1
Frame
Interviews, data spot-checks and use-case scoring. We agree the problem, the success metric and the baseline we must beat.
2–4 weeks · Decision: build a prototype?How we work
Small steps.Real evidence.No surprises.
AI projects fail when they promise too much too early. We work in short, fixed-scope phases. Each one ends with something you can test and a clear decision about whether to go further.
The process
Five stages, with a decision point after each.
You can stop after any stage and keep everything we’ve produced.
2
Prototype
A working version on your real data, used by a small group of real users. We measure it against the agreed metric.
4–8 weeks · Decision: take it to production?3
Build
Security, integrations, access control, monitoring and, where it’s justified, fine-tuned or custom models.
Scoped from the prototype · Decision: launch?4
Launch
A staged rollout with training, guides and a feedback loop. We watch real usage closely in the first weeks.
2–4 weeks · Decision: scale up?5
Run and improve
Ongoing monitoring, regular re-evaluation, model updates and new features, run by us or handed to your team.
Ongoing · Reviewed quarterlyEngagement models
Four ways to work with us.
Every phase is scoped and quoted before it starts, so you know the cost of each decision before you make it.
Discovery and advisory
Strategy, readiness and governance work with defined deliverables and a fixed price.
Best for: getting started, board approval, compliance
Prototype to production
A project in fixed-scope phases, with a go or no-go decision at the end of each.
Best for: a specific solution or custom model
Managed AI service
We host, monitor and improve your AI solutions under a service agreement with agreed response times.
Best for: teams without in-house AI engineers
Team augmentation
Our engineers join your team for a set period, building alongside your people and sharing what they know as they go.
Best for: building in-house capability fast
Working together
What we’ll ask of you, and what you can expect from us.
From you
- A named ownerOne person who can make decisions and unblock access. It’s the most important factor in whether a project succeeds.
- Access to the right peopleA few hours a week from the experts who do the work today. They define what “good” looks like.
- Secure access to dataWe’ll agree the minimum data needed and how we access it before we touch anything.
From us
- Weekly progress you can seeA short demo or written update every week, showing working software rather than status reports.
- Honest numbersEvaluation results shared in full, including bad news, as soon as we have them.
- Everything documentedCode, models, decisions and runbooks handed over, so you’re never dependent on us.
Questions
Process FAQs
How do you price projects?
Each phase is scoped and quoted up front with a clear list of deliverables. Later phases are estimated after the earlier ones, when there’s real evidence to estimate from.
What if the prototype shows AI isn’t the answer?
That’s a successful outcome: you’ve found out for the cost of a prototype rather than a full build. You keep everything we produced, including the analysis of why it didn’t work.
Can you work with our existing suppliers?
Yes. We regularly work alongside IT partners, software vendors and internal teams, and we agree responsibilities at the start.
Do you sign NDAs and data processing agreements?
Yes, before any data is shared. We’ll act as your data processor under a UK GDPR-compliant agreement.