supermemory/apps/docs/company-brain/connectors.mdx
Dhravya Shah f882f1104d
docs: restructure documentation site and update integration UI (#1331)
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
2026-07-21 20:19:30 -07:00

64 lines
3.6 KiB
Text

---
title: "Connectors"
sidebarTitle: "Connectors"
description: "Bring knowledge in with data connectors, and act in live tools with tool connectors"
icon: "plug"
---
Company Brain has two kinds of connectors. They look similar on the connections page, but they do different jobs:
| | Data connectors | Tool connectors |
|---|---|---|
| **What they do** | Bring knowledge *in* | Let the agent *act* in the tool |
| **Examples** | Google Drive, Notion, OneDrive | GitHub, Linear, Sentry, Plain, PostHog, Granola |
| **Result** | Docs land in public channel memory and stay searchable | Live reads and writes (list PRs, create issues, check errors) |
| **When it runs** | Background sync on a schedule | In the moment you ask |
## Data connectors
Data connectors sync existing files and docs into **public channel memory** so answers are grounded in real material — roadmaps, specs, handbooks, design docs.
How it works:
1. An admin connects a source (Drive, Notion workspace, OneDrive, and similar).
2. Company Brain fetches, chunks, embeds, and indexes the content in the background.
3. It re-syncs on a schedule automatically — you don't re-upload when a doc changes.
Connecting a data source is a **team-level action**. What comes in is visible org-wide, same as anything from a public channel — see the [permissions graph](/company-brain/permissions) for exactly who can read what.
<Note>
A data connector is only as useful as the docs you point it at. Start with the handful of sources people actually re-read — product specs, the handbook, the latest roadmap — rather than every folder in Drive.
</Note>
## Tool connectors
Tool connectors are live integrations (MCP-based under the hood). They don't just index past content — they read and act in the tool *right now*:
- **GitHub** — open PRs, recent commits, repo context
- **Linear** — find or create issues, check status
- **Sentry** — what's actually erroring in prod
- **Plain** — customer support tickets and history
- **PostHog** — product analytics
- **Granola** — meeting notes and decisions
- **Custom servers** — wire up your own MCP endpoint when the catalog doesn't cover a tool
You can also connect tools at two scopes — **Organization (shared)** or **Personal (yours)**. The full rule of thumb lives on [The permissions graph](/company-brain/permissions): reads prefer your personal connection and fall back to the org one; writes always run under your own account so the action is attributed to you.
If neither you nor the org has a tool connected, but a teammate does, Company Brain can ask them to **lease** temporary access for that one request — see [Leasing](/company-brain/permissions#leasing-borrowing-access-for-one-request).
## Which one do I need?
- **"What's in our Q2 roadmap?"** → data connector (Drive/Notion/OneDrive already synced)
- **"What are my open PRs?"** or **"Create a Linear issue"** → tool connector (GitHub / Linear)
- **"What did we decide in the Acme call?"** → tool connector that also brings knowledge in (Granola), or a data connector if notes live in Drive/Notion
You almost always want both: data connectors for the long-lived knowledge base, tool connectors for the live work happening this week.
<CardGroup cols={2}>
<Card title="Automations & proactiveness" icon="wand-magic-sparkles" href="/company-brain/automations">
Scheduled digests and unprompted replies that use these connections.
</Card>
<Card title="What you can do" icon="sparkles" href="/company-brain/use-cases/overview">
Walkthroughs of support, incidents, PRs, meetings, and more.
</Card>
</CardGroup>