> For the complete documentation index, see [llms.txt](https://docs.varbase.vardot.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.varbase.vardot.com/content-designers/ai-in-varbase/run-an-ai-pilot.md).

# Run an AI Pilot

Start AI small: one team, two or three features, a few weeks, and numbers at the end. A pilot shows the client what AI is worth on their own content before they commit to more.

## Steps

1. Pick one team and two or three features, such as tag suggestions and alt text.
2. Agree the provider, where data may be processed, and a monthly budget.
3. Write the brand and editorial context with the client. See [Show a Client What Context Changes](https://docs.varbase.vardot.com/content-designers/ai-in-varbase/show-a-client-what-context-changes.md).
4. Train editors: where the AI help is, what it is good at, where it goes wrong, and the rule that the editor is responsible for everything published.
5. Run it for a few weeks and measure: time per task before and with AI, how often editors changed the suggestion, cost, and whether editors want to keep it.
6. Decide what to extend.

## Questions Clients Ask

* **What does it cost?** Most providers charge per use. Small features cost very little; bulk runs, such as alt text for a thousand images, cost more and should be planned. Budget it as a running cost.

## Planning Notes

* Five questions: which provider and where may data be processed; the monthly budget; which roles use which features; who writes and owns the context; who reviews the AI's work and how often.
* Assumption: provider usage for editorial help is low; measure it in the pilot month.
