1
person approves an action before it reaches a client or a record
Company brain
Growth stops at the admin, not at the demand.
The short version
Firms with 5 to 50 staff are rarely short of work. They are short of the hours it takes to chase it. Submissions go quiet. Clients go dormant. Replies take days. The work that keeps being dropped is admin work, and the usual answer to it is to hire another coordinator.
The company brain is the other answer. In week one we connect your systems, map how the firm actually runs, and write down where the work is being dropped today. After that the brain reads, drafts the next action, and waits for a person on your team to approve it. Then the proof puts today's numbers next to the week one baseline.
Week one: connect, map, baseline
Install is a week of work with your team, not a login and a welcome email. Three things happen in it.
Connect
We link the systems the work already lives in: your ATS or CRM, email, phone, WhatsApp, the spreadsheets people actually use, and accounting. Connecting more than five systems adds a day and $500 to the install, and we say so during scoping rather than after.
Map
We write down how the firm runs. Who owns what. What a good submission looks like here as opposed to somewhere else. Which clients matter and why. What gets escalated and to whom. What you confirm in that conversation is what the brain then holds as canonical.
Baseline
We count the dropped work before we change any of it. If there is no number for week one, then “it feels better now” in month three has nothing to sit against, and neither of us can tell whether you should keep paying us.
| What a baseline counts | Typical at install |
|---|---|
| Submissions with no follow-up in five days | 34% |
| Client relationships that have gone dormant | 22 |
| Average time to respond | 3.2 days |
| Admin hours a week | around 31 |
Those are the figures a baseline typically turns up. They are an estimate of what we expect to find, not a promise about your firm. Your numbers land where they land, and they go in writing in week one either way.
What the brain remembers
The brain holds what it learns in three states, and it tells you which state a thing is in rather than presenting all of it with the same confidence.
- Working. It has understood something from your systems, and that reading is not confirmed with you yet.
- Canonical. You have confirmed it. Your process, your clients, your history, and the decisions behind them.
- Unresolved. Your systems disagree with each other, so it is held as unresolved rather than settled quietly.
Unresolved is the state owners react to. It comes out as a plain list: here are the things your systems disagree about. The two halves of a contradiction usually sit in two different systems, so the list is slow to assemble by hand. We estimate it is worth working through early, and we show it to you during the install rather than holding it back.
The loop, and the gate in the middle
After install the brain runs the same five steps, over and over, on the work that keeps being dropped.
- Reads. What changed across your systems since it last looked.
- Proposes. The next action, written out in full, as a draft.
- Gate. A person on your team approves it, edits it, or throws it out.
- Acts. Only after that approval.
- Learns. What you approved, changed, or rejected goes back into the brain.
The gate is not a setting. Every action that reaches a client or a system of record waits for a person, and there is no switch that turns that off. It costs us the demo where the machine does it alone. We would rather build the version a firm is still using in month six. More on why we work this way.
3
states the brain keeps context in: working, canonical, and unresolved
0
firms running this today. We are in build, and the first 5 get design partner terms.
Your brain is sealed
Each firm gets its own brain and it stays inside that firm. We do not pool your records with another company's, and one client's brain is not trained on another client's work.
What does travel between firms is narrower than that: a benchmark of what good looks like in the trade. It is a shape, not anyone's data, and it is what crosses between firms.
Modules sit on the brain, and only on the brain
The brain is the product. Modules are separate, bought one at a time at $750 a month each, and a module on its own has nothing to read and nothing to check against. So we do not sell one that way. It is the structural fact behind the way this is priced, and we would rather you hear it now than in month two.
The proof
The proof is the week one list again, with today's figures beside it. Follow-up gaps. Dormant relationships. Response time. Admin hours. Where a number moved, you see by how much. Where it did not move, that is in the same report, in the same size type, with what we intend to try next.
What it costs
| Firm size | Install, one time | Brain |
|---|---|---|
| Up to 10 staff | $1,500 | $1,000 a month |
| 11 to 25 staff | $2,500 | $2,000 a month |
| 26 to 50 staff | $4,000 | $3,500 a month |
- Modules are $750 a month each, bought one at a time.
- More than five systems to connect adds a day to install and $500 to the price, stated during scoping.
- We do not price per seat. What we are selling is fewer admin hours, and a seat licence would bill you for headcount and charge you more for succeeding.
- Design partners, the first 5 firms only: install at half, and the brain one band down for 12 months.
What it will not do
Three limits worth knowing before the first call. They are choices we made, not gaps we are planning to close.
01Touch a client or a record without a person approving it first
02Move your data into another firm's brain
03Run a module with no brain under it
Install runs a week. $1,500 to $4,000 one time by staff count, then $1,000 to $3,500 a month for the brain. The first 5 firms get design partner terms.