Picture your finance team a year after its first workflow went live. Every overdue invoice gets a prepared follow-up the day it falls due, the controller reviews drafts instead of writing them, and the hours that went on chasing now go on the cash forecast. That is 10X the ROI of your existing AI spend in one team, and nobody outside the company sends you an invoice for it.
Some providers would rather take a cut of that gain: little or nothing up front, then a share of whatever you save. It sounds confident, and it sounds fair. We charge a fixed fee instead, because of what each arrangement asks of you before, during and after the work.
The short answer.
Before any work starts, you and we agree the scope: which workflow is built, what it must do, what it connects to and what done looks like. The price is in that agreement, and it does not move. Then we build. We ask the people who do the work how they do it, build the workflow on the ChatGPT, Claude, Copilot or Gemini you already pay for, test it on your own examples and hand it over once the person who does the work has run it on a real case. Each sprint includes a support window after handover, and we correct anything that does not do what was agreed within it. What we test before we call an AI workflow ready sets out the standard it has to meet.
After that your team runs the workflow, and every gain it brings stays with you. The pricing page gives the same answer in three sentences; this page shows why.
What each arrangement asks of you.
Put the two side by side and read down each column. One is a single agreement made before the work starts. The other is an account that stays open for as long as the workflow runs.
| When | A share of the savings | A fixed fee |
|---|---|---|
| Before work starts | Agree a baseline both sides accept, what counts as a saving, how much of any gain is credited to the workflow rather than to a new hire, a quiet month or a price change, and the provider’s right to inspect your records. | Agree the scope: the workflow, what it must do, what it connects to and what done looks like. You know the whole price. |
| During the work | Start counting: hours, cases, who ran the workflow and on what, so there is a number to share. | Tell us how the work runs. We ask the people who do it, and nothing records or monitors them. |
| Each billing date | Reconcile the savings, check the provider’s arithmetic and pay an amount you learn afterwards. | Nothing. The agreed fee covered the agreed scope. |
| When the work changes | Reopen the baseline whenever the team, the volume or the process moves, because the saving is measured against it. | The workflow’s owner decides what changes. New requirements are quoted as new work, at a fixed fee agreed before it starts. |
| A year on | Still paying for gains your own people now produce, review and improve. | Running the workflow as your own and keeping every gain it brings. |
Every line in the middle column is work for your managers, a conversation with a provider about your own records, or a reason to count what your people do. The right-hand column asks for one careful conversation at the start, and after that the time goes on the work.
A share of savings pulls toward cuts.
The saving easiest to put on an invoice is fewer hours paid for. Faster replies to clients, fewer errors reaching a customer, a forecast ready on Monday and capacity moved to growth are often the bigger gains, and they are much harder to count. A contract paid on counted savings pulls both sides toward the gain that is easiest to count, and toward timing the people who do the work. Neither is the reason most companies put AI to work.
Faster is the floor sets out the gains beyond time. What will your team do with the capacity it gains? is the worksheet for deciding where the freed hours go. That choice belongs to you, and a fee schedule should not lean on it.
A fixed fee keeps our side simple too. We are paid to build a workflow that does what was agreed, proven on your real work, and we earn the next piece of work when the last one runs well in your team’s hands.
Measure your gains for yourself.
None of this is a reason not to count. Count for your own purposes: to see the return, to choose what to build next and to show the leadership team. Use the six-measure baseline in What is an AI workflow worth?: elapsed time, hands-on effort, review, rework, output quality and adoption, written down from your own records before the first run. Your people own the numbers, so nobody argues with them, and no invoice depends on them.
Finding out how the work runs is a matter of asking, not watching. Asked or watched explains why we ask, and what software that watches would miss.
Every service we offer is priced this way. The 3-Day Workflow Sprint and the 2-Week Workflow Sprint have published fixed prices. Larger work, across teams or to launch a new function, gets a fixed fee agreed after scoping, with the deliverables and timing written down before it starts. If you are weighing a sprint against a monthly fee for each agent, An AI employee or a more capable team? answers that question.
The workflow runs on the platform your team already uses, at the pace you set, and your organization keeps control of how the work gets done. Read the prices, then bring the workflow you want built first.