Practice useful AI skills through focused lessons.
Practice useful AI skills through focused lessons.
Practice AI skills through guided tasks drawn from everyday work.
MicrocourseIntermediateSeparate and combine project changes with Git branchesUnderstand branches and merges, choose when work needs to stay separate, and direct an agent to combine approved changes while protecting existing work.
MicrocourseIntermediateKeep AI tools on the same approved rulesYou will give the client-email rule an official file, connect Claude and Codex to it, test both tools on the same emails, and catch old copies without deleting genuine project exceptions.
MicrocourseIntermediateConnect an unlisted service to your AI assistantYou will confirm that the service publishes an application programming interface (API) or another supported route, that it covers your job, and that the app or script sending the requests, called the runtime, can sign in the way the service requires. When those checks pass, you will create a narrowly scoped credential, keep it out of chats and source files, test one small read, and revoke or replace it when it is exposed or no longer needed.
MicrocourseIntermediateCoordinate AI Assistants on a Multi-Step JobYou will plan a job with connected stages, decide which parts can run at the same time, and give each helper a clear assignment. You will combine their results, check them against the sources, and compare the arrangement with one agent before deciding whether the extra coordination helps.
MicrocourseIntermediateCompare AI Drafts and Keep the Best VersionYou will compare drafts that take different approaches to the same job, using the same written criteria and the same sample cases, called representative tests. You will keep the strongest complete draft safe, test any borrowed passage inside the whole draft, and stop when another round gives no useful improvement.
MicrocourseIntermediateCheck AI Work Against Written StandardsYou will choose checks for a finished draft, test an AI reviewer against examples you have already judged, and require evidence for each finding. You will separate a missed error from a false alarm and decide, from what an error would cost, what the AI may repair and what goes to a qualified person.
