Company Brain

Company Brain is a Slack bot that keeps a memory of what people say in its channels and searches it with each asker's own permissions. When asked, it opens issues, reads pull requests or runs code in a sandbox. It was a paid supermemory product and is now open source under Apache 2.0.

What Company Brain can draw on depends on where the question is asked. In a public channel it uses the shared memory the whole workspace can see. In a private channel it adds that channel's memory, and in a DM it can also use the asker's personal memory and every private channel that person belongs to. Tool access follows the same rule: write actions run under the asker's own connection, and if a request needs a tool only a teammate has connected, that teammate gets an approve or deny card in Slack. GitHub, Linear, Notion, Google Workspace and other services connect over MCP, so the bot can open the issue or read the PR it was asked about. It can join a conversation unprompted when it has something relevant, with proactivity set per workspace and per channel. It also posts scheduled digests to channels or DMs. Skills and a workspace prompt teach it house processes and tone. A web app at the Worker's root shows a graph of stored memories plus settings for tools, models, automations and skills. The Deploy to Cloudflare button asks for a supermemory API key and one model key from Anthropic, OpenAI, Google, xAI or OpenRouter. It provisions a D1 database, a KV namespace, Durable Objects for the agent and a Workers AI binding, and the Worker applies its own D1 migrations on the first request.

Built for: Startups and small teams that work in Slack and want a shared assistant with memory and tool access, running in their own Cloudflare account

What you can use it for

• Find who owns a project or what the team decided from past Slack threads
• Open a Linear or GitHub issue from a Slack conversation
• Post scheduled digests to a channel or a DM
• Answer support and incident questions with context from connected tools
• Run a script against a repo and get a chart or CSV back in Slack

Deploy

• One-click · main path

  Branch main

  Deploy to Cloudflare

• CLI

  Branch main

  wrangler.jsonc ships placeholder IDs. Replace the D1 database_id and the BRAIN_KV id with the values the two create commands print before running bun run deploy. The README itself documents only the Deploy button.

  git clone https://github.com/supermemoryai/company-brain.git
  cd company-brain
  bun install
  bunx wrangler d1 create company-brain
  bunx wrangler kv namespace create BRAIN_KV
  bun run deploy
  bunx wrangler secret put SUPERMEMORY_API_KEY
  bunx wrangler secret put MODEL_API_KEY

  Copy

After the button deploy, open /setup, create the Slack app from the generated manifest and paste in its client ID, client secret and signing secret. Then sign in with Slack and install the app.

What it needs to run

• Workers AIAI
• D1 databasescompany-brain
• Durable ObjectsCompanyBrainAgent, Sandbox
• KV namespacesBRAIN_KV

Good to know

Slack is the only chat platform; Microsoft Teams support is an open proposal. Memory is stored in supermemory's hosted service, so team conversations leave your Cloudflare account. Model calls go to the provider whose key you supply. The first person to sign in with Slack becomes the owner, so finish /setup right after deploying. An open bug report says that on a fresh install, DMs do not yet read private-channel memory as the README describes. The repository was published in late September 2026.

Cost on Cloudflare

Where this comes from

• Repositoryhttps://github.com/supermemoryai/company-brain
• App websitehttps://x.com/DhravyaShah/status/2103668051468300701?s=20
• Documentationhttps://github.com/supermemoryai/company-brain/blob/main/docs/guide/README.md

An alternative to Slack AI.