This is where AI stops answering and starts acting. You will build agents that plan, use tools, remember, and work together — the systems behind the most exciting products being shipped anywhere right now.
Everything you have learned so far comes together here. Prompting becomes planning. Tool calling becomes real-world action. RAG becomes an agent's knowledge. And you learn the orchestration patterns that keep autonomous systems safe, observable and useful.
You will build with the same frameworks and protocols used in production teams today — not toy versions of them.
What separates an agent from a chatbot: goals, loops, tool use, reflection and stopping conditions — and when a simple pipeline beats an agent entirely.
Building agents as explicit graphs: state, nodes, edges and checkpoints, so complex behaviour stays debuggable instead of becoming a black box.
Orchestrating teams of agents with CrewAI and AutoGen: roles, delegation, shared memory, and the coordination patterns that actually work.
The three pillars of capable agents — designing good tools, choosing what an agent remembers, and letting it break big goals into achievable steps.
Working directly with the Anthropic Claude API, the OpenAI Agents SDK and Google ADK — the vendor-native ways to build production agents.
MCP, the open standard for connecting agents to tools and data: what it solves, how servers and clients work, and building your own MCP integrations.
Agents are the frontier of applied AI — and the most sought-after skill in the field. Learn to build them properly, free, with mentors beside you.