How we work
This page is for the sponsor and the BA. It says who does which part of the work, why that changes what it costs and how long it takes, what we need from you, and what the first week looks like.
The delivery model
We deliver the same kind of work a software house delivers: designs, interfaces, systems, tests, documentation. The difference is who does which part.
| Part of the work | Who does it | Who checks it |
|---|---|---|
| Judgement, architecture, design decisions and the relationship with you | Who does itMe, always | Who checks itYou: the sponsor signs every decision |
| Anything that touches your live systems, security or data boundaries | Who does itMe, or a trusted engineer I have worked with, named in the statement of work before they start | Who checks itMe |
| Drafting specifications, flowcharts and documents; research; writing and testing code; building test cases; runbooks | Who does itAI agents I have trained and direct | Who checks itMe: every design decision and every deliverable |
Which part goes where is chosen by risk and reversibility, not by convenience. Work that could harm a live system if it went wrong stays with people.
Why it matters to you. On a software project, much of the fee pays for hours spent drafting, building and testing. In our model agents do much of that under my review, so fewer engineering hours go into each deliverable, and the senior time goes on the decisions that carry the risk. What that buys you is a price fixed before the work starts and a shorter route to go-live. How many hours it takes us is our risk, not a line on your invoice, and the hours we save go into getting you live sooner. We do not claim the work costs less than another supplier, because we have not measured that.
What the agents never do. They never touch your live systems. They never hold your credentials. They never make a design decision or sign anything off. Nothing they produce reaches you without my review.
Your material. Your documents, data samples and notes are kept on our systems on AWS in London. They go only to AI services on a business or API plan that has a data processing agreement and does not train on your data by default. Working material is deleted or returned within 30 days of the engagement ending; records of the work are kept six years.
Industrial™ does not sell AI as a service. The agents are how we deliver. Where your change itself needs AI, AI UVD provides that part under its own terms (aiuvd.com).
Small, and quick because of it
We are small on purpose. For you that means:
- Decisions in the room. You talk to the person who designs the system. Nothing is relayed through an account manager.
- No hand-offs. The person who scopes the work is the person who delivers it.
- A short line from question to answer. A question about the design is answered by the person who made it.
- Changing course without the machinery. If the project needs to change direction, you write it down in a short note, I reply with the effect on price, dates and risk, and your sponsor says yes or no. It is still written down, so nobody argues later about what was agreed. It is just not a committee.
The honest consequence: we take on a small number of clients at a time.
What we need from you
- A sponsor who can decide. One named person who signs the decision and owns the budget.
- A BA or process owner as the day-to-day contact. The person who knows how the work is really done, including the workarounds.
- The documents that exist, however out of date: specifications, process maps, interface lists, supplier contracts, previous proposals.
- Access, narrowly and in writing. Read access first, and only to what the work needs. Anything more is asked for in writing, with the reason.
- Time from the right people, in short sessions. Your people are asked for answers, not days.
- One sentence on what done looks like. If we cannot write it in week one, finding it is the first task.
Week one
Kick-off with the sponsor and the BA. Scope, what done looks like, who decides, what cannot break, and the dates that matter.
Walkthrough of the current system with the people who use it and the people who support it.
Reading. I go through the documents, the data and the interfaces, with agents doing the first pass through long documents and exports under my review. Nothing on your systems is changed.
Stakeholder map. Who is affected, who has a say, and which third parties are involved.
End-of-week note, in writing. What I understand, a first current-state flowchart, open questions with an owner for each, and the decisions that will be needed and when.
What you receive
- Discovery notes, and current-state and target-state flowcharts
- Wireframes for every screen or document that changes
- Written options, each with its trade-offs, cost, risk and what it needs from you
- A decision note for the sponsor to sign
- A delivery plan: sequence, milestones, who does which part, and what can go wrong
- If we deliver the whole: the built change, tested, with a runbook and a recorded handover. Your business owns the code and configuration built for it, on payment; our pre-existing tools built into it are licensed to you. For a hosted service, ownership follows the option chosen in the statement of work (/engagements/#ip).
How decisions get made
- The sponsor decides; I recommend. Every option comes with my recommendation and my reasons, in writing.
- If it is not written down, it has not been decided. Decisions go in a decisions log that you keep.
- Disagreements are written up as options, not settled by whoever is loudest in the meeting.
- Nothing irreversible happens to a live system without a written yes from your decision-maker.
What we won't do
- Fill a seat with no defined outcome, or offer open-ended "architecture support".
- Start work without a statement of work, a fixed price and named deliverables.
- Change a live system without written agreement, or with more permissions than the job needs.
- Let an agent near your live systems or credentials.
- Put your material into an AI service that has no data processing agreement or that trains on customer data.
- Hide a problem. If something will be late, cost more or not work, you hear it from me in writing as soon as I know.
Risk, answered directly
Prices are fixed in writing before work starts, and overruns inside the agreed scope are ours to absorb. Every output is written down and handed over, so your people can carry on without us. Changes are tested, or run in shadow alongside the current process, before they touch live work. And I don't promise what I couldn't deliver alone: I can personally deliver or rescue any part of the work.
Every engagement is architected by me, and I'm the person you deal with.
We take on a small number of clients at a time.