AI Usage Policy
Rainbook's AI runs on models hosted by Atlassian. It reads what you can see in Jira and nothing more. Your data isn't used to train models.
Models
Rainbook's own AI features run on Forge LLM: Claude models that Atlassian hosts for Forge apps. Processing follows Atlassian's published regions for Forge LLM, under the Atlassian terms that govern Forge LLM, which keep customer data out of model training. No other AI provider receives your data from Rainbook. Rainbook Mail Sync can, in a later release and only when your admin adds your organisation's own Anthropic, OpenAI or Microsoft Azure AI (Azure OpenAI or Azure AI Foundry) key and chooses it, send the prompt and the records it needs to that provider instead; nothing is sent until then.
What each feature sends
Every feature sends only what the request needs, and only what the person using it can see in Jira. Email text is always passed as quoted data, never as instructions.
| Feature | What is sent to the model | Where the result goes |
|---|---|---|
| Ask for a report | Your question and the report definitions, then the results you can see when it explains them; Rainbook runs the query against only the records you can see | Your Ask thread, and a report or dashboard if you save it |
| Thread classification and summary | A captured thread's new messages and its previous summary | The Conversation, for whoever can see it |
| Record summary and next steps | The record's fields and the recent activity on it that you can see | The record, for whoever can see it |
| Health reasons | The record's health signals | The record's health reasons |
| Draft reply and follow-up | The whole thread and the records you can open | A draft you edit and send yourself; in the Gmail and Outlook add-ins it is inserted only when you press Insert |
| Duplicate check | The two records being compared | A suggestion to merge or keep both |
| Proposals | The records and activity an agent is working on that you can see | A proposed change that the person who can edit the record applies, with Undo |
After an AI reads walled or other people's private records, its text cannot go onto records more people can open for two hours. Rainbook's own tools enforce this; Atlassian's Jira tools follow Jira's permissions.
What is kept, and for how long
- Ask history: 13 months from last use. Its creator can delete any thread, or clear their whole history at once.
- Each AI result's prompt, sources and output: 30 days, for Show sources, feedback and diagnostics, in your own site. Show sources lists to each person only the sources they can open.
- AI-written summaries and reasons live on the record in Jira, like any other field or comment.
Email and your consent
When the Rainbook agent in Rovo or another AI client asks about email, it gets summaries of your own email only after you agree once: in Rainbook, open Settings › AI and your email and turn on "Use my email in AI clients". Summaries of other people's email reach AI clients only when an admin turns on "Include other people's email summaries for AI clients" in Settings › AI and Rovo; it is off until then. Rainbook's own summaries and drafts work inside Jira for whoever can see the Conversation. Email text is quoted data for the model and never an instruction.
Your data and the model
Data goes to the model only for the request in hand, from the records the person can see. Model answers are checked as structured data and turned into changes by Rainbook's own code; text from a model is never run as a query against your site.
Rovo and AI clients
The Rainbook agent in Rovo runs on your organisation's Rovo credits. Rovo gets the same results the Rainbook agent gives that person, email summaries included once the person has agreed (see Email and your consent). Claude, ChatGPT and other outside AI clients connect through Atlassian's MCP server once your organisation joins Atlassian's early access for them. Each person sees exactly what Jira lets them see. Data a person sends to an outside AI client follows your organisation's terms with that provider.
AI allowance
Rainbook's AI is included in your subscription, with fair-use limits. Each person can run up to 25 Ask analyses a day, within a monthly fair-use limit for each person; questions that fit a report don't count. Summaries and other automatic AI draw on their own monthly pool. Your admins can lower the limits or give a person more. Reports, search and everything else keep working.
On Standard, sites of 1 to 10 users use Rainbook free, with Ask included for about 60 to 100 analyses a month; summaries and drafts need a paid plan.
Rainbook's own AI, capture and background jobs are included in your subscription and use neither Rovo credits nor Automation steps. The Rainbook agent in Rovo, Rovo Studio agents that use Rainbook's tools, and Automation rules built from Rainbook's templates or actions use your Atlassian allowances: Rovo credits from 3 Dec 2026, and Automation steps from your first renewal on or after 3 Dec, shown per rule in your admin Usage tab.
Admin controls
In Settings › AI and Rovo, admins switch off Rainbook's own AI, or the Rainbook agent and tools in Rovo, separately, and turn each feature on or off: Ask, summaries, health reasons, drafts, duplicate checks, proposals and other people's email summaries for AI clients. Each change is recorded in the activity history.
Google user data
We use Gmail and Google Calendar data only under Google's Limited Use requirements; the Google section of our privacy policy gives the details.
Questions: privacy@rainbook.humanr.ai.