Grounded answers
Every response is anchored to your real sources and cites them inline. No confident nonsense — just verifiable truth.
Open source · AI-native · MIT licensed
The AI that
The AI that
actually knows
your product.
Cap Chew captures everything your product knows — docs, code, tickets, conversations — and turns it into an always-on Product Expert your team and customers can just ask.
Product Expert
live
How do I rotate API keys without downtime?
Roll keys with the --grace flag so the old and new key are both valid during the overlap window:
$ capchew keys rotate --grace 24h
Requests keep authenticating throughout — no restart, no dropped traffic.
# docs/security/keys.md
# api/auth/rotate.ts
Ask anything about your product…
Captures knowledge from the tools you already use
GitHub
Slack
Notion
Linear
Zendesk
Confluence
Intercom
Google Drive
Cap Chew · Product Expert
One expert. Everything your product knows.
Product Expert reads your entire surface area — documentation, source code, support history, and team chatter — and reasons over it like your most senior engineer. Ask in plain language; get grounded answers with receipts.
- Grounded in your sources, never hallucinated
- Every claim links back to the exact file or ticket
- Drop it into your docs, your app, or Slack
- Re-indexes automatically as your product evolves
docs
code
tickets
chat
3.2savg. cited answer
100%traceable to source
∞questions, 24/7
Why Cap Chew
Built to be trusted, owned, and extended.
Everything you need to ship an AI expert you can actually put in front of customers — without handing your knowledge to a black box.
Ingests everything
Docs, source code, support tickets, Slack threads, PDFs. If your product knows it, Cap Chew captures it.
Private by default
Self-host on your own infra. Bring your own model. Your knowledge never leaves your perimeter unless you say so.
Open & extensible
MIT licensed, plugin-based, and hackable to the core. Write a connector in an afternoon and ship it.
Always current
Connect once and Cap Chew keeps itself in sync. Merge a PR, close a ticket — the expert already knows.
Drop-in everywhere
A one-line embed widget, a REST + streaming API, and a Slack bot. Meet your users wherever they ask.
How it works
From scattered knowledge to expert in three steps.
01
Connect your sources
Point Cap Chew at your repos, docs, and tools with a config file or a few clicks. OAuth and tokens, nothing exotic.
02
Capture & index
Cap Chew chunks, embeds, and links everything into a knowledge graph — keeping track of exactly where each fact came from.
03
Ask anything
Your team and customers ask in plain language and get expert, cited answers — in your docs, your app, or Slack.
Get started
Running in under five minutes.
Spin up Product Expert locally, point it at a source, and ask your first question. No account, no waitlist, no card.
$
npx @capchew/expert init
Prefer containers? docker run capchew/expert · self-hosting guide →
import { ProductExpert } from "@capchew/expert"; const expert = new ProductExpert({ project: "acme" }); // Ask in plain language — get a grounded answer const answer = await expert.ask( "How do I rotate API keys without downtime?" ); console.log(answer.text); // the cited answer console.log(answer.sources); // [{ file, url, score }]
Open source
Yours to read, run, and reshape.
Cap Chew is MIT licensed and built in the open. Audit every line, self-host with zero lock-in, and shape the roadmap alongside a community that believes product knowledge should belong to the people who build it.
12.4kGitHub stars
240+contributors
MITlicense
60+connectors
Give your product a voice.
Capture what your product knows today. Let everyone ask it anything tomorrow.