Skip to content
Doing AI

How we work

We build from inside the operation.

We work from inside your operations rather than from a document: we sit where the work happens, build against what is actually true there, and remain accountable once it is live.

The sequence

From first workflow to production.

Four steps, in this order. The numbers are a sequence, not decoration.

01

Week 1

Identify the constraint

We work alongside your team to locate the process that is expensive, repetitive and worth solving first.

02

Weeks 2–4

Build the first system

One narrow, well-defined system connected to your live tools and your real data. No sandbox environment.

03

Weeks 4–6

Deploy to real users

The system goes live with your team and your customers. We monitor it, correct it and harden it under real load.

04

Ongoing

Operate and extend

Monitoring, iteration and the next workflow once the first one holds steady.

Our position

The models are ready. The software around them is the work.

Most organisations already have access to capable AI. The difficulty sits everywhere else: the data a system reads, the applications it writes back to, who is allowed to approve what, the exceptions that still need judgement, and the accountability that starts the moment real work depends on it. That is not a model problem, it is an enterprise software problem, and it is the same problem in every sector — the vocabulary changes and the shape does not. That engineering is what we do, and we do it from inside the operation: an invoice that disagrees with a goods receipt note is not an illustration here, it is Tuesday. The models will keep changing. Governed software that people can be held to will not.

01Production is the deliverable

Every engagement is scoped against a system that goes live and handles real volume. If we cannot see a credible path to production, we say so before any commitment is made.

02You own what we build

Working software runs in your own accounts, fully documented, with your data exportable at any point. Ownership and access are written into the agreement, not assumed.

03We build into your existing stack

Your CRM, your spreadsheets, the ERP already in place — they stay. We integrate with the systems your teams are trained on rather than replacing them.

04The team you meet is the team that builds

The founders on your first call remain accountable through deployment: one leads engineering and one leads the relationship.

05We scope honestly, including when the answer is less

Some processes are better removed than automated. Identifying which is which is part of the work, even when it reduces the size of the engagement.

Commercial terms are open to discussion. These are not.

Why it holds

Built for production, not for presentation.

Engineering-led delivery

Each engagement produces working software running in front of your customers and your team.

Integrates with your existing stack

Your CRM, your spreadsheets, your internal tools. We build into what you already run.

Your data stays yours

Governed, auditable, and never used to train external models.

Weeks, not quarters

The first system is scoped, built and in production in four to six weeks, at a price agreed before anyone writes code.