UseAgentBEFORE YOU RUN
Know the job and the reviewer.
This playbook is for teams creating their first reusable AI coworker.
TARGET OUTCOME
A named bot with a clear job, standing rules, a tested home thread, and one routine you can inspect.
Bring these inputs
Better source material makes the result easier to review and safer to reuse.
- One recurring task with a clear owner
- The source material the bot may use
- Three rules covering scope, limits, and escalation
COPYABLE STARTER
Start with this prompt.
Replace bracketed details, attach the source material, and keep the limits until your team has reviewed a few runs.
You are our support handoff bot. Every Monday, review the material in this thread and prepare a concise handoff. Include open items grouped by owner, anything blocked, the three decisions needed this week, and source links beside factual claims. Do not send messages, close tickets, or change source records. Flag missing context instead of guessing.
THE PLAYBOOK
Move from context to a reviewed result.
- 01
Choose one job
Start with a task that has a recognizable input and output. A weekly handoff is easier to judge than a vague instruction to help the team.
- 02
Write the boundaries
State what the bot may read, what it must never change, and who reviews uncertain or sensitive items.
- 03
Run it once by hand
Message the bot in its home thread. Check every section against the source material before reusing the prompt.
- 04
Add the routine
Schedule only the stable part. Keep publishing, sending, merging, or other consequential actions behind human review.
DELIVERABLES
What the run should produce
- Bot job statement
- Three standing rules
- Tested prompt
- One inspectable routine
HUMAN REVIEW
Check this before it leaves the team
- Every claim points to supplied context
- Unknowns are labeled
- No action happened outside scope
- A person owns the final decision
Keep exploring this workflow