BASTA! 2025: Agents in Action: LLM's, Tools and Reasoning

Slides for my talk about how to implement agents at BASTA! 2025 in Mainz.

Transcript

1. Agents in Action LLMs, Tools & Reasoning Agents in Action: LLMs, Tools & Reasoning Sebastian Gingter Developer Consultant sebastian.gingter@thinktecture.com
2. ▪ What to EXPECT ▪ Overview of building blocks of an Agentic system powered by Large Language Models (LLMs) ▪ Pragmatic use cases ▪ Demos (mostly language independent) ▪ What not NOT TO EXPECT ▪ ML & AI fundamentals ▪ ChatGPT, CoPilot(s) ▪ Deep dives into SDKs, Mastra, Semantic Kernel Agent Framework etc. Agents in Action LLMs, Tools & Reasoning Agents in Action: LLMs, Tools & Reasoning
3. Agents in Action: LLMs, Tools & Reasoning Sebastian Gingter Developer Consultant sebastian.gingter@thinktecture.com
4. ▪ Generative AI in business settings ▪ AI driven Developer Productivity & Software Quality ▪ All things .NET ▪ Microsoft MVP for .NET & Azure AI Services ▪ sebastian.gingter@thinktecture.com ▪ https://www.thinktecture.com Agents in Action LLMs, Tools & Reasoning Sebastian Gingter Developer Consultant @ Thinktecture AG
5. Agents in Action LLMs, Tools & Reasoning What is an “Agent”?
6. Agents in Action LLMs, Tools & Reasoning Agent, noun, countable, /ˈeɪ.dʒənt/ ▪ someone who works secretly for the government or another organization ▪ a person who acts for or represents another ▪ a person or thing that produces a particular effect or change
7. Agents in Action LLMs, Tools & Reasoning Agent ▪ LLM with the licence to act
8. Agents in Action LLMs, Tools & Reasoning From LLM to Agent
9. ▪ LLMs are just a web request for us Agents in Action LLMs, Tools & Reasoning Large Language Models (LLMs)
10. From LLM to Agent Agents in Action LLMs, Tools & Reasoning DEMO
11. Agents in Action LLMs, Tools & Reasoning Recap
12. Agents in Action LLMs, Tools & Reasoning LLMs are stateless 🫥
13. Agents in Action LLMs, Tools & Reasoning LLMs are stateless ▪ We must add memory
14. Agents in Action LLMs, Tools & Reasoning Chat is just a loop now
15. Agents in Action LLMs, Tools & Reasoning Chat is just a loop now ▪ Message types ▪ System ▪ User ▪ Assistant ▪ We have full control over the memory! ▪ We can inject and alter history if we want/need to
16. Agents in Action LLMs, Tools & Reasoning Tools allow interaction with the real world
17. Agents in Action LLMs, Tools & Reasoning Tools allow interaction with the real world ▪ Assistant message can carry tool calls ▪ New message type: tool ▪ Tool choice can be preset or automatic ▪ Provide only few tools. OpenAI docs says: “Aim for fewer than 20 functions at any one time”
18. Agents in Action LLMs, Tools & Reasoning Interlude: Model Context Protocol (MCP) ▪ No magic, just a standardized way of ▪ providing tools to a model ▪ interpreting and execution of tool calls ▪ Specification is still work in progress and missing a lot ▪ Stateful protocol at heart ▪ Authentication is an afterthought https://julsimon.medium.com/why-mcps-disregard-for-40-years-of-rpc-best-practices-will-burn-enterprises-8ef85ce5bc9b
19. Agents in Action LLMs, Tools & Reasoning Tools in a loop allow agency
20. Ein KI-Agent ist ein Programm, das in einer Schleife zwischen dem Reasoning mit einem LLM und dem Ausführen von Tools wechselt, bis es sein Ziel erreicht. Agents in Action LLMs, Tools & Reasoning KI-Agenten - Einzeiler
21. An AI-Agent is a program, which loops between reasoning with a LLM and the calling of tools, until it reaches its goal. Agents in Action LLMs, Tools & Reasoning AI-Agent – One-liner
22. Agents in Action LLMs, Tools & Reasoning That’s it!
23. Agents in Action LLMs, Tools & Reasoning Or is it?
24. Agents in Action LLMs, Tools & Reasoning What do we have now? ▪ There are a LOT of issues with that simple approach ▪ Memory grows over time (performance, cost, context window size overrun) ▪ Unstructured trial and error (depending on model) ▪ Functions limit what the agent can do ▪ User has very little control over what happens ▪ Flow is easily interrupted by external errors ▪ etc…
25. Agents in Action LLMs, Tools & Reasoning Memory / Prompt / Context Management ▪ Memory growth ▪ Context window limits ▪ Prompt Caching ▪ Selective memory ▪ Compacting (Summarization in intervals)
26. Agents in Action LLMs, Tools & Reasoning Unstructured trial & error ▪ ReAct – Pattern (Reasoning & Acting) ▪ Reasoning models do help ▪ Prompt for a plan first, then continue ▪ Ask to create a todo-list and follow that
27. Agents in Action LLMs, Tools & Reasoning Function limits ▪ Filter functions by context ▪ Embeddings can help ▪ Sub-agents ▪ Hand a sub-task over to agents with specific tools
28. Agents in Action LLMs, Tools & Reasoning Limited control ▪ Approval gates ▪ Agent should pause before certain tasks ▪ Certain tools require user-approval (outside of agent loops) ▪ User can approve/deny/request changes ▪ Real-time monitoring ▪ User sees work and can interrupt with corrections
29. Agents in Action LLMs, Tools & Reasoning Resilience ▪ A working agent is a long-running process between different systems ▪ treat it like that ▪ Agent state is serialized / persisted between roundtrips ▪ Partial results should be stored
30. Agents in Action LLMs, Tools & Reasoning Pragmatic ideas
31. Agents in Action LLMs, Tools & Reasoning Agents Architecture Overview 31 Language Model Decides to reason, act, or respond Input User or environment Memory Preferences or prior facts Storage Session history and state Instructions Behavior policy, tool usage rules Reasoning Plan, reflect, analyze results Output Response / effect Tools APIs and actions, MCP Knowledge Domain data, RAG / Search Query / event Recall context Load instructions Save session Make decision Use tools Search knowledge Process results Generate output Central decision maker Input/output flow Core system components External interactions 1 2 3 4 7 8 6 5 5 ↻
32. Agents in Action LLMs, Tools & Reasoning Different execution modes ▪ Autonomous – Agent works uninterrupted ▪ Step-by-step – Pause before every tool call ▪ Supervised - Agents actively asks when unsure ▪ You can switch between modes if necessary
33. Agents in Action LLMs, Tools & Reasoning Progress broadcasting ▪ Using of real-time / streaming APIs ▪ Reasoning and current steps are always visible to the user ▪ User can interrupt / stop the process ▪ User can inject (corrective) messages at every time
34. Example Agent Agents in Action LLMs, Tools & Reasoning DEMO
35. Thank you! Demos: https://github.com/thinktecture-labs/basta-2025-agents Sebastian Gingter sebastian.gingter@thinktecture.com https://thinktecture.com/sebastian-gingter