We’ve moved on from the world of AI chatbots to agents that actually do things and maybe even think (kind of). Agents combine multiple skills, and often multiple agents, to perform tasks and achieve an objective.

In this advanced session we’ll go beyond prompt engineering and RAG, and look at some of the ideas behind these techniques including function calling, fine-tuning, and multi-agent systems.

We’ll use Semantic Kernel as a toolkit, which is ideally suited to work with the Microsoft technology stack, and to build custom Copilots for Microsoft 365. But the principles also apply to other technologies like LangChain and AutoGen.