Turn the docs you already wrote into the answers users ask for
Docs teams are judged on questions their pages actually resolve, not on word count. BeforeQuery layers an AI assistant on top of your existing docs (Mintlify / Docusaurus / GitBook / ReadMe / VitePress / any HTML) — so the same pages answer questions on your site, in Slack, in the IDE, and via API — and closes the loop with gap analytics that turn unanswered questions into a prioritized editorial backlog.
Sound familiar?
Traffic without answers
Docs get thousands of pageviews and a support ticket queue that suggests nobody read them. Search returns 20 links; the reader picks none and files a ticket anyway. You have no idea which pages actually resolve intent.
You keep writing pages nobody finds
New pages ship every sprint, but the same 40 questions keep coming into support. No feedback loop tells you which docs are missing, which are stale, and which are miswritten — so the backlog is guesswork.
AI is 'coming' but nobody owns it
Every stakeholder wants AI on the docs; nobody wants to own retrieval quality, prompt engineering, or an eval harness. You end up staring at a Kapa demo and a build-vs-buy spreadsheet.
Answers on the docs site aren't the same as answers in Slack or IDEs
The support team runs one bot; DevRel runs another; a customer's Cursor asks a third. Every surface has its own quality, its own hallucination rate, its own audit trail. Nothing improves in sync.
How BeforeQuery helps documentation teams
One knowledge base, every surface
The same indexed docs answer questions on the site widget, in Slack / Discord / Teams bots, inside your support form (form deflector), inside your helpdesk (Support Ticket AI), and inside Cursor / Claude via MCP. Improve a page once, every surface improves.
Gap analytics that become a backlog
Every unanswered or low-confidence question is logged. The gap report ranks missing knowledge by volume, and doc-proposal scans draft candidate pages — your team accepts or dismisses each draft. AI proposes, humans publish.
GEO-friendly by design
Answer-shaped H2s, citation-carrying responses, MCP endpoint per KB, and public API — so ChatGPT, Claude, and Perplexity can retrieve your docs directly instead of hallucinating from training data. Your docs become the source of truth outside your site too.
Runs on top of the docs platform you already use
First-class installers for Mintlify, Docusaurus, GitBook, ReadMe, VitePress, MkDocs, Sphinx, Next.js, WordPress, Shopify, Webflow, and plain HTML. No migration, no rewrite — one script tag or a webhook away.
Evals that trend, not one-off benchmarks
LLM-as-judge scores every answer for groundedness and citation quality. You see the trend across weeks — did the last docs migration improve or regress the assistant? — instead of guessing after every content push.
Relevant sources & integrations
Common knowledge sources and delivery channels for documentation teams.
Frequently asked questions
Does BeforeQuery replace our docs platform?
No — it layers on top. Keep publishing on Mintlify, Docusaurus, GitBook, ReadMe, VitePress, or your custom site. BeforeQuery indexes the published site (or the source repo) and adds the AI answer layer, widget, and bots.
How do we know if the AI is answering well?
LLM-as-judge evals score every answer for groundedness and citation quality; the score trends per knowledge base per week. Combined with user thumbs-up / thumbs-down and the gap report (which questions abstained), you see quality as a moving number, not a demo.
Does the assistant hallucinate?
Answers are generated only from retrieved chunks of your indexed content, the prompt enforces grounding in those sources, and an abstention gate refuses to answer when retrieval confidence is below threshold — you get 'I don't know' instead of a plausible-sounding lie. Every answer carries clickable citations.
Will ChatGPT / Claude / Perplexity actually cite our docs?
The MCP server + public API surface makes your docs directly consumable by AI IDEs and assistants that support external retrieval; the answer-shaped page structure, citations, and llms-friendly outputs help crawler-based AIs (Perplexity, Gemini) index and cite you correctly.
How long does setup take?
Most teams are live the same day. Connect a source — a docs URL, GitHub repo, Notion workspace, or a file upload — and BeforeQuery crawls, normalizes, chunks, and embeds the content automatically. The website widget is a single script tag, and the Slack/Discord bots install in a few clicks.
Are answers backed by citations?
Yes. Every answer includes a structured citations list with the URL, title, and excerpt of each source used, so users can click through and verify any claim against the original document.
Is my content used to train AI models?
No. Your indexed content, embeddings, and conversations are isolated to your workspace and are only used to answer your own questions. Content is never shared across customers, and data is encrypted at rest and in transit.
Is there a free plan?
Yes. The Free plan includes 1 knowledge base and 100 AI questions per month — no credit card required. Pro comes with a 14-day free trial, and Enterprise plans add SSO, audit logs, and advanced governance features.
Solutions for other teams
Support Teams
Deflect repetitive questions at the widget and the support form, and give agents cited draft replies inside the helpdesk — so humans handle only the tickets that need them.
DevRel & Docs Teams
Turn your documentation, API reference, and community threads into a cited assistant that lives on your docs site, in Discord, and inside the IDE via MCP — and shows you exactly which docs to write next.
Engineering Teams
Runbooks, ADRs, READMEs, and postmortems become a cited assistant in Slack and the IDE — so senior engineers stop being the search engine for the rest of the team.
Bring cited AI answers to your documentation teams
Connect your knowledge sources and see cited AI answers in minutes — free, no credit card required.
Get Started Free