Doc Chat
Ask questions about any uploaded document. Backed by the same AI that runs the engines — and metered the same way: one run per question.
What it does
Doc Chat gives you conversational access to any document in your project's library. Ask free-form questions ("What's the LD rate in the subcontract?" or "Summarize the force-majeure clause") and get answers grounded in the document with quotes and page references.
Doc Chat draws on the same monthly run pool as the engines: each question you ask costs one run. There is no separate chat quota. Responses include citations so you can verify the answer against the source.
When to use it
- Spot questions on a contract, spec, or RFP without running a full engine.
- Searching within a long document — "where does it mention warranty periods?"
- Comparing two concepts across multiple documents.
Step by step
Open Doc Chat
Sidebar → Documents → click a document → Chat with this document. Or from an engine result, use the inline chat at the bottom.
Ask questions
Natural language. Follow-ups are context-aware.
Verify citations
Every answer includes page / section references. Click to jump to the source.
Every control, explained
Chat inputFree-form text. Ctrl+Enter to send.
Citation linkJumps to the source passage.
Clear chatStarts a fresh context. Previous conversation is archived.
Multi-doc chatAdd multiple documents to the context — ask questions that span documents.
Export transcriptPDF / text of the Q&A session.
Limits
Doc Chat has no quota of its own. Every question spends one run from your plan's monthly pool — 15 on Trial, 100 on Pro, 300 on Business, 2,500 on Enterprise — shared with engine runs and counted across your whole organization. The Free plan includes no runs, so Doc Chat is unavailable there. Quota resets on the 1st of each month. There is also a short-term per-user and per-organization rate limit to keep one person from draining a shared pool in a burst.
Troubleshooting
Click the citation to verify. If it's wrong, reply "that's not the right section — the LD rate is in section 14.3" and the model will correct.
It's honest when content isn't present — don't try to coerce an answer. Check if the question actually requires a different document.