AI MVP Sprint

One AI product idea taken from a blank repository to something real users can use, in six weeks, on your cloud account.

$25k to $40k6 weeks

Price

$25k to $40k

Duration

6 weeks

How it is priced

Fixed scope, fixed fee. The range moves on how many external systems the product has to read from and whether it has to run inside your own cloud tenancy from day one.

Who is on it

A delivery lead who writes code, senior product engineers on the application, and an engineer on retrieval and evaluation.

Written for

Founders at seed or pre-seed, and innovation teams inside larger companies who need a working system rather than another deck.

Why this engagement exists

Six weeks is enough to answer the only question that matters at this stage: does this work on real data, with real users, at a cost you can live with. It is not enough to build everything, so we do not pretend otherwise. We pick the one workflow with the sharpest value, build it end to end including the unglamorous parts, and ship it behind auth to a named group of users.

What is in scope

Week one: problem framing, a written scope with what is explicitly out, and a labelled evaluation set built from your real examples before any prompt is written
A thin vertical slice of the product in week two: one path, end to end, deployed, so the integration risk surfaces first and not last
Retrieval and context design over your documents or records, including chunking, metadata and the query path
The application itself: auth, data model, the interface a real user touches, and the admin view you need to see what the system did
An evaluation harness that runs in CI, with the pass threshold agreed in writing in week one
Deployment to your cloud account with infrastructure defined as code, plus logging, tracing and a cost per task readout
A demo every Friday against the evaluation set, not a status update

What you hold at the end

A running product on your infrastructure, under your domain, with your users in it
The full source in your repository, with commit history, from the first commit
The evaluation set and harness, so you can tell whether a change made the system better or worse after we leave
Architecture notes, a runbook for the two or three ways it fails, and the cost model per task
A written list of what we cut and what it would take to add it

What this is not

Stated here rather than discovered in month three. Anything on this list can be scoped separately, and we will say what it would take.

A security certification. We build to your controls and document the data flows, and we will sit in your security review, but a sprint does not produce an audit report
Bespoke model training. This engagement assembles and evaluates around available models, and says so plainly if the problem genuinely needs a trained one
Mobile applications on both stores. One responsive web application is what six weeks buys
An ongoing on-call rotation. Support after handover is the Embedded AI Pod

How the work runs

Scoping call and written brief

One call with the people who know the problem, then a written brief back to you within two working days: what we understood, what we think the hard part is, what is explicitly out of scope, and which of the four engagements fits. If none of them fits we say that instead of reshaping your problem to match a price list.

Output: A written brief and a named engagement, or an honest no

Evaluation before implementation

Before any prompt is written we build a labelled set from your real examples with your domain experts, split by category, with a held back portion. The pass threshold goes in writing, with the consequence of missing it agreed at the same time. This is the week that makes every later argument about quality a number rather than an opinion.

Output: A labelled evaluation set, a harness, and an agreed threshold

A thin slice, deployed

One path through the system, end to end, running in your cloud account, usually inside the second week. Not a prototype on a laptop. Deploying early is how integration risk, credential problems and data access surprises surface while there is still time to change the plan.

Output: A deployed slice on your infrastructure and a working pipeline

Weekly build, demonstrated

Every week ends with a working demonstration against the evaluation set rather than a status document. You see the score, the failures, and what we are doing about them. Anything at risk is raised in the week it becomes at risk, not in the week it becomes a problem.

Output: Shipped increments, a score per week, and a visible risk list

Hardening

Guardrails on the output path, escalation payloads, tracing, cost and latency instrumentation, alerts wired to people who can act, and runbooks for the two or three ways this specific system fails. Where the system can act, it shadow runs against live traffic before it is granted authority.

Output: Guardrails, dashboards, alerts, runbooks and shadow run results

Handover, and then out

Your engineers make the last change while we are still there to watch. Repository, infrastructure as code, evaluation suite, documentation and access all sit with you, and none of it depends on an account we control. If you want us to stay, that is an Embedded AI Pod with its own scope, not a dependency we engineered into the build.

Output: Code, IP, docs, evaluation coverage and a team that can run it

Questions we get asked

Six weeks including or excluding the discovery?

Including. Week one is discovery and it produces the written scope and the evaluation set, both of which are deliverables. If discovery shows the problem is not a six week problem, we say so in week one and you pay for week one.

What happens if the evaluation threshold is not met?

We agree the threshold in week one and we report against it every Friday, so a miss is visible in week three rather than week six. If the gap is a data problem we tell you what data would close it. If the approach is wrong we say that too. What we will not do is ship a demo that passes on curated examples and fails on yours.

Whose cloud account does it run on?

Yours, from the first deploy. We work inside your tenancy with credentials you issue and can revoke. Nothing depends on an account we control, so there is no migration at the end because there is nothing to migrate.

Do we own the code?

Yes, in full. IP assignment is in the contract, the repository is yours from the first commit, and we do not retain a licence to reuse anything built for you.

engineering / the other three

If this is not the shape you need.

02AI Integration PodAI features built into a product that already has customers, without destabilising the thing they are already paying for.$60k to $120k · 8 to 12 weeks 03Agentic Workflow BuildA multi step workflow that currently runs on people, rebuilt as a system that acts, checks itself, and hands off cleanly when it should not act.$150k to $300k · 12 to 16 weeks 04Embedded AI PodA standing engineering capability inside your team, for the year after the build when the interesting problems start.$18k to $32k per month · Three month minimum, then rolling

Bring the task, the data and what has already been tried.

You get a written view of the problem within one working day, from an engineer, including the version where the answer is that this is not a build.

Start a build

Why this engagement exists

What is in scope

Week one: problem framing, a written scope with what is explicitly out, and a labelled evaluation set built from your real examples before any prompt is written

A thin vertical slice of the product in week two: one path, end to end, deployed, so the integration risk surfaces first and not last

Retrieval and context design over your documents or records, including chunking, metadata and the query path

The application itself: auth, data model, the interface a real user touches, and the admin view you need to see what the system did

An evaluation harness that runs in CI, with the pass threshold agreed in writing in week one

Deployment to your cloud account with infrastructure defined as code, plus logging, tracing and a cost per task readout

A demo every Friday against the evaluation set, not a status update

What you hold at the end

A running product on your infrastructure, under your domain, with your users in it

The full source in your repository, with commit history, from the first commit

The evaluation set and harness, so you can tell whether a change made the system better or worse after we leave

Architecture notes, a runbook for the two or three ways it fails, and the cost model per task

A written list of what we cut and what it would take to add it

What this is not

Stated here rather than discovered in month three. Anything on this list can be scoped separately, and we will say what it would take.

A security certification. We build to your controls and document the data flows, and we will sit in your security review, but a sprint does not produce an audit report

Bespoke model training. This engagement assembles and evaluates around available models, and says so plainly if the problem genuinely needs a trained one

Mobile applications on both stores. One responsive web application is what six weeks buys

An ongoing on-call rotation. Support after handover is the Embedded AI Pod

How the work runs

Scoping call and written brief

Output: A written brief and a named engagement, or an honest no

Evaluation before implementation

Output: A labelled evaluation set, a harness, and an agreed threshold

A thin slice, deployed

Output: A deployed slice on your infrastructure and a working pipeline

Weekly build, demonstrated

Output: Shipped increments, a score per week, and a visible risk list

Hardening

Output: Guardrails, dashboards, alerts, runbooks and shadow run results

Handover, and then out

Output: Code, IP, docs, evaluation coverage and a team that can run it

Questions we get asked

Six weeks including or excluding the discovery?

What happens if the evaluation threshold is not met?

Whose cloud account does it run on?

Do we own the code?

Yes, in full. IP assignment is in the contract, the repository is yours from the first commit, and we do not retain a licence to reuse anything built for you.