How LLM agents call tools in a loop, how to design tools they use well, and how to keep agents safe and reliable.
An agent is a language model that can take actions — searching, calling APIs, running code — and decide what to do next based on the results. This course explains the pattern behind every agent framework, then focuses on the parts that decide whether an agent works in practice: tool design, guardrails and evaluation.
The code is provider-neutral: every major model API supports tool use in a similar way, so check your provider's documentation for the exact names.
4 lessons · 1 hr 2 min
Workflows versus agents, and when each fits.
Tool definitions, tool calls and results, in a provider-neutral loop.
Names, descriptions, inputs and outputs that make agents reliable.
Limit what agents can do, keep people in the loop, and measure reliability.
No ratings yet.
Sign in to leave a rating. Sign in
Use @handle to bring somebody in, and #tag to file it.
Nothing here yet. Ask a question, or say how you used it.
Sign in to join the discussion. Sign in
2 min read
2 min read
2 min read
2 min read
From a promising demo to something you can operate: scope, evaluation, interface and rollout.
Build test sets, grade outputs with code, people and model-based graders, and catch regressions before your users do.
Turn text into vectors, build a semantic search index, and ground an LLM's answers in your own documents.