Building AI Agents & Automations

by bam

Go past the intro: build AI systems that take action. This is the technical, builder-track course for people who ship. You'll learn what an agent actually is (and the more common case where you DON'T need one), the plan-act-observe loop, how to do tool/function calling well, when a fixed workflow beats an autonomous loop, how to handle memory and state, and how to wire real multi-step automations with triggers, queues, and human-in-the-loop checkpoints. Then the parts that separate a demo from production: orchestration patterns (routing, chaining, parallelization, evaluator-optimizer), guardrails and least-privilege (the agent can take real actions: sandbox it, confirm it, limit its scope), evaluating and debugging with traces, and shipping with monitoring and cost controls. Carries the Learn.WitUS trust DNA throughout: an agent that acts is a security surface. F2 (Building with AI) is the recommended prerequisite.