This talk was recorded at NDC Manchester in Manchester, England. 
#ndcmanchester #ndcconferences #developer #softwaredeveloper    

Attend the next NDC conference near you: 
https://ndcconferences.com
https://ndcmanchester.com/

Subscribe to our YouTube channel and learn every day:   
/            @NDC 

Follow our Social Media!

https://www.facebook.com/ndcconferences
https://twitter.com/NDC_Conferences
https://www.instagram.com/ndc_conferences/

#ai #ml #machinelearning #architecture 

It's impossible to be a developer in 2025 and not feel the AI buzz. But now we're moving beyond chat clients, this is the year of Agentic AI. Agents are small, goal-driven AI systems that can act on the user's behalf.

In this session, we'll dive deep into the intricacies of pro-code Agents based in the .NET ecosystem and Azure. With Microsoft Agent Framework as the AI SDK of choice, using models hosted in Azure Foundry.

But the code is just our starting point!

We'll explore some tough questions, like:
What is the best way to let Agents solve issues in collaboration?
What happens to an Agent if you let it evolve, and how do you actually solve evolution?
What is the best way to handle memory for Agents?
Can Agents do a good job reasoning over complex topics, such as economic facts?
How do we handle web requests most economically? As those cost more tokens than anything else we can do with an AI system!
And of course, we have to discuss the best current models for building Agentic systems!

These are not theoretical puzzles. They are real-world challenges when building agentic systems that interact with the world, whether through .NET logic or calling external MCP services. Expect surprising outcomes, hard-earned lessons, and a few dragons along the way.

Join this session for tons of hard learned insights and a cool demo to inspire you!

NDC Conferences

Building autonomous AI agents requires understanding memory management, tool calling, model selection, and workflow orchestration. Key challenges include context window limitations, tool confusion, long-term memory persistence, and model-specific behaviors. Microsoft Agent Framework provides abstractions for agent creation, middleware, and multi-agent workflows. Different models excel at different tasks—Anthropic models handle tool calling well, while OpenAI and Google models vary in logic capabilities. Critical practices include implementing evaluations before production, using classical engineering alongside AI, managing token costs, and never changing models without testing. Memory strategies like scratchpads, summarization, and temporal weighting help manage context. Multi-agent systems with smaller specialized models often outperform single large models.

Advanced Autonomous AI Agents: Memory, Collaboration and Evolution in Action - Andreas Wänqvist