One role today. Your whole entity next.
Four steps take an organisation from a map of the work to a live system your own team runs. The order does not change, and no step is skipped.
Building before mapping is how pilots die.
Almost every failed artificial intelligence programme we are asked to rescue started at step three. Somebody built something impressive on top of a process nobody had examined.
The result is a system that automates a workflow which should not have existed, or one that no member of staff trusts because nobody agreed what it is allowed to decide on its own.
Understanding and decomposition are not preliminaries. They are the work. By the time we build, the argument about what the system does and does not decide is already settled and written down.
- Human
- Judgement, relationships, accountability
- AI
- Drafting, analysis, synthesis under review
- Automation
- Deterministic, rules-based, repeated
Proportions are illustrative. Yours come out of step 01.
- Step 01Understand
We sit with your people and map where the work and the hours really sit — which is rarely where the process document says. Shadowing, structured interviews, and a time-and-motion read across the function.
The output is a written map of the work as it is actually performed, with the hours attached, and a ranked view of where the return would be biggest.
Process mappingTime and effort analysisReturn rankingAI readiness assessmentFind the source.
- Step 02Decompose
Every role is broken into its constituent tasks, and each task is assigned to one of three channels: what a person keeps, what artificial intelligence does under review, and what automation runs deterministically.
This is the step clients find most uncomfortable and most valuable. It forces an explicit answer to a question most organisations avoid: which decisions are we willing to let a system make without a person in the loop?
Cut the channels.
- Step 03Build
The agent tasks become a working system. Every point where a decision carries consequence gets a gate — a defined checkpoint where a named person approves, overrides or escalates before the process continues.
Nothing goes near live data until it has been audited. Every action the system takes is logged against the person accountable for it, which is what makes the system defensible in a review rather than merely fast.
Set the gates.
- Step 04Enable
We hand over a dashboard your team runs, and we train a named keeper inside your organisation to own it. Not a user — an owner, who can change a rule, retire an agent, and explain the system to an auditor.
The measure of a finished engagement is that you could stop working with us and the system would keep running correctly. If that is not true, the engagement is not finished.
Appoint the keeper.
The AI+ Readiness Assessment.
A structured session with a senior engineer on one function of your choosing. You receive a written read on where the hours sit, which tasks are genuinely automatable today, and what would have to be true before anything is built.
Yours to keep, with no obligation to continue.
Where the method can take you.
From One Task to a Nation
The five levels of enablement, from a single job to sovereign capability.
What We Build
The systems the method produces, and how each one is gated.