One AI use case, built properly and handed over
A fixed-scope sprint that takes a single well-defined problem, builds a working solution inside your existing systems, and leaves your team able to run it without me. This is the only delivery work I take on, and it comes with a published test that governs when I am allowed to accept it.
Book a free 30-minute conversationThe problem this addresses
Most organisations no longer need convincing that AI could help them. Many can even name the place it would help most. What stalls is everything after that.
The use case gets discussed for two quarters. A pilot is built by someone enthusiastic and then nobody owns it. A vendor proposes a platform that solves a more expensive version of the problem. Or the work is scoped so broadly that it can never finish, so it quietly becomes a line in next year's plan.
What closes that gap is not more analysis. It is one contained problem, an agreed definition of done, and someone accountable for the thing working by a named date.
What it is
A short, fixed-scope build. One use case, agreed in writing before anything starts, delivered inside the systems you already own wherever that is possible.
I do the work myself. There is no team of contractors working from a brief I wrote, which is the usual reason this kind of engagement drifts. The sprint ends with a handover: documentation, a working session with the people who will use it, and a short period afterwards where anything that surfaces is dealt with as part of the engagement rather than as a change request.
The test of a good sprint is that you do not need me afterwards. If the thing only works while I am attached to it, I have built the wrong thing.
The implementation conflict test
The obvious question about this page is why an independent adviser sells delivery at all. The honest answer is that independence is only real until there is delivery revenue on the table. So the conditions are published, and you can hold me to them. All five must be met before I accept delivery work arising from advice I was paid to give.
-
1
The recommendation came first, and in writing
The advice exists as a dated document written before any delivery conversation. You can read what was recommended and when, and check that the delivery scope follows it rather than the other way round.
-
2
You were given a genuine alternative
At least two other credible suppliers are named, with an honest view of where each would be stronger than me. Not a courtesy list. A real one.
-
3
The delivery scope is smaller than the advice
If the recommended programme is larger than the piece I would build, I take the piece and you keep the programme. I do not use advisory access to win the whole thing.
-
4
No ongoing dependency is created
The engagement ends with your team able to run what was built. If the work would only function with me still attached to it, it is the wrong work for me to do.
-
5
The board or its equivalent has seen this test
Whoever approves the spend has been shown these conditions and has agreed the delivery work on that basis. This is not a page you have to find. It gets put in front of them.
If any one of them fails, I help you find someone else and take nothing for the referral. If you think a piece of delivery work failed this test, say so and I will withdraw from it.
What you get
- One working solution, live in your systems rather than sitting in a demo environment
- A written definition of done, agreed before the build starts and used to judge the end of it
- Documentation your team can actually follow
- A practical handover session with the people who will use it day to day
- A defined period after handover where anything that surfaces is dealt with, not re-quoted
- An honest view of what should and should not be automated next
How this is scoped
Delivery work is scoped tightly on purpose. An open-ended AI project is the failure mode this is designed to avoid.
What sets the size
How clean the data is, how many systems have to be touched, and whether the process being automated is documented or exists only in people's heads. Integration into an existing stack is usually cheaper than it sounds. Undocumented processes are usually dearer.
A typical shape
Two to six weeks. A scoping week, a build period with regular check-ins that do not pull your team into every detail, then a handover week. A few hours a week from one person who knows how the process really works.
What the first conversation establishes
Thirty minutes, free, on what needs building, who else could build it, and whether I am genuinely among the better answers. Frequently I am not, and I will say so.
How a proposal follows
Fixed scope, fixed fee, named handover date, and the five conditions above answered one by one where the work follows advice I was paid for. Agreed in writing before anyone starts.
Who this is for, and who it is not for
This is for you if
- You have one specific problem that AI would plausibly solve, and you want it solved
- You would rather own a small working thing than commission a large plan
- You have someone internally who will run it once it is handed over
- You have done an AI opportunity audit or already have a clear brief
This is not for you if
- The use case is not yet identified, in which case start with a review instead
- You want an open-ended programme with capacity booked by the month
- You need the supplier to carry the solution indefinitely, because I hand over and leave
- You want a proof of concept to show a board rather than something people will use
Independence
No software to sell
I do not own, licence or resell a product. The tool chosen is the one that fits, not the one I earn from.
No platform dependency
No partner tiers and no vendor relationships, so building on what you already own is a real option rather than a concession.
Delivery is accepted, not pursued
This is the only delivery service I offer, and the conflict test above governs when I am allowed to take it.
Evidence
Delivery accepted, and what it produced.
Sector experience
Documented engagements, including the ones that did not go to plan
Written up with what was done, what changed and what did not.
All case studies »Delivery record
Programme leadership up to £2.5 million, and the rationalisation of 41 government websites
More than forty years in business and technology, more than thirty of them in digital. The person in the meeting is the person doing the work.
About »What happens next
Book a 30-minute conversationNot sure where to start?
Most clients begin with a conversation. No pitch, no hard sell.
Just a straightforward discussion about where you are and whether I can help.
Book a free 30-minute call