The first project
Start with one automation, not a transformation programme
Most AI projects in construction fail for the same reason: they are too big to measure and too slow to prove anything before the enthusiasm runs out. So we start with one workflow, on live work, with a number at each end.
The deal in one paragraph. One repetitive workflow. Measured before we touch it. Built into the systems you already run rather than a new platform. Live on a real project inside about six weeks. Measured again the same way at the end, and we tell you the real number even when it is disappointing. You own everything, and you are free to walk.
How it runs
- 01
We measure the thing first
Before anything is built we establish what the workflow costs you today: how many hours, how many people, how often, and where it goes wrong. If we cannot measure it, we cannot prove we improved it, and you should not pay for a claim nobody can check.
- 02
We pick the smallest useful thing
One workflow. Not a department, not a transformation. The right first candidate is repetitive, currently manual, owned by one team, and annoying enough that people will notice when it stops.
- 03
We build it into what you already run
Inside your Microsoft 365, your SharePoint, your existing systems. No new platform for your team to adopt, because adoption is where most of these projects actually die.
- 04
It runs on live work
Not a demo on sample data. A real workflow on a real project, where the awkward cases turn up and the output has to survive contact with people who did not ask for it.
- 05
We measure it again and tell you the truth
Same measurement, same method. If the saving is smaller than we hoped we say so, with the number. A consultancy that only ever reports success is not measuring anything.
Who this is not for
Worth saying plainly, because a first project that was never going to work wastes your money and our reputation.
- You want one number to go up. If the brief is “use AI” rather than a named process that annoys a named team, there is nothing to measure and nothing to prove.
- Nobody can spare an hour a week. We need someone who knows how the work actually happens, for a few hours across six weeks. Without that we automate the version of the process written in a procedure nobody follows.
- The real problem is upstream. Sometimes the answer is that a folder structure, a template or a two-line rule change fixes it. We will tell you, and we would rather lose the project than automate a mess.
And then what?
If it works, the second one is easier and cheaper, because the measurement habit and the access are already in place. Some businesses stop after two. Others go on to a bespoke platform or a secure agent over their project records. None of that is decided now, and it should not be: you cannot sensibly scope a platform before you have proved a single workflow.
Common questions
What does a first project cost?
The price is agreed on the call, once we know which workflow and how much of it already exists. We would rather quote a number we can honour than publish one and revise it upwards after you have committed. For context, published UK prices for comparable first engagements ran from about £497 to £3,000 in August 2026.
How long does it take?
Around six weeks in most cases. Two of those are measurement and design, which sounds slow and is the reason the rest works. A project that starts building in week one is usually automating whatever was easiest to describe rather than whatever costs the most.
What if it does not work?
Then you have a measured answer about one workflow, which is worth having, and you stop. That is the entire point of making the first step small. You have not bought a platform, signed a multi-year licence or restructured a department, and everything built is yours regardless.
Do we have to keep working with you afterwards?
No, and nothing is designed to make leaving difficult. Code, workflows, documentation and credentials are handed over as part of delivery. We build on mainstream tools specifically so another developer can maintain what we leave. See the trust centre for how that works.
Will this force us onto your product?
No. Construction Metric is our product and a consultancy that sells one has an obvious incentive to steer work towards it. First projects are tool-neutral, we recommend other vendors regularly, and we take no commission from any software vendor.
Which workflow should we start with?
The one your team complains about, that happens weekly or more, that one person currently owns, and where the output is a document or a record rather than a judgement. If you are not sure, that is what the first call works out, and sometimes the honest answer is that a spreadsheet fixes it and you do not need us.
Bring one annoying process
Twenty minutes. Describe the workflow and roughly how many hours a week it eats, and you will get a straight answer on whether it is worth automating, including when the answer is no. If you would rather see it mapped before you talk to anyone, put it on the Drawing Board first. It is free and asks for nothing.