Skip to main content
Repo gives your agents safe, cited access to your company’s Slack, Notion, Google Drive, and Gmail — without making you rebuild OAuth, sync jobs, chunking, embeddings, permission checks, citations, and audit logs for every project. Connect your tools once. Repo handles the context layer so you ship agents instead of infrastructure.
Availability today. Slack and Notion are open for self-serve connection. Google Drive and Gmail are gated during Google’s OAuth verification window — click Request access on either connector card and we’ll clear you within 24h.

What this looks like

Your agent asks:
“What did the customer say about pricing in Slack and Gmail this week?”
Repo returns a grounded answer with verifiable citations:
{
  "answer": "Acme flagged pricing concerns twice this week: [1] in #cs-acme they asked about the upcoming tier change, and [2] in an email thread with Sam they pushed back on the $299/seat number.",
  "citations": [
    { "index": 1, "sourceItemId": "uuid", "title": "Acme — pricing concerns" },
    { "index": 2, "sourceItemId": "uuid", "title": "Re: Pricing for renewal" }
  ],
  "context": [
    { "provider": "slack", "url": "https://acme.slack.com/...", "syncedAt": "2026-05-30T18:00:00Z" },
    { "provider": "gmail", "url": "https://mail.google.com/...", "syncedAt": "2026-05-29T22:14:00Z" }
  ]
}
Every answer is sourced. Every source carries scope + freshness metadata. Your agent can honestly say “I don’t have access to that” instead of hallucinating.

Why Repo

Without Repo, every production agent needs:
  • OAuth installs for each provider — different scopes, different consent screens, different token refresh logic
  • Background sync workers that handle rate limits, pagination, deletions, and back-pressure
  • Document chunking + embedding + vector storage
  • Citation-grounded prompt assembly so answers don’t fabricate sources
  • Per-agent permission scoping so one agent can’t accidentally see another customer’s data
  • Audit logs that prove what was retrieved, what was excluded, and why
Repo gives you that layer as an API. You bring the agent.

Who Repo is for

  • AI automation agencies shipping Slack/Drive-grounded agents for SMB and mid-market clients
  • AI-native B2B SaaS integrating Repo as the context layer behind their product
  • Internal platform teams at 50–500 person companies standing up their first production agents
If you’ve already built the brittle version of this for one customer and don’t want to do it again, you’re in the right place.

What you get

  • Source-backed answers. Every /v1/ask response includes citations to the original Slack message, Notion page, Drive file, or Gmail thread. No black-box generation.
  • Scoped API keys. Each agent gets its own key with per-action and per-provider scopes — a customer-support agent can read Slack but not Gmail; a sales agent can read Drive but not write back.
  • Freshness + audit. Every sync run, every retrieval, every key creation, every connector change is logged. Your security team can answer “what did the agent see?” the same way they answer “what did the user see?”
  • One integration surface. OAuth, chunking, embeddings, freshness windows, citation formatting, scoped retrieval — Repo owns it. You ship agents.

How it works

Connect sources

Browser-driven OAuth into Slack, Notion, Drive, and Gmail. Your team adds the bots and shares the right pages — Repo takes it from there.

Query context

Your agent calls /v1/search, /v1/context, or /v1/ask with a scoped API key and gets cited answers back.

Stay fresh

Background workers re-sync every 15 minutes. Your agent’s context window never drifts more than that.

Prove access

Every retrieval is logged with provider scope, exclusions, and citation counts. Compliance-ready by default.

Next steps

Quickstart

Sign up, connect Slack, mint a key, and get your first cited answer in under 5 minutes.