ai

what /ai/ is

the /ai/ pillar is for engineers who already ship software and now need to ship agents. not "what is an llm." not "prompt engineering tips." every lesson assumes you can read typescript and have shipped at least one production system; we spend the time on the parts that actually break in production — runtime contracts, memory boundaries, tool schemas, identity propagation, observability, and the diff between a demo and a deploy.

how the lessons work

every lesson is a runnable typescript editor in your browser. types are stripped and code executes in a module worker — no server, no keys, no costs. for the agentcore course, we publish a mocked bedrock-agentcore sdk with the same public surface as the real one, so the code you write here is the code you'd ship, minus the deploy. for agentic patterns, the model is replaced with a scripted stand-in so you can iterate on the loop without burning tokens. nothing you type leaves your browser.

what makes this different

most agent tutorials show you the happy path on someone else's framework. the courses here cover the unhappy path — what fails first under load, where the contract between agent and tool gets ambiguous, when memory is the wrong abstraction. opinions are owned, citations are inline, and the lesson list above is updated in place as the underlying tools evolve.