Before you sign anything
Five questions to ask any AI consultancy, and our answers
Question 1 of 5
Which pricing model is this quote on, and why that one for this project?
Why it mattersThe billing model decides who keeps the time saved when work goes faster than planned. Fixed price rewards a scope you can write down. Time and materials suits work you expect to change, and leaves the risk with you.
Our answer
Fixed price for each phase whose scope is written and signed. Time and materials, with a monthly cap, for work we both expect to move.
A typical engagement uses both: a fixed-price discovery and first release, then capped time and materials for what follows. The quote names the model for each phase and says why. Both are priced from the same hours, so switching a phase from one to the other does not change the arithmetic, only who carries the risk.
Question 2 of 5
Do your estimates assume AI-assisted working, or the way the work was done before?
Why it mattersIf the hours were estimated the old way and the team now works faster, a fixed price keeps the difference. The only defence is to see how the hours were built.
Our answer
They assume AI-assisted working, because that is how we build. The hours come from the AI-native column of a model we publish, category by category.
The same model prints a second column, the hours a conventional agency typically quotes for the same scope. We show both on the quote so the gap is visible. In the model's default app, code drops from 399 hours to 121 while testing stays at 109 in both columns, and a review step appears that the old method does not price separately.
Question 3 of 5
How will AI model and API costs appear on our invoices?
Why it mattersAn app with AI features pays a model provider every month. That bill can be passed through at cost, marked up, or folded into a fee you cannot see.
Our answer
At cost, on its own line, with the provider's invoice attached. Or billed straight to your own provider account, which is what we recommend once the product is live.
We add no margin to model or API usage. During a build the usage runs on our account and appears as a separate line at the provider's price. For production we set the keys up in your account from the start, so the bill and the relationship are yours. The quote includes an estimate of the monthly usage, computed the same way the model computes it: users, requests per user, tokens per request, and the provider's published price.
Question 4 of 5
Who owns the code, prompts, evaluation sets and model configuration this work produces?
Why it mattersIn an AI product the prompts, test sets and configuration are a large part of what makes it work. If they stay with the supplier, changing supplier means rebuilding them.
Our answer
You do, from the first invoice. It lives in your repositories and your accounts while we work, not ours.
We work in repositories and cloud accounts you own and invite us into. Prompts, evaluation sets and model configuration are files in that repository, versioned with the code. When an engagement ends there is nothing to hand over, because nothing was held back.
Question 5 of 5
What happens to the price if the scope is delivered early?
Why it mattersUnder fixed price, early delivery is the supplier's margin unless the contract says otherwise. With AI cutting build time, this is where most of the speed-up goes.
Our answer
Unused hours on a fixed-price phase are credited. You choose whether the credit comes off the final invoice or rolls into the next phase.
Each fixed-price phase is quoted as hours multiplied by a stated day rate. We record the hours we work. If a phase closes with hours unused, the difference is a credit at the same rate. On capped time and materials you pay for hours worked, so the saving is yours automatically.
Ask us the same five
On a first call, put these questions to us as you would to anyone else. The answers above are how we work today, and the quote you receive should let you check each one. If something on it does not match this page, ask us why.
Further reading: how agencies price app development in the AI era (the page these questions come from), and our essay on who keeps the AI speed-up.
Start a conversation
Send us a scope, or a link to your numbers in the model. We reply with questions or a quote.