An AI that knows what your business knows.
The answer is somewhere. A doc in Drive, an email from March, a meeting nobody wrote up, or the head of the one person who is on holiday. A company brain connects the places your knowledge lives, so anyone on the team can ask a plain question and get the answer with its source in seconds, instead of going on a scavenger hunt.
What is a company brain?
A company brain is an AI assistant connected to a business’s own information: documents, email, calendars, CRM, project tools and meeting transcripts. Staff ask it questions in plain language, and it answers from that information, says where each answer came from, and says so when it does not know. On top of the same knowledge it can do routine work, like a morning briefing, a project recap, or filing every meeting transcript in the right folder.
How it works
- Start from the questions. We list the ones your team asks every week. Where is the latest version of this? What did we agree with that client? How do we do this process?
- Connect only where those answers live. Google Drive, Gmail, Calendar, SharePoint, Slack, Notion, Asana, HubSpot and the like. Read-only wherever possible.
- Keep permissions intact. People only get answers from information they could already see. The brain does not become a back door.
- Ask it where you already work. In Slack or Teams, on a simple internal web page, or by email.
- Add routines once the knowledge is there. A daily briefing, meeting notes filed and summarised, project recaps, a research brief on every new lead. Built on Claude, so each one is a short instruction rather than a new project.
There is a small one on this site
Baymax, the chat assistant on the homepage, is a public version of the same idea. It answers only from the documents I gave it, and when the answer is not in them it says so instead of making one up. A company brain is that turned inward, connected to your real tools and kept current.
The math
Ask how long people spend looking for information, or waiting for someone else to answer. If 20 people each lose two hours a week, that is 40 hours of payroll every week spent on a scavenger hunt. New hires are the other half of it: someone who can ask the brain instead of interrupting a colleague gets up to speed faster, and so does the colleague.
Where I draw the line
It answers from your information and shows the source, so anyone can check it in a click. When it does not know, it says so. It never gets access to anything a person in that role could not already see, and it runs in accounts you own.
The technology is the easy part. The harder part is the habit: a team takes a few weeks to reach for it first instead of asking around. So I build the habit in, with a place people already go and something useful from it every morning.
Who it is for
Teams of roughly 5 to 100 people whose knowledge lives across many tools and a few heads: agencies, consultancies, professional services firms, property managers, multi-site trades and clinics.
It is a capacity system. It gives your team time back; it does not bring in customers. If your real problem is not enough work, start with something that brings people in, like client intake automation.
Common questions
Is our data used to train the AI?
No. It is built on Anthropic’s commercial Claude API, which does not train its models on customer data by default, and everything runs in accounts you own.
What if it gives a wrong answer?
Every answer comes with its source, so it can be checked in a click, and it is built to say it does not know rather than guess. If a source document is out of date, the answer will be too, which is why keeping a few key documents current matters more than connecting everything.
Which tools can it connect to?
Most of the common ones: Google Drive, Gmail and Calendar, Microsoft 365 and SharePoint, Slack, Notion, Asana, HubSpot, and meeting recorders. We connect only the ones where your team’s answers actually live.
Can it do things, not just answer questions?
Yes, within limits you set. The common ones are a morning briefing, meeting transcripts filed and summarised, project recaps, and a research brief on each new lead. Anything that changes records or contacts a customer stays behind an approval.
How much does it cost?
One fixed price, agreed before I start and paid when it is built and running. No retainer. The scope depends mostly on how many tools it connects to. You cover the running costs, like AI usage, and you see those before we start.