Episodes

Most organizations believe hallucinations are the biggest risk in enterprise AI. In reality, one of the most dangerous threats is something far less visible: Indirect Prompt Injection. In this episode, we explore how trusted documents, emails, SharePoint content, Teams conversations, and knowledge bases can become attack vectors that manipulate AI systems without ever compromising the underlying infrastructure. The episode examines why Retrieval-Augmented Generation (RAG), the foundation behind Microsoft 365 Copilot, Azure AI Foundry solutions, and many enterprise AI assistants, introduces an entirely new security challenge. Unlike traditional software, large language models cannot reliably separate data from instructions. Every piece of retrieved content becomes part of the model's context, allowing hidden commands, poisoned documents, metadata, and embedded instructions to influence AI behavior. Listeners will learn how indirect prompt injection works, why system prompts are n…

Most organizations believe they are building AI agents. In reality, many are still deploying advanced chatbots that operate within isolated applications. These systems can answer questions and generate content, but they lack persistence, long-term memory, and the ability to coordinate work across enterprise systems. In this episode, we explore one of the most important shifts happening in enterprise AI: the evolution from standalone conversational experiences to connected agent ecosystems. Instead of treating AI as a feature inside Microsoft Teams, Slack, or individual business applications, organizations are beginning to build persistent agent fabrics that can collaborate, share context, and execute tasks across multiple platforms. We discuss why memory, orchestration, identity, and governance are becoming critical components of modern AI architectures. The conversation examines how enterprises are moving beyond simple prompt-response interactions toward systems where AI agents…

In this episode, we explore why many organizations are making a critical mistake when building AI solutions: creating agents that operate in isolation. While individual agents can be powerful, they often become disconnected silos that lack the ability to coordinate across systems, processes, and business functions. The conversation focuses on Azure Logic Apps as the “nervous system” for enterprise AI, providing the orchestration layer that connects agents, applications, and workflows. Rather than viewing agents as standalone tools, organizations should design them as part of a larger ecosystem where events, messages, and automated processes enable collaboration and intelligent decision-making. We discuss the principles of event-driven architecture, the role of integration in modern AI systems, and how Logic Apps can connect Microsoft 365, Azure services, business applications, and external platforms. The episode also covers governance, scalability, and operational visibility, sh…

Multi-Agent Systems, Copilot Studio, Microsoft Agent Framework, Intelligent Automation, AI Orchestration, Microsoft Foundry, Artificial Intelligence, David Lorenzo Lopez, MC65 Podcast, Generative AI, Autonomous Agents, Enterprise AI, Microsoft Azure, Microservices Architecture, LLM Agents, AI for Business, GPT-3.5, Software Development, Microsoft Ecosystem

The rise of enterprise AI has created a fundamental challenge: how can organizations leverage powerful language models without exposing their most valuable proprietary data? In this episode of M365.fm, we explore the growing adoption of Private LoRA (Low-Rank Adaptation) as a practical architecture for secure AI systems built on sensitive enterprise information. Rather than sending confidential documents, intellectual property, customer records, or internal knowledge to public AI services, Private LoRA enables organizations to adapt and customize foundation models while keeping their data within controlled environments. The discussion explains why traditional fine-tuning approaches are often expensive, difficult to govern, and introduce significant security and compliance concerns. The episode breaks down how LoRA works by modifying only a small subset of model parameters, allowing organizations to create specialized AI capabilities without retraining entire large language model…

Manual tagging is dead—and it’s quietly undermining your Microsoft 365 governance strategy. In this episode, we explore why traditional metadata management based on dropdown menus, user-selected labels, and manual classification no longer works in modern organizations. The volume of content generated across SharePoint, Teams, OneDrive, Copilot, and Microsoft 365 has grown beyond what humans can reliably classify. The problem isn’t that users are unwilling to tag content—it’s that manual tagging is inconsistent, incomplete, and impossible to scale. When metadata quality declines, governance suffers. Search results become unreliable, retention policies lose effectiveness, compliance controls weaken, and AI tools like Microsoft Copilot struggle to understand and protect organizational data. The episode examines how Microsoft Purview and AI-powered classification are changing the game. Instead of relying on users to choose the correct label, modern governance systems can analyze …

As quantum computing moves from theory toward reality, many organizations are focusing on replacing RSA and ECC with post-quantum cryptography. But in this episode of M365.fm, Mirko Peters argues that simply choosing a new algorithm is not enough. The real challenge is cryptographic agility: the ability to rapidly adapt, replace, and evolve cryptographic systems as threats, standards, and technologies change. The discussion explores why most enterprise environments are deeply dependent on cryptography in ways many organizations don't fully understand. Certificates, identity systems, VPNs, TLS connections, APIs, cloud workloads, IoT devices, and long-lived data all rely on cryptographic foundations that may become vulnerable in a post-quantum world. The biggest risk is not that quantum computers arrive tomorrow—it is that organizations cannot adapt quickly when change becomes necessary. The episode examines how crypto-agility shifts the conversation from algorithm selection to ar…

As organizations rapidly adopt Microsoft 365 Copilot, AI agents, and generative AI technologies, one challenge stands above all others: ensuring data is secure, governed, and compliant. In this episode of M365 FM, Mirko Peters speaks with Peter Rising, Senior Partner Solution Architect at Microsoft, about how Microsoft Purview helps organizations prepare for AI at scale. The discussion explores why AI readiness is not just about deploying Copilot licenses but understanding and controlling the data that powers AI experiences. Peter explains how Microsoft Purview provides visibility into sensitive information, helps classify and protect business-critical data, and enables organizations to apply Zero Trust principles across their Microsoft 365 environment. The conversation covers key capabilities including Data Loss Prevention (DLP), sensitivity labels, information protection, insider risk management, auditing, compliance monitoring, and data governance. These tools help ensure tha…

Cloud strategies often focus on scalability, cost optimization, and centralized services, but many organizations overlook one critical factor: latency. In this episode of M365 FM, Mirko Peters explores why network latency can become a major barrier to application performance and user experience, especially as businesses increasingly rely on cloud-based services. The discussion examines the concept of the “latency wall” and explains why simply moving workloads to the cloud does not automatically guarantee better performance. For applications that require real-time processing, industrial automation, IoT, AI inference, or low-latency user interactions, the physical distance between users, devices, and cloud datacenters can create significant challenges. The episode highlights how edge computing is emerging as a solution by bringing compute and data processing closer to where data is generated. Rather than sending every request to a centralized cloud region, organizations can levera…

What does the future of Azure look like, and how are Infrastructure as Code and DevOps transforming the way organizations build and manage cloud solutions? In this episode of M365 FM, host Mirko Peters is joined by Microsoft Azure MVP Maik van der Gaag for an in-depth discussion about modern cloud engineering, automation, and the evolving Azure ecosystem. Maik shares insights from his extensive experience helping organizations adopt cloud technologies and modern development practices. The conversation explores why Infrastructure as Code (IaC) has become a critical foundation for scalable and reliable cloud environments. Maik explains how tools such as Terraform and Bicep enable teams to automate deployments, improve consistency, reduce configuration drift, and accelerate delivery across Azure environments. Beyond the technology, the episode highlights the cultural side of DevOps. Successful cloud transformation is not only about tools and automation but also about collaborati…

Learn how to build low-cost AI agents in Microsoft Cloud with Azure OpenAI, Copilot Studio, Power Platform, and practical cost controls.

In this episode of the M365.fm podcast, Nathan Rose joins the show to explore how Copilot Studio, Dataverse, and the Model Context Protocol (MCP) are shaping the future of agentic AI in Microsoft 365. The conversation dives into how organizations can move beyond simple chatbots and build intelligent agents that understand intent, access business data, and take meaningful actions across systems. Nathan explains why Dataverse has become a critical foundation for AI-powered business applications, providing structured data, security, and governance that enterprise AI solutions require. The discussion then focuses on MCP, an emerging open standard that enables AI agents to connect with external tools, data sources, and business systems in a more consistent and scalable way. Listeners will learn how MCP reduces integration complexity, allowing Copilot Studio agents to interact with Dataverse, Dynamics 365, Microsoft services, and even third-party systems without relying on large numbe…

Small Language Models (SLMs) are emerging as one of the most important developments in enterprise AI. While Large Language Models (LLMs) power tools like Microsoft Copilot with impressive reasoning and language capabilities, they also introduce challenges around cost, latency, hallucinations, and scalability. This episode explores why bigger models are not always better and how SLMs can solve many of the problems organizations face when deploying AI at scale. The discussion explains that many enterprise AI tasks are highly specialized and do not require the full power of a massive LLM. Instead, purpose-built SLMs can be trained or optimized for specific business scenarios, delivering faster responses, lower infrastructure costs, and more predictable outcomes. By narrowing the scope of what a model needs to know, organizations can significantly reduce hallucinations while improving reliability. The episode also examines how future Copilot architectures are likely to evolve into m…

In this episode of the M365 Voice podcast, host Mirko Peters is joined by Parvez Ghumra to explore the critical role of Application Lifecycle Management (ALM) in the Power Platform. The conversation moves beyond simple solution deployments and dives into what it takes to build enterprise-ready Power Platform solutions that are scalable, maintainable, and governed effectively. Parvez shares his journey from traditional software development into the low-code world and explains why citizen development and professional development must work together. Listeners will learn how source control, automated deployments, Azure DevOps, GitHub, and governance practices can transform Power Platform projects from manual, error-prone processes into reliable enterprise delivery pipelines. The discussion also covers common ALM challenges, how organizations can avoid deployment pitfalls, and practical strategies for introducing DevOps principles into Power Platform environments. Whether you're a ma…

The explosion of AI-powered applications has created a new challenge: how do you efficiently search through billions of vector embeddings without exploding infrastructure costs? In this episode, we explore the “billion-vector problem” and compare two leading vector search algorithms available in Azure AI Search: HNSW (Hierarchical Navigable Small World) and DiskANN. While HNSW has become the industry standard thanks to its fast in-memory performance and high recall, it requires significant RAM as datasets grow. DiskANN, originally developed by Microsoft Research, takes a different approach by leveraging SSD storage to dramatically reduce memory requirements while maintaining excellent search accuracy at massive scale. We break down how each algorithm works, where they shine, and the trade-offs architects need to consider when designing Retrieval-Augmented Generation (RAG), semantic search, and AI agent solutions. The discussion covers performance, scalability, operational costs, up…

Artificial Intelligence is everywhere. Every conference keynote, every boardroom discussion, and every technology roadmap seems to be focused on AI. But beyond the excitement and endless headlines, one question remains: how do organizations move from AI experimentation to real, measurable business value? In this episode of the M365.fm Podcast, I sit down with Microsoft MVP, Solution Architect, Microsoft Certified Trainer, and Power Platform expert Kayode Ajayi to explore what successful AI adoption actually looks like inside modern organizations. Together, we cut through the hype and focus on the practical realities of implementing Microsoft Copilot, Copilot Studio, Power Platform, and enterprise AI solutions at scale. Kayode shares his journey from technology enthusiast to Microsoft MVP and explains how Power Platform has evolved into a true enterprise-grade platform capable of supporting complex business scenarios when backed by the right architecture, governance, and security…

In this episode of the M365 FM Podcast, we explore one of the biggest hidden risks in modern data governance: shadow data. While Microsoft Purview provides powerful visibility into governed data sources, many organizations assume that what Purview cannot see does not exist. That assumption creates a dangerous blind spot. The discussion explains how shadow data emerges across disconnected systems, unmanaged repositories, legacy platforms, third-party applications, personal storage locations, and forgotten workloads that sit outside normal governance processes. These hidden data stores often contain sensitive business information, intellectual property, customer records, and compliance-relevant content that never appears in standard Purview reporting. The episode breaks down why organizations frequently mistake data discovery for complete data visibility. Even with strong classification, labeling, and compliance controls in Microsoft 365, governance can only protect what it can ac…

In this episode, Mirko Peters explores one of the most ambitious experiments in enterprise AI: engineering Microsoft Copilot to analyze and reason across more than 35 million pages of highly complex and interconnected information, inspired by the scale and investigative challenges surrounding the Epstein files. Rather than focusing on AI as a simple chatbot, the episode examines what happens when Copilot is pushed to operate at investigative scale. Traditional search systems struggle when information is fragmented across millions of documents, relationships, entities, timelines, and hidden connections. The real challenge is no longer finding information but creating context from overwhelming volumes of data. Mirko explains how modern AI architectures combine Microsoft Copilot, Microsoft Graph, semantic search, vector databases, knowledge graphs, and Retrieval-Augmented Generation (RAG) to transform massive document collections into navigable intelligence systems. The discussion …

In this thought-provoking episode of M365.fm, Mirko Peters explores a surprising but highly practical concept: how to intentionally “Trumpify” Microsoft Copilot to better understand and control AI hallucinations. Rather than treating hallucinations as random mistakes, the episode examines them as a predictable outcome of how large language models generate responses when context, grounding, and validation mechanisms are missing. The discussion breaks down why Copilot sometimes produces confident but incorrect answers, how language models fill knowledge gaps with plausible-sounding information, and why human users often trust these outputs more than they should. By deliberately pushing Copilot toward exaggerated confidence and unsupported claims, listeners gain a deeper understanding of the architectural weaknesses that lead to hallucinations in enterprise AI systems. A key theme is that hallucinations are not simply an AI problem—they are a data, governance, and grounding problem…

In this episode of M365.fm, Mirko Peters explores how organizations can build a secure and private Retrieval-Augmented Generation (RAG) platform using SharePoint as the knowledge source and n8n as the orchestration layer. The discussion focuses on moving beyond generic AI chatbots and creating enterprise-grade AI systems that can access, retrieve, and reason over internal business knowledge while maintaining governance and security. The episode explains the core architecture of a private RAG solution, including document ingestion, chunking strategies, vector embeddings, semantic search, and AI response generation. Listeners learn why SharePoint is an ideal enterprise knowledge repository and how n8n can automate the entire pipeline without requiring complex custom development. Mirko breaks down the challenges many organizations face when deploying AI, including data silos, permission management, outdated content, and the risk of exposing sensitive information to public AI servic…

This episode explores how organizations can connect Microsoft Copilot with the predictive capabilities of Power BI to move beyond simple reporting and toward proactive, data-driven decision-making. The discussion highlights that while Copilot excels at natural language interactions and summarizing information, its real business value increases when it can access trusted analytical models, forecasts, and governed business data from Power BI. The episode explains that many organizations still use Power BI primarily as a dashboard destination. However, the future lies in treating Power BI as a semantic and analytical layer that feeds AI-powered experiences. Instead of navigating reports manually, users can ask questions in natural language through Copilot and receive contextual answers backed by governed Power BI models. A key theme is predictive analytics. By combining Copilot with Power BI datasets, organizations can surface forecasts, trends, risk indicators, and business predic…

In this episode of M365.fm, host David Warner speaks with MVP Gautam Sheth about Microsoft 365 Copilot extensibility and how organizations can go beyond the out-of-the-box Copilot experience. Gautam explains the different extensibility options available, including declarative agents, custom engine agents, Copilot connectors, and plugins, helping listeners understand when and why each approach should be used. The discussion focuses on practical steps for extending Copilot with organizational knowledge, business processes, and external systems to deliver more relevant and actionable AI experiences. Gautam highlights the importance of grounding Copilot in enterprise data, leveraging Microsoft Graph, and designing solutions that align with real business scenarios rather than simply adding AI for its own sake. The episode also covers governance, security, and adoption considerations, emphasizing that successful Copilot implementations require a balance between technical capabilities …

In this episode, Mirko Peters explores why successful Microsoft 365 strategy should be approached like building a synthetic market rather than deploying technology in isolation. The core idea is that Microsoft 365 creates an internal economy where information, collaboration, automation, governance, and AI capabilities continuously interact. Organizations that focus only on individual tools such as Teams, SharePoint, Power Platform, or Copilot often miss the larger system dynamics that drive long-term value. The discussion highlights that every platform decision creates incentives and behaviors. Poor governance can encourage content sprawl, uncontrolled workspace growth, and fragmented knowledge, while well-designed governance creates trust, discoverability, and sustainable adoption. The episode argues that strategy is not about maximizing feature usage but about shaping the conditions that allow productive behaviors to emerge naturally across the organization. A key theme is tha…

Microsoft Copilot doesn’t become a true AI assistant by adding more prompts—it becomes one when it understands your context, remembers how you work, and can act across your tools. In this article, Mirko Peters explains how he transformed Microsoft Copilot into a JARVIS-like assistant by combining Microsoft 365 Copilot, Copilot Studio, custom agents, memory, and automation workflows. The core idea is to move beyond using Copilot as a smarter search engine and instead build an AI operating layer that understands projects, priorities, meetings, documents, and business processes. By connecting Copilot to Microsoft Graph, business data, and specialized agents, the assistant can provide personalized responses, automate repetitive work, surface relevant knowledge proactively, and coordinate actions across different systems. The article outlines the architecture behind this approach, including the use of custom instructions, agent orchestration, context grounding, and workflow automatio…