NewThe 6-week AI Pilot ·  Fixed scope. Fixed price. £18k. 4 pilots per quarter.  See how it works →
AI Strategy7 MIN READ·MAY 2026·BY ASSUREPATH

What an AI pilot actually costs in the UK. A line-by-line breakdown.

Most AI consultancies will not publish their pricing. We will. Here is our £18,000 fixed-price pilot, itemised line by line. The maths behind why it is not £40,000 and the three things that are not on the invoice.

We get the same call about once a week. "We're talking to a Big Four firm, a boutique AI shop or a generalist consultancy. Their pilot is £80,000 over six months. Yours is £18,000 over six weeks. What are we missing?"

The honest answer is usually: nothing. They are charging more. That is it.

But it is a fair question and it deserves a real answer. So we are going to do something the rest of the industry will not. Itemise our pilot, show the maths and explain why the number is £18,000 and not £40,000.

"Almost nobody in AI publishes their pricing. We will."

Why no one publishes this

Most AI engagements are sold on a Time & Materials basis. A day-rate of typically £800 to £1,800 depending on seniority, multiplied by a week-count nobody commits to in writing, padded with a "discovery phase" whose main purpose is to delay scoping until after you have signed.

T&M is good business for the seller. The variance lives with the buyer. If it takes longer, that is a change request. If your data is messier than expected, that is a change request. If a meeting reveals a new requirement, that is a change request.

Fixed price moves all that variance back onto the seller. Which is exactly why most AI consultancies do not offer it. It is harder to sell, harder to scope, harder to deliver. If you mis-estimate, you eat the difference, not the client.

We made our 6-week AI Pilot fixed-price anyway. Here is what £18,000 buys you, broken into the seven things we actually do.

The seven line items

These are the equivalent T&M values for each phase, calculated at our standard senior-engineer rates. We round them into £18,000 flat and absorb the variance.

01
Scoping workshop and fixed proposal. £2,400. Week 1.
A focused half-day session with your team. We pick the one workflow, define what "good" looks like and write a fixed-scope proposal you can sign or walk away from. No discovery fee. If you do not sign, there is no invoice.
02
Architecture, data and eval blueprint. £3,200. Week 1 to 2.
Senior engineer maps the build: which model, which integrations, where the data lives, what the eval suite looks like. This is the document that prevents week-5 surprises.
03
Engineer build. £11,800. Week 2 to 5.
The main line. Senior engineer plus delivery, building the actual workflow against your data. Twice-weekly demos so you see progress in working software, not Gantt charts.
04
Eval suite and accuracy targets. £1,600. Week 3 to 5.
AI without evals is theatre. We build a measurable test harness so you can see exactly how often the workflow gets it right, how often it gets it wrong and how that changes as you tune it.
05
Hosted deployment and go-live. £1,800. Week 5 to 6.
We deploy into your environment (or ours, if you prefer) on UK infrastructure. Real users can use it by the end of week six. Not a demo URL. A production workflow.
06
Runbooks, handover docs and training. £1,200. Week 6.
Written runbooks so your team can operate it without us. Two training sessions for whoever picks it up. You own the system on day one, not after a separate consultancy engagement.
07
Go / no-go decision pack. £900. Week 6.
An honest write-up of what the pilot proved, what it did not and whether scaling it makes commercial sense. If the answer is no-go, we say so. And we tell you what the £18k actually proved on the way to that conclusion.

The maths

Add the line items and you get £22,900. That is what an equivalent T&M engagement would invoice for, assuming nothing went wrong.

We charge £18,000, fixed. We absorb the £4,900 difference and we absorb the risk of overrun. If a piece of the build takes longer than budgeted, that is our problem. Not a change request.

"Fixed price is harder for the seller. Easier for the buyer. That is the whole point."

The bits that are not on the invoice

Three things make the £18,000 worth more than the seven line items above.

01
50/50 payment schedule
Half on signature, half on go-live. We do not get paid in full until the workflow is actually running in your environment. Most pilots invoice up-front or against milestones the consultancy sets itself. Ours completes only when something works for you.
02
Live by week 6, or we keep building, free
If the workflow is not in production by the end of week six, invoice two is paused and we keep building at no additional cost until it is. We have never had to invoke it. The point is that the risk of overrun sits with us, not with you.
03
Honest go / no-go at week 6
If the pilot proves the workflow is not worth scaling, the decision pack says so. We would rather you spend the next £100,000 on something that works than the next £100,000 extending a pilot we already knew was marginal. The £18k is the cheapest way to find out.

Why fixed price is harder for us, not for you

On a T&M engagement, every hour is billable. If a meeting runs long, that is billable. If we mis-scope something, that is billable. If your data is messier than expected, that is billable. The consultant's incentive is to keep the meter running.

On a fixed-price engagement, every hour comes out of our margin. The incentive flips. We want the smallest scope that proves the value, the fewest meetings that get to the right answer and the cleanest data flow we can ship. Speed is in our interest because we do not get paid more for going slow.

This is exactly why most consultancies will not price like this. It works for clients. It is painful for sellers who would rather charge by the hour.

The fixed-price asymmetry
  • T&M: consultant is incentivised to run long
  • Fixed: consultant is incentivised to ship fast
  • T&M: client carries the overrun risk
  • Fixed: consultant carries the overrun risk

Who this works for (and who it does not)

Fixed-price pilots only work when both sides can agree on scope quickly. That means a fairly narrow profile this is built for.

  • UK or Irish business, 50 to 1,000 staff. Small enough to make a fast decision, big enough to have one workflow that genuinely deserves automation.
  • One specific workflow already identified. If you are still surveying the business for ideas, you need a discovery sprint first, not a pilot.
  • A sponsor who can say yes. MD, FD or Ops Director. Someone who can call the scale-up decision in week six without a steering committee.
  • Data we can reach in week 1. If it is air-gapped, paper-only or in an undocumented legacy system, the scope changes.
  • A fast yes. The £18k fixed assumes no multi-vendor RFP cycle. If procurement adds three months, the economics break.

If you do not fit that profile, the pilot is not right. We would rather tell you in week zero than waste both our times. The 30-minute call exists to figure that out honestly.

Book a 30-minute scoping call

We run four pilots per quarter. That cap is deliberate. Pilots are senior-engineer led from week one to week six, and four is the number we can deliver properly without thinning the bench. When the quarter fills, the next booking slot is the following quarter.

We do not run a marketing list. The call is 30 minutes, with an engineer rather than a salesperson, and you will get an honest read on whether the pilot is a fit. If it is not, we will tell you what is. Even if the answer is someone else.

See the full 6-week AI Pilot page for the proposal PDF, or book a 30-minute call with the engineer who would scope it.

Talk to an engineer. Not a salesperson.

Curious whether £18,000 buys what you would actually need? Find out in 30 minutes.

No marketing list. No sales pitch. Honest answer at the end of the call. Including whether we are the right fit at all.