Internal knowledge base software that answers from your own documents
Most internal wikis fail the same way: nobody keeps them current, so people ask a colleague instead. ChatDocument skips the rewrite. You point it at the documents your team already maintains, the handbook, the policies, the process guides, and staff ask questions in plain English. Every answer quotes the page it came from, so the source of truth stays the document, not a copy of it.
An AI knowledge base software setup in an afternoon
- Spaces per topic or department, for example HR, Finance, Sales enablement, IT
- Roles in each space: owners manage members and sources, editors add documents, askers ask
- Invites by email, with seats from 5 on Team to 50 on Enterprise
- Connected Google Drive, SharePoint and OneDrive or Dropbox folders that sync on a schedule
- Saved threads and shared answers by link
- Audit log of questions and cited pages on Business and Enterprise
Keep the documents, drop the copy and paste wiki
A classic knowledge base needs someone to turn every PDF into articles and keep both in sync. With ChatDocument the policy PDF is the article. When HR updates the travel policy in SharePoint, the synced folder brings the new version into the space, and the next answer cites the new page. Word, Excel, PowerPoint, Markdown and web pages by URL work the same way, and scanned PDFs are read with OCR.
A knowledge base that never invents a policy
A generic chatbot asked "how many days of parental leave do we offer" will produce a plausible number from somewhere. ChatDocument answers only from the documents in the space the employee is asking. Each cited passage is checked in code to exist on its page before the answer is shown. If the handbook does not cover the question, the employee reads that it is not in the documents and sees the closest passages, which also tells the space owner what is missing.
- Answers only from your files
- Document name and page on every statement
- Clear "not in the documents" instead of a guess
What teams put into their internal knowledge base
- Employee handbook, benefits guides and leave policies, see HR policy chatbot
- Product documentation and support macros, see AI support answers from documentation
- Past proposals and security questionnaires, see RFP AI
- Finance procedures, expense rules and approval limits
- Onboarding packs that new hires can question in their first week
Published prices, no sales call
Every plan is self-serve. Yearly billing is the default and saves 50%. Questions are pooled across the seats of the plan.
| Plan | Yearly billing | Monthly billing | For a knowledge base |
|---|---|---|---|
| Team | 74 USD per month (888 USD per year) | 149 USD per month | 5 seats, 2,000 questions per month, 8,000 new pages, 15 spaces, all three drives, roles |
| Business | 249 USD per month (2,988 USD per year) | 499 USD per month | 15 seats, 7,000 questions, 30,000 pages, unlimited spaces within page limits, audit log, widget and API |
| Enterprise | 745 USD per month (8,940 USD per year) | 1,490 USD per month | 50 seats, 20,000 questions, 100,000 pages, SSO (SAML), SLA 99.9, invoice billing |
A single person can start on Starter at 24 USD per month billed yearly or 49 USD monthly, with 3 spaces and one connected drive. Full table on chat with PDF pricing.
FAQ
Who can see which documents
Only members of a space see and ask its documents. Owners invite members and set each person as owner, editor or asker.
Is our content used to train AI models
No. Content is encrypted in transit and at rest and processed by AI model providers bound by contract not to train on your content. See secure AI document chat.
What happens when someone leaves the company
Remove them from the space or the account. With SSO (SAML) on Enterprise, access follows your identity provider.
How are questions counted for a team
Questions are pooled across seats, with a daily cap per plan (250 on Team, 800 on Business, 2,000 on Enterprise). One question is one cited answer.
Give your team answers with the page attached
Create the account, connect one folder and invite the people who ask the most questions.