01Listen
It reads the channels it's invited to and picks out the people, projects, decisions and customers in them.
EARLY ACCESS · IN DESIGN
Your company, remembered.
Living Brain is a teammate that turns your team's conversations into a Markdown wiki and keeps it true. Use it in Slack, in your coding agent, in your terminal. Every fact links back to the message it came from.
You're on the list. We'll write once, when there's something to try.
See how it thinks
No spam. One email when it's ready.
Use it from anywhere
- SlackfirstInvite it to channels. It answers in the thread.
- Any MCP clientplannedClaude Code, Codex, Cursor, OpenCode, Claude Desktop and others.
- The
livingbrainCLIplannedOne small Rust binary for your terminal and scripts. - App (PWA)plannedInstalls on phone and desktop from the browser. Reads offline.
Built in Rust, and fast. The CLI is a single static binary that starts instantly. The same core runs the hosted service on Cloudflare's edge.
$ livingbrain ask "who owns billing?"
Lina Haas owns Billing v2 since Mar 12. [1]
[1] #product · Tomás Reyes · Mar 12 · 11:20
$ livingbrain mcp
MCP server on stdio · reads with your own access
01 · Speaks up
It answers before anyone has to dig.
Living Brain listens in the channels you invite it to. When a question already has an answer, it replies in the thread and links to the exact message the answer came from.
Rolling back #412. The 502s on /api/routes come from the new pool config.
12 more messages
is prod down? Northwind says routes won't load
Not down. routes-api returned 502s from 09:38 to 09:52 after deploy #412. Dev rolled it back at 09:51 and error rates have been normal since.
↩ #eng · Dev Patel · 09:51
[[Route Optimizer]]
02 · How it works
Listen. Write. Evolve.
02Write
Conversations become Markdown pages with [[backlinks]]. Every fact cites the message it came from.
03Evolve
On its own schedule it merges duplicates, flags contradictions, retires stale facts and rewrites summaries.
A wiki that keeps itself true.
Every night the brain revisits what it knows. It also learns how each person works, so its answers improve every week.
BILLING-V2.MD · OWNER
Dev Patel · since Jan 08
Lina Haas · since Mar 12 [^1]
[1] #product · Tomás Reyes · Mar 12 · 11:20
03 · See how it thinks
Explore a sample brain.
This is Kestrel Freight, a fictional 20-person logistics company. Drag to orbit, scroll or pinch to zoom, and click any node to open its page.
Illustration · Sample data
Static view · click a node or use the list below
Jump to a page
04 · What it does
A teammate, not a search box.
Remembers
Turns conversations, and the mail you forward to its Owlpost inbox or connect from Gmail, Outlook or IMAP, into pages for people, projects, decisions and customers.
Answers with sources
Every answer links to the messages it came from, so you can check it in one click.
Acts in your tools
Opens the issue, reads the PR, sends the email digest, and hands bigger jobs to a sandboxed agent that returns a pull request.
Speaks up
Replies when a question in a channel already has an answer, without waiting to be mentioned.
Works while you sleep
Merges duplicates, resolves contradictions and retires stale facts on its own schedule.
Learns your way
Notices how each person likes answers, short or detailed, in threads or DMs, and adapts week by week.
Connects to your coding agents
Gives Claude Code, Codex, Cursor and other MCP agents the team's memory before they write code, and gathers all their session logs in one searchable place. How it works
Easy to use planned app
The web app installs on your phone and desktop, keeps pages readable offline, opens on one box to search or ask, and keeps the 3D brain one tap away.
One night in the brain.
While the team is offline it reads what happened, tidies what it knows and hands work to agents. By morning there's a digest waiting.
Sample data- 22:40Read 214 new messages
across 9 channels
- 23:15Merged 4 duplicate pages
billing.md folded into billing-v2.md, and 3 more
- 01:30Resolved 3 contradictions
Billing v2 ship date, PG16 window, Bluepeak contact
- 03:05An agent opened PR #218
"Set routes-api pool size back to 20", waiting for Dev's review
- 07:00Wrote the morning digest
emailed to 12 people, each in the format they prefer
05 · Coding agents
Every agent starts with the company's memory.
Connect Claude Code, Codex, Cursor, OpenCode or any agent that supports MCP. Before an agent writes code, it asks the brain for team conventions, past decisions, who owns what, and what was tried and failed.
$claude mcp add --transport http livingbrain https://mcp.livingbrain.wiki
Added HTTP MCP server livingbrain
$claude
✓ livingbrain connected · 14 tools
you> add a migration for driver shift tables
→ livingbrain.ask("what's our convention for migrations?")
Forward-only, one migration per PR, named YYYYMMDD_slug.sql. Never edit a merged migration. Large tables get a backfill job, not an UPDATE. [1][2]
[1] #eng · Maya Okafor · Jan 30 [2] brain/decisions/forward-only-migrations.md
agent> Following the team convention. Creating migrations/20260314_driver_shifts.sql, with the backfill as a separate job.
Agents → brain opt-in
The prompts your team runs and the decisions agents make along the way ("we switched to Postgres because…", "this module is fragile") become wiki pages that link to the session they came from. Secrets are redacted before anything is stored.
Brain → agents
Before an agent writes code, it asks for context: conventions, past decisions, owners, and what was tried and failed. Every agent on the team starts with the company's memory instead of a blank slate.
Every agent's logs, in one place.
Planned · Sample dataThe livingbrain CLI collects the session logs your coding agents already write: Claude Code, Codex, Cursor, OpenCode and Colonizer colonies. It redacts secrets on your machine and sends them to the brain only if you opt in. Search every session, see tokens and cost per person, repo and model, and spot where agents keep failing. Decisions found in the logs become wiki pages with a link back to the session.
| Agent | Sessions | Tokens | Cost | Ended in a merged PR |
|---|---|---|---|---|
| Claude Code | 214 | 38.2M | $61 | 64% |
| Codex | 97 | 14.9M | $22 | 58% |
| Colonizer colonies | 31 | 9.4M | $14 | 71% |
| Cursor | 122 | 6.1M | $9 | 49% |
Team prompt library
Sample dataThe team's best prompts are saved, versioned and suggested to everyone. The brain notices which ones work.
Review a PR against our conventions
Write a migration the Kestrel way
Explain a stack trace using recent incidents
Draft a customer reply from this thread
It learns from every session.
The learning layer reasons over each person's prompts and agent sessions, not just chat. It learns how each developer works, which prompts lead to merged PRs and which lead to rework, and where agents keep getting stuck. Then it improves the prompt library, fills gaps in the wiki, and briefs each person's agent the way they like.
Maya Okafor
profile · one sample month
- W1Prefers small PRs, under 300 lines
- W2Writes tests first
- W3Gets stuck on auth middleware → auth notes now in her agent's brief
- W4Wants answers as diffs, not prose
Prompts ranked by merged PRs
Sample data- Review a PR against our conventions38
- Write a migration the Kestrel way31
- Fix a flaky test using incident history22
- Refactor this file · often reworked6
Fewer tokens. Less rework. Faster agents.
Planned · benchmark comingMost agents start every task cold: they reread threads, grep the repo and rediscover what the team already decided. Living Brain does that work once, at night, and hands every agent the result.
Fewer tokens
An agent asks for a short brief with citations instead of reading thirty threads and grepping the whole repo. The context is written once and reused by every agent on the team.
Less rework
Decisions and dead ends are on record, so agents stop retrying approaches the team has already ruled out.
Cheaper models do more
The hard reading happens when the wiki is written. Day to day, a small model with the right page can do work that a large model with no context gets wrong.
Fast
One Rust binary, a local cache and Cloudflare's edge. Search answers from your machine before the network does.
Without the brain
Illustration- Read the Slack threads about billing
- Grep the repo for every place that touches invoices
- Ask a teammate who owns it
- Try the approach the team rejected last month
With the brain
Illustration- brain_context_for("billing", task)
- One page: the owner, the conventions, the decision, the dead end, each with its source
- Start coding
The benchmark, in the open
We'll measure tokens, cost, time, success and rework, with and without the brain, against a plain retrieval baseline, across Claude Code, Codex and Colonizer colonies, on large and small models. The task set, the harness and the raw results will be public, and anyone can rerun them. Until then there are no numbers on this page. Follow the benchmark
06 · Works with Colonizer
From Slack to pull request.
When a job is bigger than an answer, the brain launches a colony on Colonizer, a sister Factory Zero venture: an isolated microVM with a coding agent inside that comes back with a pull request.
-
01 · ASK
Maya Okafor #routes
@livingbrain fix #142
#142 · ETA off by an hour after DST
-
02 · BRIEF
- repo: kestrel/routes
- owner: Maya Okafor
- tz logic: src/eta/tz.rs
- tried: offset patch, reverted
-
03 · COLONY c-7f3
microVM
$ opening pull request #219
- 04 · PULL REQUEST #219 Fix ETA drift across DST +42 −9 · 213 tests pass Posted back in the Slack thread
-
05 · LEARNINGS
- +DST handling notes
- +tz.rs owner: Maya
- +Dead end: offset patch
Boots already briefed
Each colony starts with the brain's context for that repo and task.
Reports back to the wiki
When it finishes, its decisions, dead ends and the PR flow back into the brain.
Followed from the thread
Start a colony from Slack and watch it work live in the same thread.
07 · Private by design
It only knows what you're allowed to know.
Living Brain reads with the asker's own access. Where you ask decides which memory it can use.
| Asked in | Shared brain | Private channel memory | Personal memory |
|---|---|---|---|
| Public channel | Yes | No | No |
| Private channel | Yes | That channel | No |
| Your DMs | Yes | Channels you're in | Yes |
Always limited to what the person asking can already see in Slack.
Planned: each scope is encrypted with its own key. Search stays inside your scopes and stays fast, and deleting a scope's key erases that memory everywhere.
Planned: connected mailboxes (Gmail, Outlook, IMAP) stay in the owner's personal memory, encrypted with their key, unless they share a specific label or folder.
08 · You own it
Your brain is a folder of Markdown.
Read it, edit it, version it, move it. Your knowledge isn't locked in a black-box vector store. Planned: keep it in your own git repo, where the Living Brain GitHub App proposes every change as a pull request and edits you merge flow back in.
---
type: project
owner: [[Lina Haas]]
ships: 2026-04-30
updated: 2026-03-13 by living-brain
---
# Billing v2
Usage-based invoices for fleet customers.
Owned by [[Lina Haas]] since Mar 12. [^1]
## Waiting on it
- [[Alder & Finch]]
- [[Northwind Grocers]]
[^1]: #product · Tomás Reyes · 2026-03-12 11:20
Thirty days of a 20-person team
Sample data- Pages written
- 412
- Links formed
- 2,960
- Facts refreshed
- 738
Each line is scaled to its own total. The graph above fills in as the chart plays.
09 · Pricing
Planned pricingPriced per workspace, not per seat.
Self-hosting is free for small teams. Teams and hosting are paid. Living Brain is built in Rust on Cloudflare's edge, so it's fast and very cheap to run, and the price reflects that.
Community
Self-hosted$0per workspace
- Up to 5 people
- Runs on your own Cloudflare account with your own model key
- All core features: the Slack brain, the living wiki, citations, the 3D graph, MCP for coding agents, agent log aggregation, export
- Storage: your own Cloudflare account, no limit from us
Teams
Self-hosted or hosted$5/mo per workspace
- Unlimited people
- Team features: SSO, admin and audit log, the shared team prompt library, per-channel policies
- Self-hosted with a license key, or hosted by us with your own model key
- Hosted: 10 GB storage included. Self-hosted: your own storage
Crew
Hosted$9/mo per workspace
- Hosted by us, DeepSeek included, or bring your own model
- Up to 25 people
- Team features included
- 25 GB storage included
Storage covers wiki pages, sources, search indexes and aggregated agent logs. Extra hosted storage: $0.25 per GB a month. Every plan: Markdown export, cited answers, and access that follows Slack permissions. No hosted plan is free.
20 people, one month
Illustrative comparison10 · FAQ
Questions
Can it collect my coding agents' logs?
Planned. The livingbrain CLI collects the session logs that Claude Code, Codex, Cursor, OpenCode and Colonizer already write, redacts secrets on your machine, and uploads them only if you opt in. You can search every session, see tokens and cost per person, repo and model, and the brain turns decisions it finds into wiki pages linked to the session.
How much storage is included?
Hosted Teams includes 10 GB and Crew includes 25 GB, covering wiki pages, sources, search indexes and aggregated agent logs. Extra hosted storage is $0.25 per GB a month. Self-hosted plans use your own Cloudflare storage, with no limit from us. Planned pricing.
Can it connect my Gmail or Outlook?
Planned. The simplest and safest way needs no mailbox access at all: forward, BCC or auto-forward mail to your brain's own address on Owlpost, our sister email service, and only what you send arrives. If you want a whole mailbox kept in sync, you can also connect Gmail, Outlook.com or Microsoft 365, or any IMAP mailbox, plus your calendar, read-only. Your mail stays in your personal memory, encrypted with your own key, and no one else can draw on it unless you share a specific label or folder. Every message is screened for spoofing and prompt injection before the brain reads it, and mail can never trigger actions. Revoke the connection at any time and its stored mail is erased. When you self-host, you use your own Google or Microsoft app, so mail never touches our servers.
Is my data encrypted?
Planned. Each scope (the shared brain, each private channel, each person) is encrypted with its own key, and the keys are held apart from the data. Search runs only inside the scopes you can see, so it stays fast. Deleting a scope's key erases that memory everywhere, backups included. On Teams you can bring your own key. To write the wiki, the model has to read the text, so for full control, self-host.
Is my data used for training?
No. Your workspace's messages and pages are never used to train models, ours or anyone else's. If you bring your own model, requests go to your provider under your account's terms.
Does it only work in Slack?
No. Slack is the first chat integration. You can also use it from any MCP client, such as Claude Code, Codex, Cursor, OpenCode or Claude Desktop, or from the livingbrain command-line tool, a single Rust binary. MCP and the CLI are planned, like the rest of the product.
Is there an app?
Planned. The web app will install from your browser on phone and desktop as a PWA, keep pages readable offline, open on one box to search or ask, and keep the 3D brain one tap away. Like the rest of Living Brain, it is not built yet.
Can I self-host?
Yes. The Community plan is free for up to 5 people: it runs on your own Cloudflare account with your own model key. Larger teams need a Teams license. The plan and the code as it lands are public on GitHub, open core: Apache-2.0, with the Teams features under a commercial license.
Which models?
On every plan you choose your own models, per role (triage, answers, research): Anthropic, OpenAI, OpenRouter, your LiteLLM gateway, or any OpenAI-compatible endpoint, including self-hosted ones. Crew also includes DeepSeek, so you can start without a key of your own.
What does "living" mean?
Most wikis start going stale the day they're written. Living Brain revisits its own pages on a schedule: it merges duplicates, flags contradictions, retires facts that are no longer true and rewrites summaries. It also learns how each person works, so its answers improve every week.
How is it different from Slack AI or Glean?
Search tools find what already exists. Living Brain writes and maintains a wiki you own as Markdown, cites the message behind every fact, acts in your tools, gives your coding agents the same memory, and is priced per workspace instead of per seat.
Which coding agents can I connect?
Any agent that supports MCP, including Claude Code, Codex, Cursor and OpenCode. We're not affiliated with these tools; they connect through the open protocol.
Do you store my agent prompts and code?
Only if you opt in. Prompts and session notes are stored with secrets redacted first. Source code is never stored unless you allow it.