Approach
How the work actually gets done.
Most of what a buyer can check before signing is here: the stages, what each one produces, who holds the work at each point, the things we turn down, and what happens to your data. If any of it sounds unlike how your last programme ran, that is the conversation worth having.
Method
Four stages, each with an output you can hold
Discover
Find the problems worth solving and rank them by value and feasibility.
You getPrioritised use cases and a business case
Prove
Build the riskiest part first, on real data, and measure it against a baseline.
You getWorking prototype with evaluation results
Build
Engineer it into your systems with monitoring, guardrails and a human in the loop.
You getProduction system your team can run
Scale
Train the people, redesign the workflow, and extend to the next use case.
You getAdoption metrics and a roadmap for what comes next
The shape of it
The handover is the plan, not the ending.
We start holding most of the work and finish holding almost none of it. Your people are on the build from the first stage, so the transfer happens continuously rather than arriving as a document at the end.
Principles
What we will and will not do
Four commitments, each with the trade-off it costs us. A principle whose opposite nobody would choose is not a principle.
01
We build in your environment, not ours
The work lives in your repositories, your cloud and your tooling from the first week. It is slower to start — we spend days on access that a demo on our own laptops would skip — and it is the only way the thing we build is still yours when we leave.
02
The riskiest part gets built first
Not the most impressive part. If the data is unusable or the accuracy fails where it matters, that should surface in week two, not month five. This makes our early demos less flattering than they could be, and it is the point.
03
One stage at a time, priced before it starts
You agree to a stage, not a programme, and you can stop at the end of any of them and keep everything produced — including the code. We give up the certainty of a long contract for the discipline of having to earn the next stage.
04
We work to leave
Every engagement names the people who will run the system afterwards, and training, runbooks and documentation are part of the build rather than a final invoice. A firm that cannot be got rid of has an incentive we would rather not have.
Security and governance
How we handle your data
You are weighing up a firm you cannot yet check with three references, so here is the part most firms make you ask for.
Data handling
Your data stays in your environment wherever the work allows it. Where it cannot, we agree in writing what moves, where it is processed and when it is deleted — before anything moves.
Access model
We ask for the least access that lets us do the job, granted through your process and revoked at the end of each stage. Where access cannot be granted, we work with redacted or synthetic data and tell you what that costs in accuracy.
Commitments
We will sign your DPA, work to your security policy, and assign the IP in everything we build to you. Engagements end on 30 days' notice, from either side.
Questions
How engagements run in practice
How does an engagement start?
How long until something is in production?
Will this work with our existing data and tools?
How do you handle security and data privacy?
Who will we actually work with?
How is pricing structured?
Next step
Have a problem worth solving?
You will hear back within a working day, from the person who would do the work rather than someone selling it. If we are not the right firm for what you need, we will say so and tell you who is.

