AI Agents
Agree what it returns. Pay from that.
We size the build to one year of measured return and hold the balance until you sign off. If the number is not worth it, we tell you and stop there.
Most AI proposals ask you to fund a build and hope. You pay for hours, the system ships, and whether it was worth it becomes an argument nobody can settle because no one agreed the baseline.
We do it the other way round. Before any paid work starts, we sit down and put a number on what the agent will return in its first year — and we write it down, together, in a document you both sign.
That agreed number becomes the build fee. You are not buying our hours. You are buying an outcome we have already agreed is worth what it costs.
How it works
Four steps, and a free one at the front
No surprise fees, no hidden costs, no bait-and-switch. The price you discuss in consultation is the price you pay.
01 — Free Savings Workshop
We map the workflow with you and document what it costs today: the hours, the error and rework rate, the cycle times. Then we agree a conservative first-year return figure in a written Savings Basis. No cost, no obligation, and no paid engagement exists until you approve the plan.
02 — The build fee is the agreed return
Sized to one year of what we jointly measured. If the workshop concludes the return does not justify the build, we say so and that is the end of it. We would rather lose the engagement than manufacture a business case.
03 — Pay as we deliver
Payments track milestones you approve, with the balance due only when you sign off against the acceptance criteria. You are never funding work you have not seen — which directly addresses the objection we hear most: full payment demanded up front.
04 — Year two is yours
Year one nets to zero: the return covers the build. Everything it returns after that stays with you. If measured first-year return lands materially short of plan, you receive a capped rebate — and if it beats plan, we do not ask for more. That upside is entirely yours.
What counts as a return
Savings are one kind. Recovered revenue, avoided penalties, and released capacity are others. What matters is that it is measurable and that we agreed how to measure it before the build, not after.
When we use a fixed price instead
When the return genuinely cannot be pinned down — exploratory work, or a first deployment into an unfamiliar process — we scope a fixed-price engagement with a clear floor instead, so you still get a predictable number.
Questions
Frequently asked
What if it does not return as much as projected?
You are protected. If measured first-year return falls materially short of the agreed Savings Basis, you receive a capped rebate. The symmetry matters to us: we also never charge more when the agent beats the plan.
Does this apply to governance work too?
No, and we will not pretend it does. A governance readiness assessment does not generate a savings stream to bill against. Applying an outcome model where there is no measurable return would make the number arbitrary and would undermine the model everywhere it genuinely applies. Governance engagements are fixed fee, scoped and agreed before we start.
Who runs the agent after launch?
Ongoing monitoring, maintenance and MLOps are available as a separate managed service, billed separately from the build. We keep it separate deliberately — bundling run costs into a build fee sized to first-year return would quietly consume the savings the model is supposed to hand you.
What do we pay to find out whether this is worth doing?
Nothing. The Savings Workshop is free, and it is genuinely a workshop rather than a pitch. You leave with a documented view of what the workflow costs you today, whether or not you engage us to change it.
Interested in this agent?
Let's scope it against your compliance requirements and agree what it returns.