AI data use

When you draft, revise, interview, check consistency or import a document, the app sends only the text needed for that request:

• Project facts from intake (name, sponsor, scope, governance level)
• Meeting notes and interview answers you add to the project
• Text of upstream documents in the same project that the draft draws from
• Your organization's templates, drafting guidance and knowledge-base documents
• The current document text when you revise, check or import it

Requests go from our servers through the Lovable AI Gateway to OpenAI models. Your browser never talks to the AI provider directly, and AI keys never leave our servers.

We do not use your content to train any model of our own.

Whether the AI gateway or the model provider keeps prompts and outputs, for how long, or uses them for training is governed by their terms. We have not yet confirmed this in writing, and we will update this page once we have.

• Every organization has a monthly AI allowance. Calls stop once it is reached.
• Admins see AI usage by project, person, action and model, with a warning at 80% of the allowance.
• Every AI call is logged with who made it, when, which model, and the token count. Prompt text is not stored in the log.
• Admins control what goes into prompts by editing their templates, drafting guidance and knowledge base.

AI output is a draft. It becomes official only after a project manager submits it and the approval committee, and any named document approver, reviews it and the chair approves it. Each version is kept, so you can always see what the AI wrote and what people changed.

On PM Pro, a solo project manager may self-approve documents and sign off gates; the approval record shows these as “Self-approved”. Committee and chair approval applies on Team and above.