20 minutes · No commitment
Your firm changes every quarter. The context it runs on has to change with it.
Cisco's 2025 AI Readiness Index found that 13% of organisations have a mature, repeatable process for generating and scaling AI use cases. The rest have people who know how, which holds until those people are busy, and does not survive them leaving.
Writing it down is the part firms buy. Keeping it current is the part that gets dropped, because by default it is nobody's job.
Your AI Business Brain is accurate on the day the install hands it over. Then a licensee condition changes, a template is rewritten, a platform is swapped, and a decision gets made in a meeting that nobody writes down. By the end of the quarter it is describing a firm that has moved.
The Context Layer is what keeps it true. It is counted per seat, per month, and nothing else on the MethodWorks ladder runs without it.
What it does, in 4 parts
| What it does | What that means |
|---|---|
| The brain stays true | A weekly pass over the brain proposes corrections where it has drifted from how the firm now works. Nothing is applied until a person says yes. |
| The brain stays awake | A note each month, per seat: what moved, what stalled, what is waiting on you, and what has gone unactioned past 30 days. |
| The number is proven | The 5 hours a week per person is counted against the tasks you named at the start, at week 6, at week 12, and each quarter after that. |
| Someone owns it | A quarterly review per seat, plugin updates, and new skills as they are released. |
The rhythm is what separates this from a document. A record corrected once a year describes the firm as it was the last time somebody had a spare afternoon. Corrected every week, it stays close enough to the work that people keep reaching for it instead of rebuilding the answer from memory.
What arrives each month
One note per seat, covering 4 things: what moved, what stalled, what is waiting on you, and what has gone unactioned past 30 days.
Every line names a piece of work and its state. Something that has stopped appears with what it is waiting on. An opportunity appears once the clock on it passes 30 days. The note exists so that the things quietly not happening get named on a schedule, instead of surfacing in a client call.
What is allowed into that note is governed by the line below, and that line does not move.
Where the reading stops
The Context Layer reads your firm's own material. It reads the shared brain, which is your record of how the firm works, sitting in your files and your tenancy. It reads whichever of your work systems you have connected to it: mail, calendar, the CRM, the ledger, the project tool.
It never reads what an individual types into their own Claude.
That is the line, and it comes with a test you can run yourself on anything we propose later. Take any signal and ask which question it answers: does it report what a person did with Claude, or does it report whether their machine can run Claude. The first is refused. Whether a seat is set up and working falls on the second side of that test. What somebody asked, drafted or explored in their own environment falls on the first, and does not get built.
The boundary goes to your staff in writing at every install, before anyone is asked to use it. People work out quickly whether a new system is there to help them or to watch them, and a team that has decided it is watching them stops putting anything real into it. Written down and handed over, the boundary is something they can hold us to.
None of the intelligence the product is sold on needs to cross that line. What moved, what stalled, what is late, where two people are pulling against each other, and what nobody has picked up are all visible in the firm's own systems and in the brain itself.
Nothing above it runs without it
Each rung is bought on its own, and none of them is a discount on the next. The Context Install is always first, because there is no brain to maintain until one is built. The Context Layer is second, and it is required before anything above it.
Decision Deployment and the Decision Layer both read the brain and run a decision loop over what they find. A brain that has stopped being true produces a decision session about a firm that no longer exists, so neither product starts until the install and the Context Layer are live.
The Context Layer is bought at every size. Under 15 staff it is the top of the stack, and the firm runs its own loop on it between quarterly reviews. At 15 to 50, Decision Deployment sits above it. At 50 plus, the Decision Layer does. The full ladder, and what each part is counted by, is on how it works.
How it is counted
Per seat, per month, on the same seats the install covered. The install sets up a structured Claude environment for each person against what that person actually does, so keeping it true is work on each person's environment rather than one job for the firm.
The quarterly review runs inside it. There is no separate quarterly fee.
Connecting a system Claude cannot reach natively is quoted separately, per build, scoped. The figure for the Context Layer follows your seat count, and you get it on the call.
Start with the hours
The Context Layer sits on top of a Context Install, so the install is the first thing you buy and the first thing to price. The booking page opens with a calculator rather than a form. Put in your seat counts by role and it returns the hours the install is built to give back, at 5 hours a week per person across 48 working weeks. Add your own hourly rate and it prices those hours in your numbers. Then you pick a time.