Thanks for attending our session at aMP Day Rennes 2026!
Like I said in my post about AI Maitri Global Summit 2026, I wanted to explore more opportunities to present at different events. I've never been a high-flying speaker, with conferences, workshops or customer visits in multiple countries each week - but I guess this year, I've come closer than ever 😅
Less than a month after the ever-so-fun HELish Summit, I got to present at the global AI Maitri event. And only a few days later I flew to France for aMP Day Rennes 2026. Add the AI-Native Workplace Summit of 2026 to the mix, and in 5-6 weeks I've presented at 4 different events.
And while it takes a lot of energy and planning to make sure I bring interesting takeaways (and only share the more well-considered opinions of mine...) to the audience each time, I've also found it to always be incredibly rewarding.
Reflections
Growing up in the countryside and starting my career in a small town in Finland, I never had many chances to take part in community events or conferences.
Apart from a few larger conferences early in my career, the SharePoint Saturdays I attended while living in the United States and then Montréal were my first real exposure to the tech community. They also showed me how valuable it is to hear other people's stories, both technical and personal.
... well, since I AM a tech guy, I mostly attend sessions where I can hear how other people are getting the most out of some tech that is at least somewhat relevant to me.
And while I eventually figured I'd like to take the leap and applied to speak at some of these events, I've been lucky enough to have employers who support it (by picking up the bill) - first Valo, then Omnia, and now Meltlake. I wrote more about how I got started in my story of becoming an MVP.
Session details
Just to rehash, here's what we presented at the event:
M365 Copilot is the UI, Foundry is the Brain, but what glues it together?
Everybody is building agents.
Most of them are glorified prompt wrappers.
Meanwhile, the interesting engineering is happening somewhere else!
Microsoft 365 Copilot is rapidly becoming the interface layer for AI experiences across Microsoft 365. But the real innovation is increasingly moving into Microsoft Foundry, where developers can choose models, evaluate models, orchestrate agents, combine tools, and build systems that extend far beyond traditional chatbot experiences. Microsoft 365 Agents SDK then connects those systems back into Microsoft 365, Teams and Copilot.
In this session, we'll build a real agent from the ground up and use it as an excuse to explore the most interesting parts of the modern Microsoft AI stack. Along the way we'll experiment with model selection, evaluations, orchestration patterns, agent memory, Work IQ, custom engine agents and agent-to-agent collaboration.
This is not a session about prompt engineering: This is a session about software engineering.
Expect fewer marketing diagrams, more code, more demos, more architectural trade-offs, and at least a few deliberately questionable decisions made in the name of learning.
Because in 2027, Copilot may be the interface. But Foundry is increasingly where the brains live.
Takeaways:
✅ Learn where Foundry creates real value
Understand when model choice, evaluations, orchestration and memory matter, and when they are just unnecessary complexity.
✅ Know when Copilot Studio is enough
Recognize the point where low-code tooling stops being productive and code-first approaches start paying dividends.
✅ Avoid expensive architectural mistakes
Learn which patterns scale, which patterns are maintainable, and which patterns only look impressive in conference demos.
✅ Leave with ideas you'll actually want to try
Walk away with practical techniques involving Foundry, custom engine agents, Work IQ and Microsoft 365 extensibility that you can apply immediately.
References
You can find the slides right here (as an attachment, since SlideShare uploads have casually been down for over two months since first being reported in August)!
See you around - or online! 😎
Comments
No comments yet.