# AI and your diagram data

AI features send data only when you use them. Generating a diagram sends your prompt; editing one sends the prompt and the diagram source, because otherwise it cannot edit anything.

Nothing is sent in the background, and nothing is sent when AI features are switched off.

---

## When this matters

* Policy reviews covering AI use.
* Deciding whether to enable AI for a pilot.
* Explaining to a team what happens when they use the generate button.

---

## The detail

| **Thing**               | **What happens**                                  |
| ----------------------- | ------------------------------------------------- |
| Generating              | Your description of the diagram you want.         |
| Editing with chat       | Your instruction plus the current diagram source. |
| The Rovo agent          | What you ask it, within your own permissions.     |
| Browsing pages normally | Nothing. AI runs only when invoked.               |
| AI switched off         | Nothing, and the surfaces are removed entirely.   |

---

## A few things that catch people out

* Generated diagrams are ordinary diagrams afterwards. Nothing about them stays tied to AI.
* The Rovo agent operates with your permissions and asks before it changes anything.
* The switch is site-wide and lives in [Administration](https://help.gocapable.com/diagrams/administration.html).

---

## Related

[Hosting and data residencyWhere the services run.](https://help.gocapable.com/diagrams/hosting-and-data-residency.html)

[Security overviewThe one page to send to a security review.](https://help.gocapable.com/diagrams/security-overview.html)

[ExamplesWorking diagrams you can copy.](https://help.gocapable.com/diagrams/examples.html)

---

_Ask us the awkward questions. They have answers._