RAG chat with citations, not hallucinations

AI answers wherever your users work

Index your docs, GitHub repos, Shopify store, SharePoint, and internal wikis. Surface cited answers in your widget, helpdesk, Slack, or via API. Automate support, build multi-agent workflows, and connect any AI tool via MCP — all from one platform.

5min
To first answer
20+
Integrations
99.9%
Uptime

CONNECTS TO YOUR EXISTING STACK

Slack
Discord
Teams
GitHub
GitLab
Notion
Confluence
Google Drive
SharePoint
Shopify
Zendesk
Freshdesk
FEATURES

Everything you need to build a knowledge-powered product

From crawling your first docs URL to running AI evals on every answer

AI Chat with Citations

Every answer links back to the exact source — doc page, ticket, or repo file — so users trust the response and can dig deeper.

Support Automation

AI copilot drafts replies inside Zendesk, Freshdesk, Front, and Jira Service Management. Deflect tickets before they reach an agent, auto-triage the rest, and test with simulation mode first.

Agent Platform

Build coordinator-to-specialist agent graphs with human-approval steps and full trace observability. Open protocols — MCP, A2A, and the @beforequery/agents TypeScript SDK — let any AI tool connect.

Website & Docs Crawler

Point BeforeQuery at any URL. It discovers pages via sitemap and links, extracts clean markdown, and re-syncs automatically when content changes.

AI Actions

Call your own customer APIs — order status, account lookups, subscription details — directly from chat. Set auto-approve or human-approval rules per action type.

20+ Connectors

GitHub, GitLab, Notion, Confluence, Google Drive, Shopify (products and pages), SharePoint, Slack, Discord, Teams — plus website crawl and file upload. One index, everywhere.

Widget, React SDK & APIs

One script tag for the hosted widget, a <BeforeQueryChat /> React component with a headless hook, raw REST/SSE APIs, and @beforequery/agents for building custom agent workflows.

Smart Retrieval & Re-ranking

LLM query rewrite and expansion surfaces results even when users phrase things differently. Cross-encoder re-ranking puts the best chunks first, every time.

Collections & Versioned Docs

Group sources into collections with per-collection client-key permissions. Serve v1, v2, and latest independently — users always get answers from the right version.

Analytics & AI Evals

See which questions get asked most, which go unanswered (your docs gaps), track daily conversation trends, and run automated groundedness and citation-quality evals.

HOW IT WORKS

From docs URL to AI-powered answers in minutes

01

Crawl

Add a docs URL, connect a GitHub or GitLab repo, or link a Notion/Confluence workspace. BeforeQuery crawls and extracts clean content automatically. Git-push webhooks keep repos in sync on every commit.

02

Chunk & Embed

Content is split into semantic chunks and embedded with a high-quality embedding model. Stored in a hybrid vector + keyword index.

03

Answer

User asks a question — in your widget, via the React SDK, in Slack, or Discord. BeforeQuery rewrites and expands the query, retrieves and re-ranks the best chunks, and generates a grounded answer with citation links.

04

Improve

Analytics show which questions succeed and which expose gaps. AI evals score groundedness. Your docs get better over time.

RAG WITH CITATIONS

Answers your users can verify

Every BeforeQuery response includes inline citations pointing to the exact doc page, changelog entry, or GitHub file the answer was drawn from. Users click through to the source — not away to Google.

  • Hybrid semantic + keyword retrieval with RRF fusion
  • LLM query rewrite, expansion, and cross-encoder re-ranking
  • Suggested follow-up questions and related articles
  • Confidence scoring and transparent fallback messaging
  • Multi-collection queries across versioned knowledge sources
Learn about AI Chat
BEFOREQUERY CHAT
USER
How do I authenticate with the SDK using an API key?
BEFOREQUERY
To authenticate using an API key, pass it as the apiKey option when initializing the client:
const client = new BeforeQuery({ apiKey: process.env.BQ_API_KEY });
Sources:Authentication GuideSDK Reference
TESTIMONIALS

Loved by developer-facing teams

BeforeQuery cut our support ticket volume by 58% in the first month. The citation links mean developers trust the answers — they're not second-guessing the AI.

Priya Nair
Head of Developer Relations, Supabase

We hooked it into our Slack workspace and Confluence in an afternoon. Our on-call engineers now get accurate answers to internal process questions instantly.

Tom Richter
Engineering Manager, Grafana Labs

The analytics dashboard showed us exactly which questions our docs were failing to answer. We fixed those gaps and our activation rate jumped.

Ana Castillo
Product Manager, Clerk

Ready to give your docs an AI brain?

Connect your first knowledge source and get your first AI-powered answer in under 5 minutes.

No credit card required