Skip to content
New: simulate any flow against real records before it goes live. See what's new →

AI and your data

Last updated 15 October 2026

Open Settings → AI assistant to see how AI is used in your workspace.

What is sent, and when

  • Content goes to the AI provider only when an AI feature runs, such as the assistant and its tasks, the chat and text bots, AI steps in flows, drafting and voice input.
  • It sends what that feature needs: the request, and the records, emails or articles it reads. The assistant reads only what the person asking can see.
  • In chat, the assistant asks before it sends anything, deletes anything or changes who can see what, including adding a webhook or adding someone to a team.
  • Every workspace uses the same provider, named on our sub-processor list with what it keeps and for how long.
  • Switch the assistant off on that page and nothing is sent to an AI provider.

Where your credits went

The same page shows the last 30 days by what AI was used for and by which model, and lists things people asked the assistant to do that it only answered in words. A saved skill often fixes those. Spend by person is under Billing.

AI drafted

Help articles, quick answers, email templates and social posts the assistant writes for you carry an AI drafted label in their lists. It stays after you edit them: it says where the words started, so whoever publishes them knows to read them first.

The weekly AI report

Each Monday from 08:00 in your workspace's time, owners and admins get a short report: the questions your chat bot couldn't answer that week, each with an Add an answer link that opens a new quick answer with the question filled in, and what people asked the assistant to do that it couldn't. Nothing is sent in a week with nothing to fix. Switch it off under Weekly AI report in your notification settings.