When Sarah leaves, six months of training leaves with her.
Every customer she handled, every refund she approved, every escalation rule she remembered — it's gone. Her replacement asks the same 50 questions for three months.
Company Brain · the missing layer between your data and AI that actually worksEarly access · guided setup
Your AI doesn't know how YOUR business works. We'll get it there with you.
The biggest blocker to AI automation isn't the models anymore — they got too good too fast. The real blocker is the knowledge inside your business. Company Brain reads your Slack, email, tickets, and handbooks — and turns them into rules the AI employees on your team can follow. New messages land in seconds. Old knowledge cleans itself up. Every answer shows you where it came from. And when AI does something real (issues a refund, sends an email), your team signs off first.
We're shipping in public. The brain is real and runs production work for early customers today — and we onboard every workspace with a hands-on call to make sure your first week ends with three rules live, citations clicking, and your team using it. If the fit isn't right we say so on the call.
Included on every paid plan · from $29/mo · 200 credits free · cancel any time · your data stays yours
Already signed up? Jump into your workspace and click Load demo brain on the empty state to see three example docs, three rules, and a chat — in ten seconds.
If any of this hits
You're running on tribal knowledge. The AI you bought can't see it.
Every successful business has critical know-how scattered everywhere — in Slack threads, in old emails, in support tickets, in people's heads. It works because humans vaguely remember where it lives. AI agents can't operate like that. So they don't.
Your AI doesn't know your refund policy.
It knows the entire internet. Not your business. Ask ChatGPT what to do about a $4,000 refund and you'll get generic customer-service advice. Useless.
Knowledge lives in Slack DMs and people's heads.
The escalation path is in a thread from 2024. The pricing exception lives in a manager's memory. Your handbook hasn't been updated since you hired.
You can't tell what your team's AI is actually doing.
Some staff use ChatGPT. Some don't. There's no record of what got shipped, what got asked, what got copy-pasted into customer-facing emails.
How it works
From scattered knowledge to AI that knows your business — in 4 steps.
01
Connect what you already use
Slack, Gmail, Outlook, Teams, Zendesk, Notion — three clicks each. New messages land in your brain within seconds, no waiting for a scheduled import. Or paste your handbook / upload a PDF or Word doc up to 10 MB — we read the whole file in the time it takes you to grab coffee.
02
We read it and write your rules — your way
Pick the shape that fits. A plain rule ("refunds under $250 auto-approve, $250–$1k get manager sign-off") or a step-by-step playbook ("when X happens → Sarah does Y → if it's over $1k, escalate"). The brain drafts both from your actual data. You review and turn on with one click.
03
Every AI chat now knows your business
Pick any of 25+ AI employees. Every chat reaches into your brain, finds the rules and history that matter, and answers using YOUR playbook. Every fact the AI quotes is tappable — click to see the actual Slack thread, ticket, or handbook page that informed it. No black box.
04
Big calls get a second opinion — and a sign-off
Refund over $1,000? Lawsuit threat? Hiring decision? A second AI re-reads the response and flags what's missing. And when AI does something real for you (issues a refund, sends an email, adjusts pricing), an admin signs off first — never after the fact.
What's included
You get the brain. Not a chatbot.
One subscription. Six chat-tool connectors + dual-mode rule induction (flat policy or step-by-step playbook) + auto-currency that keeps the brain fresh + click-through citations for full lineage + per-seat usage tracking + tailored AI per role with rule scoping + challenger on high-stakes turns + approval queue for autonomous AI + clones of your real teammates + public API + audit log + weekly digest. We'll be honest about the seams in the "where it is today" section below — what's live, what's in flight, and how we cover the gap with a setup call.
The brain itself
Six apps connect with three clicks each: Slack, Gmail, Outlook, Microsoft Teams, Zendesk, Notion. Plus your handbook — paste it in or upload a PDF or Word doc up to 10 MB. New messages land in your brain within seconds, not 15 minutes. When a source re-syncs, existing messages get refreshed in place — no duplicates piling up.
Two ways to write a rule
A plain rule when you want "what's our policy" (refunds, pricing, compliance). A step-by-step playbook when the work has an order (incident response, onboarding, escalation). Pick the shape that fits — the brain writes either one from your data.
A brain that stays current — by itself
Stale info ages out after 30 days unless we see it again. Every active rule gets re-checked every day against your latest data. If your refund threshold changed and the rule didn't, it auto-pauses and the owner gets an email. Your brain self-corrects without you watching.
Every answer shows its work
When the AI quotes a fact, the source is right there next to it — tap to see the actual Slack message, support ticket, or handbook paragraph it came from, with timestamp. No black box. If the answer's wrong, you can see why in two clicks.
See exactly what AI is doing per teammate
One credit pool, every teammate spends from it. The owner dashboard shows jobs run, credits used, hours saved per person — plus their top AI employees and last activity. The "who's secretly using ChatGPT" question answers itself.
Different AI for different roles
Each teammate has a title — Bookkeeper, Sales Rep, Founder. The same AI employee leans differently for each. And rules can target specific roles — the refund policy applies to your Sales Reps but not your Bookkeepers. One workspace, the right answer for each person.
A second-opinion AI on the big calls
When a chat mentions refunds, lawsuits, hires, contracts, or big dollar amounts, a second AI auto-reviews the answer and flags what's missing. You see both side by side. No one ships a one-sided AI take to a decision that matters.
Sign-off before AI takes real action
If you give an AI agent the keys to issue refunds, send emails, or change prices, an admin approves in a queue before the action fires. Each request is risk-scored and saved to the activity log. AI does the work; humans keep the authority.
AI versions of your real teammates
Pick a teammate. The brain reads their messages, tickets, and decision patterns and builds an AI version that answers the way they would. When Sarah goes on leave (or leaves for good), the AI version stays.
Knowledge graph + ask-the-brain
A live, drillable map of every source, teammate, rule, and fact your brain knows — color-coded by topic. Plus a search box that takes plain English: "What do we know about refunds over $1k?" returns a cited answer pulled from across your team.
Your own subdomain + one log-in
Your team signs in at acme.automatebusinessnow.com. One log-in carries across the main site and every subdomain — no second password to remember. We handle the certificates and DNS; you pick the name when you sign up.
Sign-off log, developer access, weekly digest
Every AI action that needs sign-off lands in a queue with the actor, the request, and the outcome — full trail. Developers can read your brain from your other tools through a simple API. And every Monday the owner gets an email: what got used, what got built, what got flagged, how much was spent.
The thing other tools don't do
A brain that stays current — and corrects itself.
Every other "AI over your data" tool is a snapshot. They read your stuff once and slowly go out of date. We built the opposite.
Auto cleanup at 30 days
Old knowledge ages out
Anything we learned more than 30 days ago either gets re-confirmed or quietly removed. The AI never quotes a Slack thread from 18 months ago as if it happened yesterday.
Daily rule check
Stale rules pause themselves
Every active rule gets re-checked every day against your latest data. If reality has moved (you changed your refund threshold and forgot to update the rule), it auto-pauses and the owner gets an email. Nothing goes silently out of date.
Re-syncs are clean
Doesn't bloat over time
When a connected source re-reads — Slack channel, Gmail thread, Notion page — existing messages get refreshed in place, not re-stored. The brain only grows when there's something genuinely new to learn.
The bigger picture
The new primitive every business is about to need.
The biggest blocker to AI automation of companies is no longer the models. They got so good so quickly. The real blocker now is the domain knowledge inside companies.
Every business has critical know-how scattered everywhere. Some of it lives in people's heads. Some of it's buried in old email threads, in Slack accounts, in support tickets, in databases. The company works because humans vaguely remember where that knowledge lives and how to apply it. But AI agents can't operate like that.
If every company in the world is going to run on AI automation, we need a new primitive — a company brain. A system that pulls knowledge out of those fragmented sources, structures it, keeps it current, and turns it into an executable skills file every AI on the team can read.
This isn't a company-wide search. It isn't a chatbot over your documents. It's a living map of how the company actually works — how refunds get handled, how pricing exceptions get decided, how engineers respond when something breaks. AI systems use that map to do the work safely and consistently.
The company brain is the missing layer between your raw data and AI automation that actually works. Until you have it, you're running on a smart-sounding stranger guessing at your business. You shouldn't be guessing. Neither should your AI.
Why not just use…
Here's what you'd compare us against. And why none of them solve this.
ChatGPT / Gemini / Claude
Generic. Smart but doesn't know your refund threshold, your escalation path, or how Sarah handles a pissed customer.
Notion AI / Glean / RAG chatbots
Search box on top of your docs. Tells you where the answer is. Doesn't decide. Doesn't enforce. Doesn't catch a stale policy.
Hire a consultant ($80k+)
They write a doc you'll read twice. Their advice doesn't show up at chat-time when your sales rep is on a call.
Build it in-house (12-18 months)
By the time you ship it, your competitor was live here in 5 minutes.
No risk
If it doesn't work, you owe us nothing.
Cancel any time
One click in your dashboard. No phone call, no retention script. Top-up credits never expire — keep using them.
Your data stays yours
Skills, facts, audit log — all exportable. We never train models on your data. Standard zero-retention contracts with our model providers.
200 credits free to start
No card required to sign up. Connect a source, build a rule, run a chat — see the brain work before you commit a dollar.
Where Company Brain is today
Shipping in public. Here's what's live and what's next.
We'd rather tell you exactly what works and where the seams still show than oversell. This is the same status board our team works against — kept current, not aspirational.
Live and used in production
- Multi-tenant workspace with shared credit pool + audit log
- SOP upload (PDF / Word / CSV / TXT / MD up to 10 MB) + paste-text ingestion
- Slack, Gmail, Outlook, Microsoft Teams, Zendesk, Notion connectors
- Plain-policy + step-by-step playbook rule induction with citations
- Streaming chat with rule + fact retrieval and click-through citations
- Daily rule re-verification + 30-day fact freshness
- Challenger pass on high-stakes turns (refund, legal, hire, large $)
- Approval queue for autonomous AI tool calls
- Stripe billing on every paid plan ($29 → $599/mo) — credits scale with how much your team uses
- Public read API + bearer-token developer access
- Weekly digest email (pool spend, top employees, new skills, challenges)
In flight · hands-on with our team in the meantime
- First-run setup checklist (workspace → sources → rule → chat → invite → billing)
- Source-onboarding redesign (gating new sources behind OAuth completion)
- "This answer is wrong" / "Teach this as a rule" correction loops in chat
- Brain-health dashboard (sources indexed · rules active · unresolved gaps)
- Owner-only vs admin-only vs member permission boundaries
- Per-company answer-quality test prompts (regression on rule activation)
- Trust UX (what the AI knows, who can see what, data retention controls)
- Custom subdomain wildcard certs for production tenants
Until these are self-serve, the setup call is where we make sure your workspace ends day one with rules live, citations clicking, and your team using it.
FAQ
Questions. Answered in plain English.
How long does setup actually take?+
5 minutes if you paste a handbook. 15 minutes if you connect Slack or Gmail. The brain ingests instantly when you paste; OAuth flows for chat tools take a couple minutes. Either way, you have your first AI rule running before your coffee gets cold.
What if my team's AI starts giving bad advice?+
Three lines of defense. First, every cited fact in chat is clickable — the user (or you on review) can verify the source in one tap. Second, the daily drift check auto-pauses any active rule that no longer matches reality and emails the owner. Third, when AI agents take destructive actions, admins approve in the queue before the action fires — never after. The brain self-corrects without you watching.
How do I know the AI is actually using our brain instead of making things up?+
Every chat turn surfaces which active rules and which facts informed the answer — visible inline, color-coded by source. Click any citation chip to see the original Slack message, ticket comment, or handbook paragraph with timestamp. If a rule applied, you'll see it. If no relevant rule existed, the AI says so rather than improvising.
Do you train on our data?+
No. We never train models on your data, and the AI providers we route through don't either — they keep nothing after the response is sent. Your rules, knowledge, and decisions stay private to your workspace.
What if someone leaves?+
Their AI version + every rule they helped build stays. If you cloned them as an AI version, that AI keeps answering in their voice using their decision patterns. The vision is: "oh the person with most context left two weeks ago" stops being a problem.
Can I export everything?+
Yes. Skills + facts + the audit log are exportable. Your data is yours. Cancel any time and take it with you.
Why is Company Brain included on every paid plan?+
Because seats are free and credits are how we earn. Charging extra for a brain that every AI employee shares is the wrong tax — it gates the thing that makes the AI work. So every paid plan (Starter $29 through Power $599) ships with the full Company Brain, every connector, every rule type, every safeguard. The only thing that changes between tiers is how many credits you get per month. Unlimited teammates on every paid plan, no per-seat fees, no surprise charges.
Pricing
Included on every paid plan. From $29/mo.
Company Brain is a feature, not a tier. Starter, Pro, Scale, and Power all ship with the full brain, every connector, every rule type, and unlimited teammates. The only thing that changes is how many credits your team has to spend per month. Top up any dollar amount when you need more — top-ups never expire. Cancel any time. No surprise charges.
200 credits free on signup · no card required · 5 minutes to your first AI rule