What You Could Build This Weekend

You now know more about what actually makes something an "agent" than most people posting about AI on the internet. Here's what that knowledge is actually good for.

Three real starting projects

A research digest agent. Give it a topic, have it search, pull the most relevant sources, and summarize what changed since last time you checked. This is a perfect first project because the loop from the last lesson maps onto it almost exactly: search, look at results, decide if you have enough, search again if not, summarize.

An inbox triage agent. Have it read incoming messages, decide which ones actually need a response today versus which can wait, and draft replies to the urgent ones for you to approve. This one teaches you something the crash course didn't cover: how agents handle situations where they shouldn't act alone, and need a human to approve the risky step.

A personal research assistant. One that remembers what you've asked before, builds on past conversations instead of starting fresh every time, and can pull in outside information when it needs to. This is where memory and retrieval — the stuff that turns a one-off tool into something that actually gets more useful the longer you use it — start to matter.

The honest gap

Here's the thing this free crash course deliberately doesn't cover: how you actually build any of these. What "planning" really looks like inside the model. How tool-calling is implemented under the hood, with real failure modes and how to handle them. How to give an agent memory that doesn't fall apart after a few steps. How to know if what you built actually works, instead of just seeming to work in a demo.

That's the entire content of Intro to AI Agents — six modules, eighteen real lessons, quizzes that actually check you understood it, and a final exam, built for someone who's never written a line of code. If any of the three projects above sounded exciting rather than intimidating, that's the course to go build them in.

And if you already know how to code and want to go straight to the deep end — coordinating multiple agents together, production-grade guardrails, evaluation, deployment — Multi-Agent Systems & Orchestration is the advanced course built for exactly that.

You just finished the free preview. The next step is up to you.