Self-hosted · Groq-powered

Your docs, answering for you.

Paste your Groq key, drop in your help docs, and publish. The bot answers from your sources only — and says so when it cannot. Nothing leaves your stack except the inference call.

Build your first bot I already have an account

Free to run. You only pay Groq for the tokens your bot uses.

From signup to published bot in under five minutes

01
Name your bot
Give it a name and a persona in one short form.
02
Add sources
Upload files, paste a URL, or write Q&A pairs.
03
Paste your Groq key
Validated inline before it is stored.
04
Test & publish
Chat with it live, then grab the embed snippet.

Everything the MVP needs, nothing it doesn't.

Train on your own content

Drop in PDFs, DOCX, Markdown or plain text, crawl a site or sitemap, and add hand-written Q&A pairs that outrank everything else in retrieval.

Your Groq key, your bill

Each bot uses the key you paste in. It is validated on save, encrypted at rest, decrypted only server-side at inference time, and never reaches the browser.

Isolated per chatbot

Every bot gets its own vector collection. One tenant’s documents, embeddings and conversations can never surface in another tenant’s answers.

Publish in one click

Get a shareable chat link and a one-line script tag. The widget floats on any site and never exposes a credential.

Streaming answers

Token-by-token responses over SSE, with Groq’s inference speed putting the first token on screen in well under two seconds.

Logs and analytics

Every exchange is stored per chatbot: full transcripts, fallback rate, latency, token usage and the questions people actually ask.

Ready when you are.

Create a free account