Working services for organisations that can’t afford to get it wrong.
We take high-stakes ideas and problems to a working, validated service — fast, using agentic AI, with your sensitive data never leaving an approved boundary. A small, senior team, carrying the thread end to end.
The problem we remove
In a large organisation, this work is split across four roles — a product manager, a service designer, a business analyst and a solutions architect. The handoffs between them leak time, money and meaning. Every boundary is a place for the intent to get lost.
We’re a small, senior team that carries the thread end to end — without the handoff tax of a big supplier or systems integrator. One team holds the whole problem, from the first conversation to a service you can put in front of real users.
How we work
A way of thinking, not a fixed methodology. We adapt it to the problem in front of us, but it usually runs like this:
-
01
Understand the real needs
We start with the people doing the work, the constraints they operate under, and the lines they can’t cross.
-
02
Shape the problem
We turn it into stories, service blueprints, high-level architecture and PRDs — so the intent is explicit and survives a committee.
-
03
Build a living prototype
Not slideware. A working, evolving prototype you can use, test and argue with.
-
04
Prove it with real users
We validate it with the people who’ll actually use it, in conditions close to the real thing.
-
05
Hand over a working service, with a roadmap
You’re left with something that works up to a defined scale, plus an honest roadmap of what it takes to grow it further.
AI-native, security-first
We use agentic AI to compress what used to take a team months. That speed is only useful if it never costs you control of your data — so the two halves are welded together.
Your sensitive data never leaves an approved environment. We work either with your own approved enterprise AI inside your boundary (for example, Claude via Copilot), or with open-weight models — DeepSeek, Qwen, Gemma — that we run locally on our own controlled hardware. Either way, the boundary holds.
Selected work
All work →Government
FCDO — secure collaboration across boundaries
Helping civil servants communicate, collaborate and build workflows around sensitive information, often across security boundaries.
Read more →Public safety
Offmon — public-safety mobile applications
AI-powered mobile applications helping UK law enforcement monitor high-risk individuals in the community, to prevent reoffending and keep people safe.
Read more →What this is (and isn’t)
What this is
We take you from fog to a validated, working service — usable up to a defined scale — plus an honest roadmap for taking it further. You get clarity, a thing that works, and a credible plan.
What this isn’t
- Not a full delivery house.
- Not multi-year staff augmentation.
- Not an enterprise systems integrator.
Who you’ll work with
A small, senior team led by Tom Inglis. You work with the people doing the work — not an account manager and a rotating cast behind them. You can engage us through the DOS7 framework or outside IR35.
More about us →Common questions
How are you different from a big consultancy or SI?
We’re a small, senior team. You get the people doing the work, not layers of account management — and none of the handoff tax that comes with a large supplier.
How do you use AI without exposing our data?
Either your own approved enterprise AI inside your boundary (e.g. Claude via Copilot), or open-weight models — DeepSeek, Qwen, Gemma — we run locally on our own controlled hardware. Your sensitive data never leaves an approved environment.
Can you work inside our secure environment, or across security boundaries?
Yes. We’ve done exactly that in government, building workflows for sensitive information across security boundaries.
Can we engage you outside IR35, or through a framework?
Yes — through DOS7 (RM1043.9) or outside IR35.
What do we actually get at the end of an engagement?
A working, validated service that’s usable up to a defined scale, plus a clear roadmap of the product, technical and service work needed to take it further.
How long does this take?
It depends on scope, but our AI-native approach compresses what traditionally takes a team months.
How do you price it?
£600 a day plus VAT, outside IR35 or via DOS7. Defined pieces of work can be scoped to a fixed fee, agreed up front.
Occasional notes, worth reading
Notes on secure, AI-native service design. Infrequent, no noise.