unthinkable
How we work and price

Every engagement starts from the number you want to move.

However you choose to pay, we agree the business objective first, report against it in every demo and tell you when work isn't moving it. Most clients pay by time and materials; where the gain can be measured, part of the fee can be linked to the result.

How we start

How every engagement starts.

The same five steps, whichever way you pay.

1The number

We pick the moment that matters most (where customers wait, give up or chase) and write down today's figure.

2The measure

We agree how the gain is measured, from which system, and what counts as done.

3A dated plan

After a short discovery you get a plan with dates for the first release. How long depends on what you are building and what already exists; with AI the range is wide, so we don't quote a typical number.

4Working software every week

From the first week you see working software, every week, and the first moment goes live measured against the starting figure.

5The next moment

Once the result holds, we move to the next moment, or hand over to your team.

Ways to pay

Choose how you pay.

Four ways to work with us, in the order most clients choose them. Each one stays tied to your business objective.

Time & materials

A senior AI team billed for time, within an agreed monthly budget, with sprint goals and a weekly demo.

Even on time and materials we never lose sight of your objective: every sprint plan starts from it, every report shows the number next to the hours, and we tell you when work isn't moving it.

Best whenScope will change as you learn; most engagements work this way.

One fixed outcome

One moment, one agreed result, one price. Acceptance criteria, including the measure, are written before we start.

Done means the agreed result is reached and accepted, not that hours are used up.

Best whenA defined first step, or to see us work before committing further.

Dedicated team

A complete senior AI team for a monthly fee, working as part of yours.

Reviewed each quarter on what shipped and what moved, not on hours.

Best whenA continuous roadmap you want to run yourself.

Linked to the result

Where the gain can be measured from your own systems: a reduced implementation fee plus a percentage of the expected gain.

Our fee moves with your number, so we only do well when you do.

Best whenThe number is clear, measurable and mostly within the software's reach. We'll tell you when it isn't.

Stages

What "done" means at each stage.

AI makes a first version fast. We name what each stage includes, so a demo is never mistaken for production software.

StageWhat "done" meansWhat it isn't yetWhen
Working on test dataThe moment works on your test data, for the main paths, and the measure is agreed.Not yet secured for real data; edge cases not handled.In the plan
First moment liveReal data and edge cases, evaluation thresholds met, security reviewed, monitored and documented; measured against the starting figure.Not yet tuned by months of real use.In the plan
Measured and extendedThe result holds; the next moment is chosen from what the numbers show.Never a big-bang rebuild.Month by month
Run and improveKept accurate, safe and affordable as data, usage and models change.Never "finished"; agents need care like any system.Ongoing
Always included

Whichever way you pay.

Your number, in every report

The objective agreed at the start, reported next to the work in every demo and monthly report.

A senior AI lead

One person accountable from week one, with a weekly demo of working software.

Quality is part of done

Tests, evaluations, security and documentation are reserved in every sprint and reported openly, never sold as extras.

Yours to keep

You own the code, IP, data, prompts and evaluations. We never train models on your data.

Problems raised early

Raised the day we see them, including when work isn't moving the number.

Straight answers

Questions about working with us.

How much does it cost to build an AI agent or an AI feature?

It depends on how many systems it works in and how exact its rules must be. A first moment that does one job in one system is usually a fixed outcome; larger or changing work is time and materials within a monthly budget. We give you a written estimate after a short discovery.

If we pay by time and materials, how do we know the work is moving our business?

Because the objective is written down before the first sprint and reported next to the hours in every demo and monthly report. If a piece of work isn't moving it, we say so and propose what to change.

How does a fee linked to the result work?

Where the gain can be measured from your own systems, we agree the starting figure, the measure and the period up front. You pay a reduced implementation fee, plus a percentage of the expected gain. We offer it only where the number is clear and mostly within the software's reach.

Can we start one way and switch later?

Yes. Many clients start with one fixed outcome, continue on time and materials or with a dedicated team, and link part of the fee to the result once a number is proven.

Who owns the code and the data?

You do, whichever way you pay: code, prompts, evaluation sets, data and runbooks live in your accounts from the start.

What drives the cost of an AI project?

Mostly the number of systems the software has to work in and how exact its rules must be, not the AI model. Each integration, approval step, compliance review and evaluation set adds work, so a moment that reads and writes in one system costs far less than one spread across five. Model running costs are measured separately, per transaction. Unthinkable shows which of these drive your estimate, so you can choose what to start with.

What does AI cost to run once it is live?

It depends on how often the software calls a model and which model it uses, so we measure it rather than guess. During the first release, Unthinkable meters every agent action with its model, tokens and cost, and reports it per order, claim or conversation. Caching, smaller models for simple steps and limits per user keep it down.

How is a dedicated team different from staff augmentation?

A dedicated team is a complete senior AI team for a monthly fee, with a senior AI lead accountable for what ships; staff augmentation adds individual developers you manage yourself. Unthinkable's dedicated teams work as part of yours, demo working software every week and are reviewed each quarter on what shipped and what moved, not on hours.

What does support after launch cost?

Support after launch is priced the same way as the build: time and materials within a monthly budget, or a dedicated team for a monthly fee. It covers keeping the software accurate, safe and affordable as your data, usage and models change, including rerunning evaluations and fixing what they catch. If your own team takes over instead, we hand over the code, runbooks and evaluation sets and stay until they run it without us.

Ask Unthinkable

Which number would you move first?

Tell us the objective and where your customers wait today. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.

Send us a note instead

Required

Thanks. A senior AI engineer will reply to you.
u
Ask UnthinkableClarity first, then the right team
Your conversation

An AI agent. This prototype uses prepared answers and sends nothing.