing it essentially with that little package

196
00:15:58,400 --> 00:16:03,200
cost over here. Now, to be sure, there's other hidden elements there if you're trying to do something

197
00:16:03,200 --> 00:16:08,320
on the enterprise, you need to go further, you scale out, you have to touch the Azure sites,

198
00:16:08,320 --> 00:16:12,960
but here's the thing, before you had to do that almost the whole package Azure. Now,

199
00:16:12,960 --> 00:16:20,560
almost the entire project can be controlled or hosted, so to speak, from the copilot studio side,

200
00:16:20,560 --> 00:16:25,040
and now when you have to do a nuance and you have to do something that's really strong underneath

201
00:16:25,040 --> 00:16:29,760
the covers, that's the time you go pro code. And if you're a pro coder, you can just go under the covers

202
00:16:29,760 --> 00:16:34,000
and look at the stuff anyway. This is the part I like what they did in the copilot studio is like,

203
00:16:34,000 --> 00:16:40,240
they allowed the pro coders to not feel that they have to drag and drop everything. If they feel

204
00:16:40,240 --> 00:16:44,640
like writing code somewhere, go ahead, write the code over here and now we'll connect it. Hey,

205
00:16:44,640 --> 00:16:49,360
let's make it even better. Use your own claw tool to actually write the code. It's like, whoa,

206
00:16:50,160 --> 00:16:55,360
wow, you're accommodating everybody, aren't you? So I think that's the big, the big plus that they have

207
00:16:55,360 --> 00:17:02,000
right now is the fact that this was a low, what was considered low code, you know, personal productivity,

208
00:17:02,000 --> 00:17:07,520
now could actually do for real enterprise stuff. And I've actually seen some applications running

209
00:17:07,520 --> 00:17:15,600
in that particular way. I think there are some really cool examples or showcases from Microsoft.

210
00:17:15,600 --> 00:17:24,720
Yes. And I think a lot of companies work copilot well, but I think there are some companies,

211
00:17:24,720 --> 00:17:32,480
yeah, that's not so well. Well, yes. What are some common misconceptions

212
00:17:32,480 --> 00:17:38,400
organizations have before deploying copilot? So a part of the misconceptions is they think they

213
00:17:38,400 --> 00:17:43,600
think that if they're doing maybe the low code and they're doing the, I paid one price and I've

214
00:17:43,600 --> 00:17:49,200
get all I can eat, quote, unquote, for that particular price. The misconception is they're going to do

215
00:17:49,200 --> 00:17:54,480
that and they're going to save money. They're going to save time and they're going to get what they

216
00:17:54,480 --> 00:18:00,080
wanted done. Essentially, they're thinking they're going to win Formula one with their, with their

217
00:18:00,080 --> 00:18:05,680
Honda because they have now this capability of putting a great engine into it. It's like, well,

218
00:18:05,680 --> 00:18:09,440
that's not the way it works. They're sort of like,

219
00:18:10,240 --> 00:18:16,400
misconstrueing what one thing can do to solve the other thing that was never meant to be solved

220
00:18:16,400 --> 00:18:20,960
in that particular way. So this is where some people get into trouble. They'll see a demo and

221
00:18:20,960 --> 00:18:26,560
they believe it applies to them. And again, not ironically, the same tools can tell them it's like,

222
00:18:26,560 --> 00:18:30,480
well, yeah, I see that you want to use Power Platform to do that, but you know, that's not what

223
00:18:30,480 --> 00:18:35,680
Power Platforms for. Well, who could have told me that? Well, but Microsoft says it can do that.

224
00:18:35,680 --> 00:18:41,840
Well, you saw it there, but you probably had to apply it in this particular context where some companies

225
00:18:41,840 --> 00:18:47,040
not exactly miss, especially when you have a committee of so many people and this is not AI's fault

226
00:18:47,040 --> 00:18:52,080
or anything else or even the people's fault is the fact that you have all these ideas you want to

227
00:18:52,080 --> 00:18:57,760
try to everyone wants to get pleased. Everyone wants to be made happy. And in so doing, you make everyone

228
00:18:57,760 --> 00:19:04,160
else unhappy. So you got to just stop and think and sometimes you just have to have the one

229
00:19:05,520 --> 00:19:12,000
person that can see it all and bring it together. You were talking about having the big picture in

230
00:19:12,000 --> 00:19:16,560
there. That is the key. The big picture do make sure you're not going to go stray.

231
00:19:16,560 --> 00:19:28,320
And can companies before they enable and co-pilot, co-pilot, sorry, is there anything they

232
00:19:28,320 --> 00:19:36,320
should review first? So a lot of it is prepared, a preparedness going in. I think Microsoft and

233
00:19:36,320 --> 00:19:40,880
everybody has been saying this for AI is like you don't just turn on AI. You want to make sure you're

234
00:19:40,880 --> 00:19:51,040
ready. And here's not a, again, a surprise that to be ready for AI, essentially bringing in a creature

235
00:19:51,040 --> 00:19:56,800
that is going to be able to do what you can do at scale, like a thousand use, a million use,

236
00:19:56,800 --> 00:20:02,880
a billion use, all right? Is if you didn't have your data house in order, you didn't have your

237
00:20:02,880 --> 00:20:06,880
security in order, the things that people have been talking about for years and years and years

238
00:20:06,880 --> 00:20:12,640
since the start, if that's not in order, then when you bring it in, that's the first thing that you're

239
00:20:12,640 --> 00:20:18,880
going to notice breaks, essentially what it does. Who said this? Someone said that when you drink

240
00:20:18,880 --> 00:20:24,640
alcohol, it basically shows what your actual personality is, situation. It's sort of like that.

241
00:20:24,640 --> 00:20:28,720
It's like you're going to get drunk with the power and they when you, but the active getting drunk

242
00:20:28,720 --> 00:20:36,240
will reveal, let's just say all your potential failings, your gaps, other issues that are out there.

243
00:20:36,240 --> 00:20:40,320
What is the way to get into copilot? They give you the tools. They tell you what the tools are.

244
00:20:40,320 --> 00:20:44,240
There's some things out there that they said, oh, that costs extra money to be able to do that.

245
00:20:44,240 --> 00:20:48,400
They say, well, here's the thing. We always tell people, oh, you don't want to do purview.

246
00:20:48,400 --> 00:20:52,720
You don't want to do that particular stuff. I don't want to pay for e5 or what's going to be

247
00:20:52,720 --> 00:20:57,360
any one of those things. Here's the thing. You don't have to necessarily pay that. Maybe you don't

248
00:20:57,360 --> 00:21:02,320
want to pay that. You feel it. That's too much, but you have to do something that does what those

249
00:21:02,320 --> 00:21:07,760
things do. Whatever you do, you should have some sort of security. Whatever you do, you cover yourself.

250
00:21:07,760 --> 00:21:12,320
Whatever you do, you have, you should have policies at the very least if you're not going to pay someone

251
00:21:12,320 --> 00:21:17,440
to give you a product that does the thing. If there's a preparedness, that's the first one because

252
00:21:17,440 --> 00:21:22,160
that's the one where things usually break down. That's the first one. After that, it just becomes

253
00:21:22,160 --> 00:21:27,920
knowing when and how to use it. There's a lot of documentation about when and how to use it. To

254
00:21:27,920 --> 00:21:34,800
be sure, the tools now can tell you, if I'm going to put something together, for example, in Copa

255
00:21:34,800 --> 00:21:40,000
of Studio, I'm not going to ask it to make me a prompt to make me a good Copa of

256
00:21:40,000 --> 00:21:44,880
Peace, but some people think that's how to do it. Actually, I'll ask it before. Is it even a good idea

257
00:21:44,880 --> 00:21:49,200
to do this in Copa of Studio? I'll ask the Copa of Pilot. It might say, no, would you probably

258
00:21:49,200 --> 00:21:54,240
should do something else after I give it the information. It's like, aha, I'm already doing that in

259
00:21:54,240 --> 00:21:59,920
advance. Am I ready to do this? Sometimes that's even a question you can ask. There's tooling now that

260
00:21:59,920 --> 00:22:05,120
can make that happen. Again, it pays that you've done what you should have done in the first place.

261
00:22:05,120 --> 00:22:11,920
Set up your security, set up your data policies, your data governance, the classifications and all

262
00:22:11,920 --> 00:22:16,800
that stuff. People are saying, oh, we're only just doing this now. Well, then don't complain. You

263
00:22:16,800 --> 00:22:23,360
waited. You procrastinated this long. So if things are going to be a little bit hard, this is the

264
00:22:23,360 --> 00:22:29,200
price you have to pay to go through. I know it's a little harsh, but better that you finish and

265
00:22:29,200 --> 00:22:33,280
get through that. And after a while, you should be able to at least get on the other side and at

266
00:22:33,280 --> 00:22:37,520
least feel confident that whatever is happening is not going to bring down your organization.

267
00:22:37,520 --> 00:22:46,400
Yeah, I had a client who has, by the test run, let's go pilot.

268
00:22:46,720 --> 00:22:54,960
Some license and then he say, wow, it's 30 bucks per per people and say, that's too expensive.

269
00:22:54,960 --> 00:23:03,600
Ask about this. And I'll say, okay, we see they save 10 hours every week. How much did you pay then?

270
00:23:03,600 --> 00:23:15,280
Less than $3 for loss per hour. What? Yeah, I think when we see what we get on production,

271
00:23:15,280 --> 00:23:24,640
it's so, so amazing. And there are also, I think, on LinkedIn, we have two kind of people I've

272
00:23:24,640 --> 00:23:29,840
found. The people they say, there's absolutely no risk with scope pilot.

273
00:23:29,840 --> 00:23:36,480
Well, and we ask the other, they say, oh, it's the risky thing in the world.

274
00:23:36,480 --> 00:23:43,680
Yes. What are the biggest security risk organizations overlook?

275
00:23:43,680 --> 00:23:47,680
The biggest security risk is a good one. The business is the same risk you had before.

276
00:23:47,680 --> 00:23:54,960
Sorry to say, but it's the people. The one that says, hey, you're going to win a million dollars,

277
00:23:54,960 --> 00:24:01,920
click here. And no matter what technology you have, if you click here, yes, what?

278
00:24:01,920 --> 00:24:09,040
It's going to, this is the biggest risk. And it's sometimes like, there's people who have somehow

279
00:24:09,040 --> 00:24:14,880
taken the training every year for, you probably do this every time you enter into a client organization

280
00:24:14,880 --> 00:24:20,720
that gives you like a little thing that you have to at least watch a video or do a training to make

281
00:24:20,720 --> 00:24:25,040
sure that you're not going to give up their secrets and everything else. And it might be tedious

282
00:24:25,040 --> 00:24:29,920
and everything else. But you know what? I think it's as they say, important for people to remember

283
00:24:29,920 --> 00:24:35,040
that no matter how often or how smart you might think you are, there's still someone that's going

284
00:24:35,040 --> 00:24:42,960
to find your weakness somewhere. And I have been caught once or twice on a great fishing attempt.

285
00:24:42,960 --> 00:24:49,040
Thank goodness they were like somebody in IT trying to see if they could come up with a way to do it.

286
00:24:49,040 --> 00:24:53,680
Well, here's the thing. As I say, if they know everything enough about you, they will get through.

287
00:24:53,680 --> 00:24:59,200
The biggest, the biggest risk, of course, in the end of the day, are the people that used the tools.

288
00:24:59,200 --> 00:25:04,320
If they're not responsible, these are very powerful tools. We can try putting safeguards. In fact,

289
00:25:04,320 --> 00:25:09,280
the Microsoft Copilot side of things as safeguards were like if they try to do things

290
00:25:09,280 --> 00:25:14,400
sort of nefariously, I need to cheat on this test. I need to do stuff. If it detects those kind of

291
00:25:14,400 --> 00:25:20,320
things, it attempts to block you, to stop you or to fall down. In fact, one of my things that the

292
00:25:20,320 --> 00:25:26,160
Power Platform Conference was a, I did like a little show called Whose Prop is It Anyway? Where,

293
00:25:26,160 --> 00:25:30,320
of course, it's in Vegas, you can say whatever, but most importantly, if you can say something that

294
00:25:30,320 --> 00:25:35,920
makes Copilot cry or stop you win points. Well, yes, there's a way to make Copilot say I will not

295
00:25:35,920 --> 00:25:41,280
talk to you again ever, at least back then, because you could give it a prompt that essentially tells

296
00:25:41,280 --> 00:25:45,600
it to be as evil as possible. I'm not going to tell you what that prompt is. It might still or

297
00:25:45,600 --> 00:25:51,120
may not work anymore or might have been caught, but this is the idea. The human can make the worst

298
00:25:51,120 --> 00:25:56,400
decisions and make it do very bad things. And we're sort of seeing that in the world today a little

299
00:25:56,400 --> 00:26:00,960
bit. If they're again, the biggest risk, the human in the end of the day, how can you protect it?

300
00:26:00,960 --> 00:26:07,760
The governance is your first, first measure to stop it. The next one, of course, is this means the,

301
00:26:07,760 --> 00:26:14,400
how you be, how you are with your people is actually more important than, or as important as ever.

302
00:26:14,400 --> 00:26:20,400
Be good to your people, have good people, do whatever you can for the people to be in the,

303
00:26:20,400 --> 00:26:25,440
in the way that will work well with the, with the technology, whatever it is.

304
00:26:26,400 --> 00:26:34,000
And then there are some, some tools are, I think, the defender for Co-Pilot, where is,

305
00:26:34,000 --> 00:26:46,240
my favorite for it's, it's pure view. Yes. What, think you, what tool helps most to get these AI

306
00:26:46,240 --> 00:26:54,640
readiness in Co-Pilot? Ah, AI readiness, there, there are some nice third party tools that are out there.

307
00:26:54,640 --> 00:27:01,920
And a lot of cases, AI readiness, I think Microsoft's own tools that get you there,

308
00:27:01,920 --> 00:27:08,320
purview to me is one of the key ones. Not exactly a, a cheap tool to get it in place,

309
00:27:08,320 --> 00:27:13,680
but it's almost like it's that kind of tool, whether you use purview or something that does what

310
00:27:13,680 --> 00:27:18,160
purview does is sort of mandatory. Before purview existed, there was quite a few third parties that

311
00:27:18,160 --> 00:27:23,120
went in there. I mean, even all the, the tools that you might have heard of just that we used to work

312
00:27:23,120 --> 00:27:27,600
with SharePoint, for example, it's the share gate tool, the ab point tools, the, all the other,

313
00:27:27,600 --> 00:27:32,640
folks that are in that particular area, does kind of tooling as they say, pay the extra,

314
00:27:32,640 --> 00:27:38,800
pay it a little bit extra to get the piece of mind. And I find that those tools work all very well.

315
00:27:38,800 --> 00:27:43,280
I haven't seen one of those that like, oh, it leaves a lot of gaps and everything else. They usually

316
00:27:43,280 --> 00:27:48,640
get you most of the way there. Again, most of the way there means that they're still the work of the

317
00:27:49,280 --> 00:27:55,760
the organization to do to close the gap. So if the tools say, do this, you better do it. And if you don't

318
00:27:55,760 --> 00:28:01,440
then sign over here, I didn't want to do it. Okay. And that's, that's on you. You didn't do it.

319
00:28:01,440 --> 00:28:10,400
Follow the tool. Yeah. Often I see that people use, I don't know, a tool and they

320
00:28:10,400 --> 00:28:17,760
the losing a little bit of awareness. Yes. Because then they have a tool for, I don't know,

321
00:28:17,760 --> 00:28:22,240
for governance and teams. And all the other tools are not guaranteed.

322
00:28:22,240 --> 00:28:29,920
One drive or, or I don't know, streams and what we all have. So I think it's also a little bit

323
00:28:29,920 --> 00:28:38,240
risky, but there are, I like, what I use are true tools. I like, like, Rancor, I don't know,

324
00:28:38,240 --> 00:28:42,240
it's really cool. Oh, Rancor, of course. My friends from Rancor, I remember those guys. Yes.

325
00:28:43,280 --> 00:28:49,440
Yes. They're actually really good too. I remember. And reminding me of Rancor. And the script,

326
00:28:49,440 --> 00:28:55,760
script, runner, I also like the causes. You can do all this directly to the graph. Ape and

327
00:28:55,760 --> 00:29:01,440
yes. Yes. It was scripting and and and it's, yeah. Okay. It's, it's more for the script.

328
00:29:01,440 --> 00:29:07,440
No, I'm still script person. Is that people don't realize that the only reason why I do low

329
00:29:07,440 --> 00:29:21,440
codes, because I got lazy. That's all. Yeah. But when we think, I think a, a eye policies and

330
00:29:21,440 --> 00:29:27,440
what your recommendations around a eye policies and responsible use.

331
00:29:27,440 --> 00:29:34,880
It's, it's going to the next level, right? There is, there's the, the, the first thing about

332
00:29:34,880 --> 00:29:41,120
A policies is of course, understand where the industry is at that, that, at that point. It's not

333
00:29:41,120 --> 00:29:47,120
just the responsible AI piece that Microsoft puts out my own company with Cap Jam and I and

334
00:29:47,120 --> 00:29:52,720
Sugetty. We have, we had our own manifesto, but even before Microsoft came up with it, on what

335
00:29:52,720 --> 00:29:58,720
responsible AI and what governance should look like for that particular area. It was actually quite,

336
00:29:58,720 --> 00:30:02,960
the, it's still actually out there. It's still valid and I'll detail. And there's quite a few of

337
00:30:02,960 --> 00:30:08,080
those now out there in the industries, in the industry. There's the well-architected pieces that

338
00:30:08,080 --> 00:30:13,920
we look at on following these kind of policies going through. At the end of the day, it is the

339
00:30:13,920 --> 00:30:21,040
organization that decides how they're going to be towards those policies. And my, my, my point,

340
00:30:21,040 --> 00:30:26,880
the, with a lot of the, the companies is like, do, as I say, do one to others is the, is the mantra

341
00:30:26,880 --> 00:30:32,720
over here. If, if you don't want something done to you in a certain way, then don't do it to them.

342
00:30:32,720 --> 00:30:38,160
If you want, if you want to allow something to happen to you, then don't allow it from there either.

343
00:30:38,160 --> 00:30:43,360
It works in the reverse. What's the first policy that you have to put into place? If you're not sure

344
00:30:43,360 --> 00:30:49,200
what that is, then that's a bit of the homework over here. At some level, I believe every organization

345
00:30:49,200 --> 00:30:54,960
should have maybe a compliance officer or perhaps on the, at these, at the CIO level, essentially,

346
00:30:54,960 --> 00:31:02,080
on the CIO level, especially. There should be some sort of person in there that looks at that

347
00:31:02,080 --> 00:31:08,320
particular aspect that is responsible for that, that is the author of, or the controller, at least,

348
00:31:08,320 --> 00:31:13,600
of the information that goes through that. Again, if it's not a, and there was a person because it's a large

349
00:31:13,600 --> 00:31:19,040
organization, it's a function that needs to happen. You were talking about, like, you have the tools

350
00:31:19,040 --> 00:31:22,800
over here, but if you don't see what the tools are doing or you're, you're relying on the tools,

351
00:31:22,800 --> 00:31:27,440
you feel like you feel like they're safe. It's not just safe just by turning on the tool. That's the

352
00:31:27,440 --> 00:31:33,200
other, the other, here's the other part about AI responsibility. It's like software responsibility.

353
00:31:33,200 --> 00:31:38,640
If you let things run and you claimed that they're in monitors, say, "Oh, I have Azure Insights.

354
00:31:38,640 --> 00:31:43,600
It's watching everything. I have the SRE agent from Azure and it's going to tell me what

355
00:31:43,600 --> 00:31:48,480
something goes wrong and proactively fix things." Well, that's all fine and well, but who's the

356
00:31:48,480 --> 00:31:54,080
human watching over that stuff? Who's looking at the dashboards? Who's reading those reports?

357
00:31:54,080 --> 00:31:58,560
What's your cadence? When do you look at them? Only when something goes wrong? Or are you looking at

358
00:31:58,560 --> 00:32:04,560
them every day, every week? What's your cadence? If there's a one responsible policy on AI,

359
00:32:04,560 --> 00:32:10,240
is that you need to review your reports, those monitors every so often. And whenever you can,

360
00:32:10,240 --> 00:32:16,160
ask for insights. It's like, "Hmm, what insights can I get from this today?" This existed for quite a

361
00:32:16,160 --> 00:32:21,760
while. I remember when performance monitor came out on the Microsoft side, it was the concept was

362
00:32:21,760 --> 00:32:27,520
to get insights that you didn't think of asking before. Well, that was old technology already.

363
00:32:27,520 --> 00:32:32,560
That's about three, four years old now. Guess what? There's more power now. So you could do that.

364
00:32:32,560 --> 00:32:38,240
A responsible AI person doesn't sit there and rest on their laurels. They're constantly looking

365
00:32:38,240 --> 00:32:43,600
to see what they can, that's much better, but what they can prevent also.

366
00:32:45,120 --> 00:32:54,720
Yeah, we have, I think, when we go to build co-pilots, I think we have two bigger options. We have

367
00:32:54,720 --> 00:33:02,240
the co-pilot studio. We have AI Foundry. Yes. Microsoft Foundry, yes.

368
00:33:02,240 --> 00:33:10,480
How are the difference between both and when should I use what?

369
00:33:10,480 --> 00:33:14,720
What's it? It's a good question. It's a solution. One of my favorite questions because we always get

370
00:33:14,720 --> 00:33:19,520
that. We get that a lot. I get that a lot on when to use what tool. In fact, if you want to think about it

371
00:33:19,520 --> 00:33:24,560
on the beginning end is like, "When do I use app builder? The little baby I can make an app and I

372
00:33:24,560 --> 00:33:30,800
can look like I'm very good all the way to AI Foundry." Or Microsoft Foundry is there reminding me what

373
00:33:30,800 --> 00:33:39,520
to call it. Here's the thing. The AI, the Microsoft Foundry is sits in Azure. This is the deep dive I

374
00:33:39,520 --> 00:33:45,120
want to tweak everything that I can tweak, have full control to do whatever I need. In the end,

375
00:33:45,120 --> 00:33:52,720
still get the agent or the pieces that I can deploy to wherever, in this particular case, the one

376
00:33:52,720 --> 00:34:00,240
target that I'm looking at is M365. Co-pilot studio is more for what's happening in M365, even though it

377
00:34:00,240 --> 00:34:06,400
can create the endpoints, it can be used elsewhere. Co-pilot studio essentially pre-packages quite a

378
00:34:06,400 --> 00:34:11,840
bit of those pieces. In other words, it takes care of a lot of the plumbing and it makes it's right away

379
00:34:11,840 --> 00:34:17,600
ready to connect and deal with pieces of the enterprise that are within the M365 tenants.

380
00:34:17,600 --> 00:34:23,840
That's the piece of the belong too. I can't make a co-pilot studio piece and operate multiple tenants.

381
00:34:23,840 --> 00:34:30,640
However, I could make a Microsoft Foundry piece that could be existing in multiple tenants,

382
00:34:30,640 --> 00:34:36,000
multiple customers, multiple areas. So right off the bat, it's like, where is your stuff going to live?

383
00:34:36,000 --> 00:34:40,320
And then the other one is like, how much power do you really need to tweak? Maybe I want to

384
00:34:40,320 --> 00:34:45,840
tweak down to the L-alam and multiple L-alam and orchestrate these pieces and MCP servers. You're probably

385
00:34:45,840 --> 00:34:50,560
doing it in the Azure. If I'm like, I just want to connect these pieces of the enterprise together

386
00:34:50,560 --> 00:34:58,640
and make sure that I fulfill this particular workflow you're fine in the co-pilot studio. And if you

387
00:34:58,640 --> 00:35:01,280
just want to do something for yourself, then you go up one level.

388
00:35:03,520 --> 00:35:09,040
I think there are also two, I say buzzwords. That's okay.

389
00:35:09,040 --> 00:35:16,640
One are custom co-pilots and the other are agents. What are the different between both and when

390
00:35:16,640 --> 00:35:27,040
you use what? Ah, who? This is interesting. The custom co-pilot, when you're making an agent,

391
00:35:27,840 --> 00:35:34,480
all right. An agent is just, these are just terms. And here's the, maybe it's not ironic that the product,

392
00:35:34,480 --> 00:35:40,880
a co-pilot, what co-pilot studio used to be called was Power Virtual Agents. And so almost like we

393
00:35:40,880 --> 00:35:45,280
came full circle and said, oh, what's it doing? It's making agents. It's not making bots.

394
00:35:45,280 --> 00:35:49,440
For a while, we were calling it bots. And then all of a sudden it started being calling agents.

395
00:35:49,440 --> 00:35:54,320
It's like, oh, I think agents is going to be the thing. So then they call it co-pilot studio.

396
00:35:54,320 --> 00:35:59,520
It's like, well, why do they do that? Because we don't want to call it co-pilot, bot maker or something.

397
00:35:59,520 --> 00:36:04,160
I guess I don't know what they did. But this was, this was crazy over here. So what's the difference

398
00:36:04,160 --> 00:36:12,480
between a custom co-pilot versus an agent? A custom co-pilot is more like, here's your main chat

399
00:36:12,480 --> 00:36:16,400
interface. And I've actually seen, quote, and, quote, custom co-pilot's not called that.

400
00:36:16,400 --> 00:36:23,280
So remember what co-pilot is. Here's your interface. You ask it whatever you make it, you make it

401
00:36:23,280 --> 00:36:30,160
go through whatever kind of statement. It's not focused necessarily on a particular topic.

402
00:36:30,160 --> 00:36:36,560
But it's meant to handle a much bigger scale of information and it's customized. Maybe it's

403
00:36:36,560 --> 00:36:41,680
customized to never allow you to ask this kind of question ever. Maybe it's customized to work

404
00:36:41,680 --> 00:36:47,120
on the organization in a particular way. To which you could tweak cost of co-pilot in general to do that.

405
00:36:47,120 --> 00:36:52,720
But this, you want to go beyond that. So now you have a custom co-pilot. The agent is just simply

406
00:36:52,720 --> 00:36:56,880
something that's more focused. And by the way, you don't want to have an agent that does everything.

407
00:36:56,880 --> 00:37:00,880
An agent that can answer any kind of question. That's not the point of the agent. The agent is

408
00:37:00,880 --> 00:37:04,640
supposed, if I get an agent, I have an agent that's going to handle, oh, travel arrangements. I have

409
00:37:04,640 --> 00:37:11,760
an agent that's going to do my tasks for this particular area. All right. That's the idea. It's

410
00:37:11,760 --> 00:37:16,880
that the agent should be focused while the custom co-pilot should be like the bigger area.

411
00:37:16,880 --> 00:37:21,680
So if you think about it, I have a business with so many things I can do. There's your custom co-pilot.

412
00:37:21,680 --> 00:37:24,560
I have the people that are doing the work. Those are the agents.

413
00:37:24,560 --> 00:37:33,360
And did you see some limitation? We have talked about co-pilot studio.

414
00:37:33,360 --> 00:37:36,080
Seeing some limitation in co-pilot studios today.

415
00:37:36,080 --> 00:37:44,960
So there's quite a few limitations that would have you like go outside of it to make the changes.

416
00:37:44,960 --> 00:37:50,880
I mean, some of the key limitations is, well, if there's any one limitation, it's the course of the

417
00:37:50,880 --> 00:37:55,040
fact that it's only going to work on your tenant. It's going to work in the context of what you're

418
00:37:55,040 --> 00:38:01,360
trying to do in M3C5. Is that a limitation for some people? Yes, it could be considered a limitation.

419
00:38:01,360 --> 00:38:08,080
Current limitations include the fact that once you select an LLM for its activity behind the scenes,

420
00:38:08,080 --> 00:38:14,320
it can only use that LLM. So if it's if you're if something is better done, maybe with an

421
00:38:14,320 --> 00:38:22,000
on-tropic LLM versus say a GPT or something else, you can exactly switch. All right, if you had to do

422
00:38:22,000 --> 00:38:28,480
that currently, you would have to maybe connect separately on a different agent, as they say within,

423
00:38:28,480 --> 00:38:33,760
that actually connects to the different LLM. That was the only way to really do it. But if you can

424
00:38:33,760 --> 00:38:39,120
imagine that current limitation likely won't be a limitation for long. Here's the thing. If I

425
00:38:39,120 --> 00:38:44,480
mention a limitation now, probably in three, two, one, Microsoft will find a way to get past that's

426
00:38:44,480 --> 00:38:50,800
limitation. So by the time we hear this again, the limitation that I was talking about, I don't know,

427
00:38:50,800 --> 00:38:54,960
I'm not giving in these secrets. Maybe that limitation is not going to be there. Maybe there'll be

428
00:38:54,960 --> 00:39:03,520
a way to do more choose choose the right LLM for the right job for the right thing within

429
00:39:03,520 --> 00:39:08,800
a copo studio. I know that you can do that in Azure. That means sorry in the in the Microsoft

430
00:39:08,800 --> 00:39:15,520
Foundry, technically. So why not in the copo studios? If it's not there, so because they haven't got to it yet,

431
00:39:15,520 --> 00:39:22,640
probably. And no secrets over here, you can think of what a limitation could be. Maybe it's like

432
00:39:22,640 --> 00:39:27,840
the number of things that happen through the tools. I'll give you a good example here. So the

433
00:39:27,840 --> 00:39:33,840
limitation is not copo studio, but I decided to use power automate or maybe a power app within to do

434
00:39:33,840 --> 00:39:38,480
something and that has a limitation that says only two thousand records at a time or something on that,

435
00:39:38,480 --> 00:39:43,920
all of a sudden it has a code and code limitation. But that limitation wasn't really copo studios.

436
00:39:43,920 --> 00:39:49,600
It was the architect that decided what of copo studio they were going to do that caused the limitation.

437
00:39:49,600 --> 00:39:55,040
And maybe it was because they had no choice. Okay, so we'll copo studio fix that. No,

438
00:39:55,040 --> 00:40:00,960
that limitation to go by studio. Not exactly, but that's the limit that you chose that architecture that

439
00:40:00,960 --> 00:40:07,760
will force you into those decisions. All right, so again, for what we do over here, we want to make

440
00:40:07,760 --> 00:40:12,560
sure that we're open, we understand that there's interconnectedness. If there is a limitation,

441
00:40:12,560 --> 00:40:20,240
there's always a way around it. Yeah, when we think about autonomous agents of the

442
00:40:20,240 --> 00:40:28,400
Microsoft ecosystem, how many of them are there are more to seven year old schoolboy or this more

443
00:40:28,400 --> 00:40:37,360
the PhD guy? Oh, dear. Yeah, so that was one of them. In fact, that's one of those features that

444
00:40:37,360 --> 00:40:44,080
came out quickly. The fact that you could do autonomous agents just by simply putting a trigger,

445
00:40:44,080 --> 00:40:48,400
it's almost like, oh, that sounds like something they had before. Oh, what was called power automate,

446
00:40:48,400 --> 00:40:53,600
where something happens, it's triggered, something will follow through. So now the autonomous agents

447
00:40:53,600 --> 00:40:59,840
are now being triggered. What it almost feels like they're being proactive. So a proactive, what's a

448
00:40:59,840 --> 00:41:05,520
proactive agent? I think those are the ones that are more of the concern. An autonomous agent is just

449
00:41:05,520 --> 00:41:12,880
simply rack going on a trigger, something told it to do its thing. It's doing its thing. It's no different

450
00:41:12,880 --> 00:41:19,840
than any automation. So the autonomous agent is simply maybe a smarter automation, maybe it will

451
00:41:19,840 --> 00:41:25,600
reason and say, oh, I was triggered. And this time, I'm going to use reasoning to decide what my next steps

452
00:41:25,600 --> 00:41:30,960
are going to be and what I should do as opposed to do this step, step, step, step, step, step.

453
00:41:30,960 --> 00:41:35,360
Correct. So there's your autonomous agent. But now comes the next part. It's like, well,

454
00:41:35,360 --> 00:41:42,000
the autonomous agent wasn't triggered, but the trigger is, I see a condition over here and I'm

455
00:41:42,000 --> 00:41:48,320
always looking, always watching, always watching like the cartoon. And then they said, oh, I see a

456
00:41:48,320 --> 00:41:54,800
condition over here that was not a part of what you asked me originally to do other than look this

457
00:41:54,800 --> 00:42:01,600
over. And I feel or I see or I assume or whatever it is somehow it comes up to the idea that it

458
00:42:01,600 --> 00:42:07,280
found the problem that it can fix or deal with. And perhaps it's going to run. That's the kind of

459
00:42:07,280 --> 00:42:13,120
agent that we're like, oh, yikes, how is that going to work? How do you test such a beast? How do you

460
00:42:13,120 --> 00:42:18,400
protect from such a beast making it maybe a rash decision? The human in the loop is now, we haven't

461
00:42:18,400 --> 00:42:22,800
talked about yet, this yet, the concept of the human in the loop is any of those kinds of agents you

462
00:42:22,800 --> 00:42:28,800
want to make sure has that human in the loop aspect of it that does not allow it to proceed.

463
00:42:28,800 --> 00:42:34,880
And that's much autonomously, but irresponsibly, I guess is the best way to put it.

464
00:42:36,480 --> 00:42:43,200
I think that's what we're looking at when we're trying to see what we can do to keep the so-called

465
00:42:43,200 --> 00:42:49,840
proactive agent capabilities that are clearly available. We've seen already not in the Microsoft

466
00:42:49,840 --> 00:42:55,840
necessarily, but in the other tooling, this happened. I think there was like, I saw something

467
00:42:55,840 --> 00:43:03,120
not too long ago where they sort of added avatars of the different LLMs and they were like in towns

468
00:43:03,120 --> 00:43:09,920
and they had to coexist and various kinds of outcomes happened based on how they act with each other.

469
00:43:09,920 --> 00:43:16,720
It's like, whoa, the idea that if they, uh, honestly or on their own come up with certain kinds of

470
00:43:16,720 --> 00:43:22,400
decisions in a certain way because there's no human to guide them, then things would happen. A lot

471
00:43:22,400 --> 00:43:27,680
of those cases, by the way, all ended up with bad outcomes, every single one as far as I know, without

472
00:43:27,680 --> 00:43:34,480
the human in, expect a bad outcome. Let's put it that way. Yeah, I like, I think Microsoft has these

473
00:43:34,480 --> 00:43:40,000
Co-Pilot Champion in their description. I like this concept. Where? Yeah. We'll leave

474
00:43:40,000 --> 00:43:49,760
people. What did you think for the future? We have, I don't know, for three years, Co-P,

475
00:43:51,360 --> 00:43:59,360
front engine hearing was the hype job and you're, I don't know, a six other thousand dollar for a year.

476
00:43:59,360 --> 00:44:06,640
This is a top one. You know, there was a time that I could see what was coming, right? Or it was,

477
00:44:06,640 --> 00:44:10,640
and it was clear also and, and remember, like I said, Microsoft telecasts what they're doing,

478
00:44:10,640 --> 00:44:16,640
quite a long time ahead and the MVP summit sort of gave us one year down the line, up to one year

479
00:44:16,640 --> 00:44:22,960
down the line. Right now, even that looks murky. They seem to come up with stuff so fast that I don't

480
00:44:22,960 --> 00:44:27,920
eat it. I don't think even our summit has enough time to give us that information. It's there coming

481
00:44:27,920 --> 00:44:32,400
up with stuff faster. The only way to actually be in the know is actually being in those projects

482
00:44:32,400 --> 00:44:38,240
or working with Microsoft directly. There's just too much information. So for, let's just, just

483
00:44:38,240 --> 00:44:44,000
for my, my area for the, for the power platform side, maybe there's six or seven workstreams of new

484
00:44:44,000 --> 00:44:50,000
stuff. I have no time to be in each one of them to see what each one of them does. I'd love to be

485
00:44:50,000 --> 00:44:55,280
because all of them seem to affect me, but I only have time maybe for one or two and I hope that I

486
00:44:55,280 --> 00:45:01,840
can hear something about one of those maybe on a PGI that they, they give us for the partners or for

487
00:45:01,840 --> 00:45:08,080
the, for the MVP's to at least hear that it's coming. So do I, can I see what's going to come up in

488
00:45:08,080 --> 00:45:11,600
in the next year? Other than the fact that if we're going at the trend, we're going,

489
00:45:12,400 --> 00:45:16,880
it feels like things that are whatever's limitation now, it doesn't make a nice PowerPoint would be,

490
00:45:16,880 --> 00:45:22,000
maybe one of those that people complain about a lot. I'm going to imagine that it better be making

491
00:45:22,000 --> 00:45:26,160
a great PowerPoint a year from now. Well, two years ago, I said it should make a better PowerPoint

492
00:45:26,160 --> 00:45:31,920
next year. It's still not doing it. So maybe I'm wrong, but this time I feel like it's finally got

493
00:45:31,920 --> 00:45:38,880
it happen. I'm going to see a beautiful PowerPoint maker in the next year. Awesome. We are talking about

494
00:45:38,880 --> 00:45:45,920
the PowerPoint and then we have this concept from Procode, low code, no code, wipe coding,

495
00:45:45,920 --> 00:45:54,640
makes it sense to be longer a Procoder or is it a job that will dying?

496
00:45:54,640 --> 00:46:00,720
Well, here's the thing. When we say Procode, they so called professional developer. The professional

497
00:46:00,720 --> 00:46:08,240
developer better not be, let's just say, very narrow-minded on how things should work. The professional

498
00:46:08,240 --> 00:46:14,640
developer is so called full stack, a person when they say full stack. That means full stack. It means

499
00:46:14,640 --> 00:46:22,720
you need to understand all the pieces around you and how and when to do what, where to do what,

500
00:46:22,720 --> 00:46:29,520
and how to go. You may have a specialty. I do.net. I do this kind of stuff. Okay, fine and well.

501
00:46:29,520 --> 00:46:35,040
The tools right now, essentially, the AI tools that are available to you now, the vibe coding and

502
00:46:35,040 --> 00:46:40,000
everything else allow you to go beyond that. But you as the professional developer should know how

503
00:46:40,000 --> 00:46:47,840
all of that stump comes together. It basically means that if you're a Procoder, then you really are

504
00:46:47,840 --> 00:46:53,600
a Procoder when you understand what it is that's being shown to you. Until then, you're still getting

505
00:46:53,600 --> 00:46:59,760
to that point. The low code is different. The low code over here has an objective. Wants to get

506
00:46:59,760 --> 00:47:04,720
a certain thing done. They're concerned not necessarily is what's going on under the covers.

507
00:47:05,200 --> 00:47:11,200
They just need to get from here to there and they got to get there efficiently and with the

508
00:47:11,200 --> 00:47:15,520
assurance that whatever it is that they get from here to there is going to do what it is that they do.

509
00:47:15,520 --> 00:47:22,000
And once it does it, they're done. All right. Again, what's the assurance? The assurance is as a

510
00:47:22,000 --> 00:47:26,800
low coder, I feel confident. If I put this together that if there's something under the covers,

511
00:47:26,800 --> 00:47:33,520
that's not right, they'll be someone else or something else that's going to be there to catch me.

512
00:47:34,240 --> 00:47:38,640
Going through. And that's, I think that's one of the big differences over here.

513
00:47:38,640 --> 00:47:44,800
If I'm doing some development, for example, for Power Platform, and they say,

514
00:47:44,800 --> 00:47:49,440
why are you doing Power Platform over here? Shouldn't that be a low coder type person? Why is a

515
00:47:49,440 --> 00:47:56,480
Procoder doing Power Platform? The Procoder, just like in the case of the administrative assistant

516
00:47:56,480 --> 00:48:02,080
using good old Microsoft Access because it's an easy tool. Yet here they are talking about data

517
00:48:02,080 --> 00:48:08,400
concepts that usually is in the realm of the data administrator. You're not probably going to come up

518
00:48:08,400 --> 00:48:14,640
with the best design necessarily. If the tooling is not there to support you to give you those

519
00:48:14,640 --> 00:48:22,560
smarts or to prevent you from making the mistakes that maybe a data administrator would understand.

520
00:48:22,560 --> 00:48:27,600
You have the same issue with low code. There are some things you might do because you have the power

521
00:48:27,600 --> 00:48:34,720
and then essentially do things that hurt. Again, it's training governance all of those pieces that come

522
00:48:34,720 --> 00:48:41,600
into play. We talked about it earlier that are needed to make that work. To me, if there's like,

523
00:48:41,600 --> 00:48:46,160
if you're going to a client project need to build something, it doesn't matter if it's Power Platform

524
00:48:46,160 --> 00:48:53,440
or Azure or anything like that, you're bringing a developer, that developer is a quote-unquote professional.

525
00:48:54,160 --> 00:48:59,840
So they will do things professionally in whatever tool that they're told to do it in or that they're

526
00:48:59,840 --> 00:49:05,680
they're going to be doing it in. They have to be professional. The other person who is that that's

527
00:49:05,680 --> 00:49:11,680
not their day job and they're doing something else, they should not be expected to do things at that

528
00:49:11,680 --> 00:49:16,880
professional level. They're going to do things with the tool as the tools designed to work.

529
00:49:16,880 --> 00:49:25,840
Finding well, we should not expect anymore. So we can say, I don't know, we can develop with the pro

530
00:49:25,840 --> 00:49:35,840
quota, we can develop company-wide solutions and the other things are more for my personal use.

531
00:49:35,840 --> 00:49:42,480
So yes, so actually this is a good, that's actually a good point. So sometimes a business user will

532
00:49:42,480 --> 00:49:48,160
come up with something that's like, hey, this is a great idea for the whole company and then they go

533
00:49:48,160 --> 00:49:53,280
ahead and using their tools come up with something that looks like, wow, this is great. I'm going to show

534
00:49:53,280 --> 00:50:00,160
it to my boss. Yeah, wow, we can now we can solve our end of month of finance, you know, a balancing sheet

535
00:50:00,160 --> 00:50:06,240
and all that stuff. And of course, that's not exactly a personal thing. That's your affecting the

536
00:50:06,240 --> 00:50:10,640
whole company by touching the balance sheet and you're doing all that stuff. So well, hold on a moment,

537
00:50:10,640 --> 00:50:17,920
you have to step back and say, okay, that was a great idea, son. You put together a system that's

538
00:50:17,920 --> 00:50:24,800
going to give you that. And then your manager and she looked it over and said, this is great, this is

539
00:50:24,800 --> 00:50:29,440
a good idea. I think it should be something you should implement. At that point then, it's up to

540
00:50:29,440 --> 00:50:34,640
the top leadership to say, you're touching something that's at the enterprise level. At that point is

541
00:50:34,640 --> 00:50:41,760
where the professional needs to come in at the very least to give the blessing that yes, it can be

542
00:50:41,760 --> 00:50:46,480
done. Let's package it so it's truly enterprise and it's maintained at the enterprise level and

543
00:50:46,480 --> 00:50:51,040
that person give them a reward, give them something over here, but take it away from them and give it to

544
00:50:51,040 --> 00:50:55,840
actual people who will now support and maintain it going forward because that person that had the

545
00:50:55,840 --> 00:51:01,040
day job, unless they feel like becoming a developer themselves to do it forever, that's, it shouldn't

546
00:51:01,040 --> 00:51:06,800
be for them. They came up with the idea. They should be rewarded and go on. If they want to still be like,

547
00:51:06,800 --> 00:51:12,000
I still want to be involved in it, find whatever you want to do to deal with that. That's fine, but at some

548
00:51:12,000 --> 00:51:17,120
point when something is enterprise, there needs to be the enterprise professionals that will run

549
00:51:17,120 --> 00:51:28,880
and maintain it and take it forward. This is cool. So I like to ask, have you, you have things seen a lot

550
00:51:28,880 --> 00:51:38,800
of co-pilot at kept Gemini or at your clients? What are one co-pilot that was cool or

551
00:51:38,800 --> 00:51:48,640
very special use case? Yes, seem. That's one. I haven't seen the one use case to rule them all.

552
00:51:48,640 --> 00:51:56,080
The thing is though, I deal with all industries at this point. So there are different ones that have

553
00:51:56,080 --> 00:52:01,760
the things that are good for them. It could be legal. It could be medical. It could be manufacturing. It

554
00:52:01,760 --> 00:52:09,120
could be retail. And they all have a lot of their own things that seem to be the use case that works

555
00:52:09,120 --> 00:52:13,760
for all. However, there are common use cases. And I'm going to call them the business use case.

556
00:52:13,760 --> 00:52:19,920
If there's one common set of use cases that I see for every single of the ones where they bring in

557
00:52:19,920 --> 00:52:24,160
their whole organization to us, but sometimes we only see one department maybe sometimes, right?

558
00:52:24,880 --> 00:52:29,520
If they bring the whole organization to us, the folks that seem to approach us that I've seen more

559
00:52:29,520 --> 00:52:36,320
in these at least workshops have been folks from HR, which is pretty interesting. It's like HR.

560
00:52:36,320 --> 00:52:39,920
And then you look at the Microsoft use cases. A lot of HR use cases. I have a feeling those

561
00:52:39,920 --> 00:52:47,280
same people are being approached by Microsoft, well, or vice versa. HR type use cases, finding out

562
00:52:47,280 --> 00:52:54,800
information. Maybe once I get paid, what are my bits? What can I get? Where can I get these

563
00:52:54,800 --> 00:52:59,680
things fixed? The things that have been forever and forever the reason for an internet from for years

564
00:52:59,680 --> 00:53:06,640
back, for everything else, it seems like those are the most common and the common use cases that

565
00:53:06,640 --> 00:53:13,040
are more personal, okay? HR is a personal thing. Human resources, you are the human and you are

566
00:53:13,040 --> 00:53:19,040
the resource. You probably as the human end of the resource need things. And then the organization

567
00:53:19,040 --> 00:53:23,920
needs to handle you and all of these others in a certain way and at scale.

568
00:53:24,880 --> 00:53:29,040
I think that's the first thing that I've seen as the most common across all organizations

569
00:53:29,040 --> 00:53:37,440
regardless of industry, regardless of where they are in the world. Yeah, thank you. This was

570
00:53:37,440 --> 00:53:49,440
so interesting to talk. I hope you come back another time. So, my last question is with what

571
00:53:49,440 --> 00:53:58,480
you tell me now, it's when you had unlimited budget resources and time for an AI project, what

572
00:53:58,480 --> 00:54:06,880
will you build? Oh, I have all the money I can do whatever I want. This is not so hard,

573
00:54:06,880 --> 00:54:12,480
necessarily. It's time is the key. It's never mind the resources because time seems to be my

574
00:54:12,480 --> 00:54:17,920
enemy for all these things. Other than building a time machine with my AI would be lovely.

575
00:54:18,640 --> 00:54:22,880
It's cool. It would probably be the first thing I would think of if that could ever happen.

576
00:54:22,880 --> 00:54:30,320
I think in a lot of cases, what I would do with that kind of resource is to essentially make

577
00:54:30,320 --> 00:54:36,720
that actual agent that I can rely on to finally take care of whatever admin so I can just get on with

578
00:54:36,720 --> 00:54:43,920
my game. Awesome. Thank you for your time and this was really, I really enjoyed it. It's a lot of

579
00:54:43,920 --> 00:54:48,560
fun. Thank you so much. You're welcome. I can't wait until you're to show. And maybe a year

580
00:54:48,560 --> 00:54:54,320
later, let's find out if any of my any of our predictions came true. Yeah, we can look at the

581
00:54:54,320 --> 00:54:59,760
in six months. Let me do it. I'll show you. Look at it. You're like so different. And we all have

582
00:54:59,760 --> 00:55:06,720
horns and be like, oh, wow, we evolved. Okay. It's a pleasure, sir. And we'll be seeing you

583
00:55:06,720 --> 00:55:13,360
sometime over the next year. Hopefully I'll get to travel. Thank you. Bye and have a nice day.

584
00:55:13,360 --> 00:55:13,760
Thank you.

585
00:55:13,760 --> 00:55:15,760
Thank you.

Mirko Peters

Founder of m365.fm, m365.show and m365con.net

Mirko Peters is a Microsoft 365 expert, content creator, and founder of m365.fm, a platform dedicated to sharing practical insights on modern workplace technologies. His work focuses on Microsoft 365 governance, security, collaboration, and real-world implementation strategies.

Through his podcast and written content, Mirko provides hands-on guidance for IT professionals, architects, and business leaders navigating the complexities of Microsoft 365. He is known for translating complex topics into clear, actionable advice, often highlighting common mistakes and overlooked risks in real-world environments.

With a strong emphasis on community contribution and knowledge sharing, Mirko is actively building a platform that connects experts, shares experiences, and helps organizations get the most out of their Microsoft 365 investments.

Ralph Rivas

Copilot Junkie

Ralph Rivas is a proud Copilot junkie who’s obsessed with one question: how do you go from a neat AI demo to something your team actually uses on Monday? A Microsoft MVP and senior leader at Sogeti (part of Capgemini), Ralph helps organizations tame Copilot chaos — turning prompts into workflows, workflows into agents, and agents into governed systems that don’t blow up Security’s inbox. He’s the co‑founder of the M365Chicago events community, co‑leader of the Chicago Microsoft Power Platform User Group (CMPPUG), and co‑host of the long‑running CloudTalk Show (running since 2017), where he spends a lot of time breaking, fixing, and rebuilding Copilot on purpose.

Maximizing Microsoft Copilot: Beyond the Demo with Ralph Rivas [MVP]