Generative AI at Work / Unit 9 GAI 219 review, 2 to 3 minutes
Your Team's First Month
The idea
The sixth capstone, constructed for this course, turns the course into a first month for a team starting to use generative AI deliberately, in the spirit of the method's advice to pilot before scaling.
In practice
In week one, the team chooses two tasks, one low-stakes and one moderate: drafting internal meeting notes, and summarizing client documents for account managers. For each, it writes the one-sentence use description and runs the twelve-risk scan, and it confirms which tool is approved for client material. In week two, it writes its rules: approved tools, what may never be pasted, forbidden uses, the review level for each task, and the logging rule, on two pages, with a named owner. In week three, it builds and tests prompts. For each task the team writes success criteria and collects real test cases, including an awkward one, then improves the prompts one change at a time. The prompts that pass go into a shared library with their test cases. In week four, it runs both tasks for real, logs every output as accepted, edited or rejected with a reason, and meets at the end of the month to read the log. Edits on the summaries are mostly for missing points, so the summary prompt gets a required-points list. The meeting notes need almost no edits, so their review stays light. The team agrees what to try next, and when to run the risk scan again.
Where people slip
The month produces two working uses rather than twenty experiments, along with rules, a library and a log that make the third use faster and safer. Starting small is not caution for its own sake. It is how a team learns where its own risks actually are.
Your place on the trail
Every lesson rests on earlier ones and opens later ones. Tap any to jump.
Builds on
Unlocks
- Nothing yet depends on this.
Something wrong in this lesson? Flag it · Sources for Generative AI at Work