AI Agent Frameworks
For engineers who can call a model API and now need to choose and use an agent framework for real work. You will build agents with LangChain and LangGraph, understand autonomous loops, CrewAI crews and Semantic Kernel's function registry, and pick the right tool for each job by its constraints.
Course Content
Why agent frameworks exist, how LangChain's create_agent and LangGraph's state graphs work, and three hands-on exercises comparing a tool-calling agent with a graph workflow. 6 lessons, about 75 minutes.
The goal-stack and task-queue patterns behind AutoGPT and BabyAGI, built from scratch with the budgets that make them stop, plus role-based crews in CrewAI. 3 lessons, about 40 minutes.
How Semantic Kernel's function registry, plugins, automatic function calling and memory work, and how they carry over to its successor, Microsoft Agent Framework. 3 lessons, about 40 minutes.
A way to choose a framework by your constraints, then a hardened multi-tool research agent you build, deploy and benchmark honestly. 3 lessons, about 40 minutes.