The open-source database and workspace for AI agents
Busabase is an open-source (MIT) database and workspace for AI agents. It captures what agents produce — data, docs, skills, and apps — so it stays with your team in a reusable, structured form instead of evaporating in personal chat histories. The same engine runs as a local desktop app, self-hosted via Docker inside your own network, or as Busabase Cloud.
Store customers, projects, and inventory in tables with fixed columns, so every agent reads and writes the same fields.
Shared team memory: answers you can find, with a source, that someone has checked.
Internal tools built on the same data, so every screen shows the latest version.
Save a workflow once as a skill, and any agent can run it next time.
Document how your team does things; agents check the playbook before each task and follow it instead of improvising.
Connect 40+ external agents — Claude Code, Codex, Cursor, Hermes, OpenClaw, n8n, Gemini CLI, Zapier, LangChain, or your own — that all read and write the same workspace; you can also chat with a connected agent inside Busabase.
Every change carries who made it, what changed, and full history. Writes that matter can be held for a person to approve; rejected work never enters canonical data, and each decision stays attached for explainability.
The local engine is open source (MIT): Busabase Desktop keeps data in a folder on your machine with no usage tracking and offline support, or deploy the Docker image inside your company network.
Claude Code, Codex, Cursor, n8n, or your own agent connects through Agent Skills, MCP, OpenAPI, or the CLI — nothing to switch, agents keep working the way they do today, they just write into Busabase.
Each change carries who made it, what changed, and a full history. Writes that matter can be held for a person to review.
The next agent, a teammate, or an AirApp reads the same data, so nobody starts from zero.
Open source and free forever locally; Cloud starts free and grows per seat. Free plan at $0, Plus at $10/seat/month (billed $120/seat annually), and Pro at $20/seat/month (billed $240/seat annually), with 2 months free on annual billing.
$0
For one person keeping what their agents produce. Includes 1 space you own, 3 seats per space, 5,000 records per space, 1 GB attachment storage, and 1 self-hosted connection online.
$10 /seat/mo (billed $120/seat annually)
For small teams building on the same agent-written data. Includes 3 spaces you own, 10 seats per space, 20 GB attachment storage, and OpenAPI, webhooks, and MCP. Billed yearly with 2 months free.
$20 /seat/mo (billed $240/seat annually)
For growing workspaces that need room to scale. Unlimited spaces you own, unlimited seats, 100 GB attachment storage, and priority support (WhatsApp/WeChat/Email). Billed yearly with 2 months free.
Yes. The local engine is open source — you can run it yourself, no SaaS required. It is licensed under MIT.
Yes. Busabase Desktop keeps data in a folder on your machine, and the Docker image runs on your own server inside your network. Cloud is optional — it is the same engine either way.
Any agent that can call a REST API — Claude Code, Codex, Gemini CLI, n8n, Zapier, LangChain, or your own. Paste the SKILL.md and go.
Notion and shared drives are cabinets for what people write, and AI chat tools forget when the session ends. Busabase is where agent work lands: typed data, docs, skills, and apps that every agent reads and writes through the same structure, with a history of every change.
Rejected work never enters your canonical data. Every decision — approve, request changes, reject — stays attached so the record remains explainable later.
Yes. With the local engine your data never leaves your machine; on Cloud it stays in your own space. You stay in control.
The open-source local engine is free forever. Cloud starts free (1 space, 3 seats per space, 5,000 records), then Plus at $10/seat/month and Pro at $20/seat/month, billed annually with 2 months free.
Answer your AI agent's questions asynchronously in one UI
