Last updated: 28 July 2026

I’ve been building out an AI agent, Iron Mind AI, a personal trainer agent, incrementally using the Microsoft Agent Framework.

Each post in this series adds a new capability: function tools, memory, human-in-the-loop approval, MCP support, RAG, and more.

Each part is isolated enough that you can jump in wherever it’s relevant to you, but they’re designed to build on each other. If you’re starting from scratch, work through them in order below.

If you want the fast-track version, the whole series is also packaged as a free video course with source code for every step.

~

The Series, In Order

Here is each part that forms the series.

1. Microsoft Agent Framework: First Look The core concepts, components, and patterns behind shipping AI agents with the Agent Framework.
2. Microsoft Agent Framework: Conversations and Threads Creating conversations with and without threads, serialising/deserialising threads, and working with multiple threads per agent.
3. Microsoft Agent Framework: Extending Agent Intelligence Using Function Tools Giving your agent access to data and capabilities beyond its training data using function tools.
4. Microsoft Agent Framework: Using Agents as Function Tools Building discrete agents and exposing them as function tools to other agents, using .AsAIFunction().
5. Microsoft Agent Framework: Implementing Human-in-the-Loop AI Agents Enforcing approval checkpoints and guardrails, essential for regulated industries and anything that modifies state.
6. Microsoft Agent Framework: Giving Agents Contextual Memory Using AIContextProvider Persisting context across conversational threads so your agent stops being stateless.
7. Microsoft Agent Framework: Using Background Responses to Create an AI Researcher and Newsletter Publisher Handling long-running agent tasks without blocking the UI, using continuation tokens.
8. Model Context Protocol (MCP): Building and Debugging Your First MCP Server in .NET The foundation for exposing agent capabilities to any MCP-compatible client.
9. Microsoft Agent Framework: Exposing an Existing AI Agent as an MCP Tool Taking an agent you’ve already built and wrapping it as an MCP tool over HTTP.
10. Microsoft Agent Framework: Implementing an AI Email Marketing Agent Extending the agent with email marketing capabilities via a third-party provider.
11. Microsoft Agent Framework: Adding RAG to Your AI Agent Using TextSearchProvider and In-Memory Vector Store Grounding agent responses in your own documents instead of relying solely on training data. (Also part of the RAG in .NET series if you’re focused on RAG specifically.)
12. New Free Course: Understanding Microsoft Agent Framework The whole series above, packaged as a free video course with full source code.

Dig in!

~

Building Production RAG?

If you came here specifically for the RAG post and want to go deeper on production RAG patterns, chunking, retrieval quality, observability, and the tooling I’ve built to manage it, see the RAG in .NET series.

~

Questions about any post in this series, or want to see a topic covered? Drop a note in the comments, or schedule a call to discuss consulting and development services.

M.G. Siegler from Spyglass is back for our Montly discussion of the latest tech news. We cover: 1) Demis Hassabis steps down as DeepMind CEO (Alex solo) 2) Apple's memory crunch 3) Should Apple have known better? 4) Will iPhone prices go up? 5) How much can Apple raise prices without losing sales 6) Does that eventually hurt its services business? 7) Microsoft is spending less on AI... but there's an interesting wrinkle 8) How much cloud growth is driven by OpenAI and Anthropic? 9) The divisions within Google's AI division 10) Can Google get it together?

---

Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice.

Want a discount for Big Technology on Substack + Discord? Here’s 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b

Learn more about your ad choices. Visit megaphone.fm/adchoices

Summary

In this episode, Andy welcomes Cheryl Strauss Einhorn, founder and CEO of the decision sciences company Decisive and author of The Human Edge: Smarter Decisions in the Age of AI. Cheryl has spent decades helping people make better decisions, and her message here is a simple one: AI can gather, summarize, and compare, but it can't know what matters most unless we do the thinking first.

Andy and Cheryl talk about why "AI first" can quietly turn into AI only, and why a poor answer from AI is actually useful feedback about your own thinking. Cheryl explains why separating the what of a decision from the why can change the path forward, shares her vision of success question, and walks through how human bias and AI bias reinforce each other. You'll also hear a practical way to use AI personas to pressure test your thinking before a big stakeholder conversation, plus what parents can do to help kids strengthen their decision-making muscles instead of letting them atrophy.

If you're looking for practical ways to make smarter decisions in the age of AI, this episode is for you!

Sound Bites

• "AI is about patterns, but we're about purpose."
• "We need to be the chief deciders in our own life."
• "We need to make sure that we actually have a way to push back and check the veracity of what it's giving us because it's a known liar."
• "And so the authenticity of human interaction is what actually builds trust, strengthens relationships, and really gives us a lot of the connections that are important to us in our home life and our work life."
• "You don't need to have a perfect prompt for AI."
• "AI is going to give you other people's answers."
• "'Knowledge is power' isn't as true now. It's what you are able to do with the knowledge, and it's why you want to collect the knowledge in the first place."
• "I almost never would accept AI's first answer."
• "We tend to have evolving hypotheses unless we really build an audit trail of our thinking."
• "So in the world of medicine, a fever tells you something is wrong, but it tells you nothing about what is wrong or where to look."
• "So first we know AI's a sycophant, right?"
• "If it's 99% accurate in its conclusion, but its underlying data is the wrong data set, you've got the wrong answer."
• "I would say the more that I've studied these cognitive biases, these mental shortcuts, the more I realize that we see the world through a dirty windshield."
• "Our brains are muscles, and just like we exercise to strengthen our muscles, if we don't exercise, they atrophy."

Chapters

• 00:00 Introduction
• 02:00 Start of Interview
• 02:12 Growing Up Surrounded by Questions
• 04:30 What Leaders Should Watch Out for in "AI First"
• 07:07 When AI Confidently Makes Things Up
• 09:04 How AI First Turns into AI Only
• 09:20 Why Polished Doesn't Mean Authentic
• 11:59 A Prompting Problem or a Problem Definition Problem?
• 14:37 The Vision of Success Question
• 15:45 Why a Bad Answer Is Useful Feedback
• 19:06 Separating the What from the Why
• 20:32 Research: More Information Isn't Better Judgment
• 24:35 Progressive Prompting and Never Accepting the First Answer
• 25:18 Fabricated Quotes and the Trouble with Attribution
• 26:50 Documenting Assumptions and Building an Audit Trail
• 29:25 Committing to Your Own Thinking
• 32:29 How Human Bias and AI Bias Reinforce Each Other
• 34:55 Knowing About Biases Doesn't Make You Immune
• 36:51 Using AI Personas to Challenge Your Thinking
• 40:20 AI and Stakeholder Management
• 42:42 Helping Kids Become Better Decision-Makers
• 44:25 End of Interview
• 44:51 Andy Comments After the Interview
• 48:13 Outtakes

Learn More

You can learn more about Cheryl and her work at AREAMethod.com.

For more learning on this topic, check out:

• Episode 460 with Joe Sutherland. It's an interesting look at the intersection of AI, data, and decision-making, and a great follow-up to this discussion.
• Episode 381 with Jim Loehr. One of Andy's favorite conversations about decision-making, from a remarkable figure in the human performance world.
• Episode 99 with Mike Roberto. A conversation about his book Why Great Leaders Don't Take Yes for an Answer, recorded long before ChatGPT showed up, and still insightful today.

Chat with PMeLa

You can chat directly with PMeLa, the podcast's AI persona, to get episode recommendations and answers to your project management and leadership questions. Visit PeopleAndProjectsPodcast.com/PMeLa to chat with her.

Pass the PMP Exam

If you or someone you know is thinking about getting PMP certified, we've put together a helpful guide called The 5 Best Resources to Help You Pass the PMP Exam on Your First Try. We've helped thousands of people earn their certification, and we'd love to help you too. It's totally free, and it's a great way to get a head start.

Just go to 5BestResources.PeopleAndProjectsPodcast.com to grab your copy. I'd love to help you get your PMP this year!

Join Us for LEAD52

I know you want to be a more confident leader–that's why you listen to this podcast. LEAD52 is a global community of people like you who are committed to transforming their ability to lead and deliver. It's 52 weeks of leadership learning, delivered right to your inbox, taking less than 5 minutes a week. And it's all for free. Learn more and sign up at GetLEAD52.com. Thanks!

Thank you for joining me for this episode of The People and Projects Podcast!

Talent Triangle: Power Skills

Topics: Decision Making, Artificial Intelligence, Leadership, Project Management, Critical Thinking, Cognitive Bias, Problem Solving, Stakeholder Management, Judgment, Curiosity

The following music was used for this episode:

Music: Echo by Alexander Nakarada
License (CC BY 4.0): https://filmmusic.io/standard-license

Music: Tuesday by Sascha Ende
License (CC BY 4.0): https://filmmusic.io/standard-license

Robin Heinze and Tyler Williams break down React Native’s AppRegistry! From Expo and app entry points to brownfield apps and more, see what’s happening under the hood and build mobile apps with even more confidence after this exciting episode.

Connect With Us!

• Robin Heinze: @robinheinze
• Tyler Williams: @coolsoftware.dev (Bsky)
• React Native Radio: @ReactNativeRdio

This episode is brought to you by Infinite Red!

Infinite Red is a premier mobile app consultancy, especially focused on Expo and React Native, located fully remote in the US. We’re a team of 30 with highly experienced mobile app developers and have been doing this for over a decade. We are also one of the first development teams to adopt agentic coding in a way that keeps high quality standards and aren’t afraid to do things the old school way if we need to. If you’re looking for mobile app or React Native or Expo expertise for your next project, hit us up at infinite.red/radio.

Share Episode         
         
Pushkar Gopalakrishna, Senior Staff Software Engineer at Snap, previously Cruise and AWS, joins to explore why engineering organizations pivot away from Terraform and HCL toward Kubernetes-native tools even when they might not be better. We unpack how developer friction, copy-pasted control structures, and misaligned organizational incentives create massive tech debt—often forcing SREs to manually update infrastructure repositories for compliance, resulting in broken pipelines and severe operational friction.

We debate over the mechanics of Crossplane, detailing how its continuous reconciliation loop and Custom Resource Definitions (CRDs) allow teams to express cloud infrastructure as YAML alongside their application manifests at potentially the cost of async validation. Pushkar pulls back the curtain on how Cruise managed infrastructure at scale using a custom internal platform called 'Juno' to bootstrap GCP projects, repositories, and permissions, while leveraging Crossplane for application-level resources. We also dive into the dangers of using CI tools for continuous deployment, detailing a terrifying incident where a pipeline bug accidentally marked three production Kubernetes namespaces for deletion, and how moving to ArgoCD and Argo Rollouts helped prevent future outages for autonomous vehicles.

Finally, we touch on the realities of non-production environment isolation, testing against live APIs, and why platform teams must balance providing a seamless developer experience without stripping away developer accountability.

• Crossplane
• Podcast Guest Request for Principal Engineer — What work are you doing?
• Amazon Multi-level fullyment center for drones
• ✨ Episode: Terraform vs OpenTofu

• Warren - Books: The Murderbot Diaries
• Pushkar - DJI mini drone