What these pages cover

Each use case describes one workflow as it runs in production: what triggers it, what context it needs, which decisions it is allowed to make on its own, which ones stop for a human, and what it writes back to your systems. They are patterns rather than packages — the shape holds across businesses, the rules and thresholds inside it are set during the audit against your own data.

How to pick the first one

The first workflow to automate is rarely the largest one. It is the one that runs often enough for the saving to be measurable within a quarter, has a rule set someone can state out loud, and fails visibly rather than quietly when it gets something wrong. A workflow that runs twice a week, or whose correct answer is a matter of judgement no one has written down, will not repay the work of automating it and will make the next project harder to fund.

What every pattern includes

Regardless of the workflow, the same controls ship with it: a defined owner, a success metric agreed before the build, permission scopes limited to the records the automation actually touches, human approval on anything that sends money or a commitment outside the company, an audit log of what the system did and why, and a documented fallback for when a model or an API is unavailable.

Discuss which one fits