Learning roadmap
AI App Development
Ship a real AI app. From your first prompt to a tool-using agent with a budget cap.
5 stages~40 h
That leaves 5 stages, ~40 hours of focused work. You can clear it in any order, but the stages compound — try them top-to-bottom your first time.
- You'll pick up:
tokens, context window, temperature, top p, streaming, cost per token
Practice quizzesLab taskA "prompt lab" page: send the same prompt to two brains side by side, diff the answers, and log which patterns changed the quality.
Build it in the LabCheckpointYou can predict (within an order of magnitude) how a prompt change will affect output and cost.
What you'll be able to do
- Wire up a Claude/GPT-class API into a working app.
- Design prompts that reliably produce structured output.
- Build a tool-using agent with a critic / evaluator loop.
- Ground answers in your own data with a vector store.
Or try a related path