# Turn a good bot run into a saved routine

> Tested by AI Setups That Work. Small business · Intermediate · Operations. Updated 2026-11-02.
> Receipt: Ran on Grok Bot, Nov 2, 2026: a good lead-sweep run was written up as a 193-word routine and re-run on new sample emails against a real Google Sheet. It worked, but missed a customer saying yes and re-drafted old leads, so it was fixed and not scheduled.

Canonical page: https://www.aisetupsthatwork.com/setups/turn-a-good-bot-run-into-a-saved-routine/

## What it does
Save a bot run that worked as a written routine, test it on a second input, then schedule it with outward actions kept behind your approval.

## Who it's for
Owners who get a great result from a bot once and cannot get it again next week.

## Tools
- Grok Bot or your agent tool
- A second test input
- Your bot's scheduling or routines feature

## Steps
1. Right after a good run, use the prompt below to have the bot write the routine.
2. Read it and fix anything vague, especially what done looks like.
3. Run the routine on a different input, such as another week's emails or a different client.
4. If it works twice, schedule it.
5. Keep send, purchase, delete, publish and record changes behind approval. xAI documents approvals for these actions in Grok Bot.
6. Add the routine to your list so the monthly usage review covers it.

## Copy-paste prompt
Adapt bracketed fields to your situation.
```text
That run worked. Write it down as a reusable routine.

Include:
1. Name and one-line purpose
2. Inputs it needs and where they come from
3. Numbered steps, exactly as you did them
4. What done looks like
5. Actions that need my approval (send, buy, delete, publish, change records)
6. What to do if a step fails: stop and tell me
Under 200 words. Do not schedule it yet.
```

## Guardrails
- What it touches: Connected mail, calendar, sheets, or chat. Scoped to this job (Grok Bot or your agent tool; A second test input).
- Needs your OK: Send, pay, publish, and account changes stay with you. Treat agent output as drafts first.
- Safety level: Read-only
- Scheduling after one lucky run.
- Vague done-looks-like, so results drift.
- Scheduled routines that send without approval.

## Tips
- Two successful runs before scheduling.
- Name routines by the job, not the tool.
- A routine nobody uses is a usage cost; review monthly.

## Estimates (labeled, not measured)
- Time saved: About 10 minutes per routine.
- Money saved: Not stated in sources.
- Tool cost: Prices: Not stated in sources.

## Sources
Summaries use the public sources below, not independent verification.
- xAI: https://docs.x.ai/grok-bot/approvals-security-and-privacy (Grok Bot approvals, security and privacy: approvals for sends, spend and purchases; shared computer, local execution settings and connector hygiene (as cited elsewhere in this catalog))
- Anthropic: https://www.anthropic.com/engineering/building-effective-agents (Building effective agents: agents check progress against ground truth such as tool results at each step, can pause for human feedback at checkpoints or when blocked, and use stopping conditions such as a maximum number of iterations (read 2026-11-02 from a search result summary of the page))

## Get the free starter pack
Five beginner copy-paste AI setups, free by email: https://www.aisetupsthatwork.com/starter-pack/?utm_source=ai-agent&utm_medium=markdown&utm_campaign=setup-turn-a-good-bot-run-into-a-saved-routine

Related:
- [Make bot corrections stick](https://www.aisetupsthatwork.com/setups/make-bot-corrections-stick/index.md)
- [Organize your bots like an org chart](https://www.aisetupsthatwork.com/setups/organize-your-bots-like-an-org-chart/index.md)
- [Stop one job eating your bot usage](https://www.aisetupsthatwork.com/setups/find-which-bot-routine-burns-your-usage/index.md)
