Built for enterprise scale
.Open for everyone.
Brigade is a crew of self-improving AI agents on a real org chart, sharing Tideline, our proprietary memory with a built-in database switch. They plug into the tools you already run and launch in one click with nothing to host, behind five layers of security.
Windows installerDirect
Signed installer. Double-click and you're set.
Brand-new machine? One line and you’re in.
This installs Node for you if it’s missing, then sets up Brigade. Nothing else to do.
Detects Node, installs a pinned version if it’s missing or too old, then installs Brigade.

What is Brigade
An AI crew that grows with you
Brigade sets up in minutes, works around the clock, learns your preferences, and gets better every day. It is open source and runs on your own machine.
A memory that grows
Tideline links every fact into a knowledge graph you can open in Obsidian, so your crew remembers what matters and recalls it when it counts.
One-click install
A single command and your crew is live in under a minute. Nothing to host and no servers to wire up.
On duty 24/7
Scheduled tasks and always-on chats keep your crew working while you sleep.
On every device you own
Talk to your crew from your phone, watch, and desktop, all the way out to smart glasses and a Meta Quest.
Five layers of security
Sealed credentials, sandboxed tools, and approvals before anything risky. Your data stays on your machine and you set the limits.
Headroom, built in
47kBrigade ships with Headroom, the context-compression layer, so your crew runs on up to 95% fewer tokens with the same answers. Faster replies and smaller bills.
See Headroom on GitHub
Your whole crew, in one chat.
Text Brigade like a colleague. It hands each task to the right specialist and checks with you before anything touches your accounts.

Layers of
security and scalability
Five layers of security and three built for scale. You always know what each agent can touch and where your data lives, however big your crew gets.
Brigade keeps its state in your encrypted Convex backend instead of scattered local files. Credentials are sealed before they leave your machine, so the agent never sees your real keys. Wipe your device and your whole crew restores itself, with nothing sensitive left behind on disk.
👀 I am all around you.
Tideline
our best-in-class memory, built for scale.
Not a notes file replayed into the prompt. Tideline is a real long-term memory that keeps what matters and lets the rest fade. It builds a memory graph, self-improves under your sign-off, runs loop engineering that knows when to stop, and catches AI slop before it ships, across any model: Claude, GPT, Gemini, Llama, GLM, DeepSeek, Qwen, MiniMax, and more.
Long-term recall. 100% on the hard cases where the usual approach hits 43%.
Self-improving. Confirms patterns and proposes changes for your approval.
Loop engineering. Runs autonomously toward a goal, but knows when to stop.
AI-slop detection. Rewrites generic, low-effort output before it ever ships.
Carrow handoff. Move a live task between models with zero context loss.
A real memory graph. Typed links between facts, walked for connected context.
Obsidian-style vault. Plain markdown notes that open in Obsidian; your edits survive.
Poison-resistant. Untrusted pages and tools cannot rewrite what it believes.
The models are interchangeable. The intelligence, memory, and judgment around them are Tideline.
I am always self improvingI control real
devices.
Brigade actually uses a computer and a phone, clicking, typing and tapping like a person. Run it on your devices, or give it its own. It even gets its own email and phone number.
Not just an assistant. It actually has hands on a real computer and phone.


Just say, join my call.
Message Brigade in any chat: WhatsApp, Slack, iMessage. Say 'jump on my 2pm' and it shows up to the meeting for you.
Brigade · AIWhy Brigade?
| Where Brigade pulls ahead | |||||
| Memory | Tideline, enterprise-grade managed memory | Plugin memory, opt-in | SQLite + markdown, self-managed | CLAUDE.md + auto-memory | Plugin memory, opt-in |
| Self-improving agents | Agents learn and improve, you approve | No | Auto skill-learning, unreviewed | No | No |
| Security | Five layers, built-in | Built-in primitives, self-configured | Self-managed | Sandbox + permissions, no managed layers | Built-in primitives, self-configured |
| Database & encryption | Switchable, encrypted (Convex) | DIY, no encryption at rest | SQLite, encryption on you | Proprietary, closed | DIY, no encryption at rest |
| Goals & org chart | Goal-based agents on a real org chart | No | No | No | No |
| Native channels | 20+ chat apps, plus phone, watch, Meta glasses & Quest | 20+ chat apps, software only | 20+ chat apps, software only | CLI, desktop, web, IDEs, mobile | 20+ chat apps, software only |
| Live design whiteboard | Agents debate and draw the design live (coming soon) | No | No | No | No |
| Headroom (token saver) | Fewer tokens, better results (enterprise) | No | No | No | No |
Questions, answered.
What is Brigade?
- Brigade is a self-hostable crew of AI agents arranged on a real org chart. They delegate work, hand tasks off between models with no context loss, remember across sessions through a shared memory called Tideline, and act inside the tools and channels you already use.
Is Brigade free and open source?
- Yes. Brigade is MIT licensed and free to self-host today, with a managed cloud service coming soon. You only pay for the AI model usage you run through it, or nothing at all if you use a local model.
Can I self-host Brigade and run it locally?
- Yes. Brigade is CLI-first and runs on your own machine or server, so your data and credentials stay with you. It works with local models through Ollama and any OpenAI-compatible endpoint, as well as hosted providers.
How is Brigade different from Claude Code or OpenClaw?
- Brigade runs agents on a real org chart with goal-based routing, ships Tideline managed long-term memory, plugs into 1,000+ apps via Composio, and lets agents self-improve under your sign-off behind five layers of security. Most alternatives leave memory, integrations, and security for you to wire up.
What is Tideline?
- Tideline is Brigade's proprietary long-term memory, built in-house. It is a real memory graph with typed links between facts, not a notes file replayed into the prompt. It keeps what matters, lets the rest fade, self-improves under your approval, and resists poisoning from untrusted pages.
What apps and models does Brigade work with?
- Brigade connects to 1,000+ apps through Composio, including Gmail, Slack, GitHub, Notion, Jira, and Linear, plus 20+ chat channels. It runs across any model, including Claude, GPT, Gemini, Llama, GLM, DeepSeek, Qwen, and MiniMax.
How do I install Brigade?
- Install it from npm with npm i @spinabot/brigade (Node 22.12 or newer), or use the one-click self-host flow. There is nothing to host for the managed service when it launches; today you can run the whole crew on your own machine for free.









