Good fit if you need
- Governed AI assistance embedded in real workflows (not a chat demo or a workflow with a single LLM step)
- Clear accountability: who approved what, when, and why
- A system that can survive handover, scrutiny, and operational drift
Engage NullProof Studio
We design governed operating systems for teams that deliver — where speed matters and mistakes are expensive.
First track: Insurance & Risk (Signal Intelligence, exposure scoring, claims operations) — live pilot running on PFAS signal monitoring, results available under NDA. Professional services and founder studios follow the same model.
This is not “AI automation.” It’s an operating system.
Led by Andy Woolterton — two decades in enterprise architecture and AI strategy. Building governed agent systems for insurance and professional services.
Fit check
Is this the right shape of work for you?
Not a fit if
- You want prompt packs or a chatbot
- You can’t assign an internal owner for decisions and escalation
- You want speed without governance (we won’t ship that)
Wherever you are with it, you'll recognise one of these.
Engagements
Three ways we engage
Each engagement compounds. Most teams start with the sprint, then move into governance setup or a build.
FAST · GOVERNED · DECISION-READY
Architecture Sprint
A blueprint you can build from — built around constraints, not optimism.
What you get
- Clarity on who does what — you, your team, and the AI. No ambiguity.
- A picture of how work flows through your operation. You’ll recognise it immediately.
- The engineering that means you never have to explain what the AI did. It’s already documented.
- A plan that says what “done” looks like — before we start, not after.
If you already have a system, we’ll start by stress-testing it: where it fails, how it lies, how it drifts, and how to instrument it.
TRUST · AUDIT · OPERATIONS
Governance Setup
The safety nets installed into your workflows — so you can move fast without looking over your shoulder.
What you get
- Clear lines: where the AI can act alone, where it needs you. Mapped to your risk profile.
- A system that checks its own work — three times, with escalation — before anything reaches production.
- A record of who approved what, and when. Before anyone asks for it.
- Simple instructions for the routine stuff, and a clear path for the moments that need a human.
- A dashboard that tells you what’s happening, not what happened.
Often includes TRUST (cryptographic identity for agents and artefacts) and MY. NULLPROOFSTUDIO (accountable lead + multiple agents + full audit trail).
PRODUCTION · ITERATIVE · OWNED
Build & Run
Ship it. Run it. Sleep at night. The engineering handles the rest.
What you get
- The operating model, built and running. Not a prototype — production.
- A rhythm for your team: what happens daily, weekly, quarterly. Everyone knows their beat.
- The engineering that catches errors, recovers from failures, and tells you before you have to ask.
- A clear path to the next workflow, the next agent, the next capability — without starting over.
Optional components: RADAR Risk Intelligence, EN-QUIRE governed editing, Custom LLM Models.
Book a 30-minute call →
Not sure which tier fits? The call decides — that's what it's for.
Underneath all of it sits governance you'll rarely think about — trust boundaries, a record of who approved what, a clear path when something needs a human. It's the reason moving fast here doesn't mean moving exposed. You won't have to become an expert in any of it. That's the point.
Optional reading
Want the version that speaks your language?
Same call, same booking link — these pages just speak to a particular shape of operation or assessment. Optional, not a required step.
Framework
Technology Capability Assessment
Evidence-based review of whether a capability creates advantage, where it is mature, and how much complexity it introduces.
Read the track →Track
Insurance & Risk
Signal Intelligence, claims operations, exposure scoring — governed signal infrastructure for teams whose risk picture changes overnight.
Read the track →Track
Professional Services
Delivery OS for firms with QA, client risk, and utilisation leakage — production agent systems inside billable workflows.
Read the track →Track
Founder / Creative Studio
One accountable lead, governed agents, creative throughput. The model we run NullProof on — transferable to your studio.
Read the track →How it works
How an engagement runs
- We talk (30 minutes). You tell us what’s keeping you up at night. We listen for the decision owner, the constraint, and the one move that would make a real difference. No pitch. No jargon. Just a conversation.
- We think (within a week). You get a written assessment — what we’d build first, why it matters, and what it would take. No obligation. Even if you never build with us, the memo is useful.
- We build together (weeks). If it makes sense, we start. A sprint, a governance setup, or a full build — scoped to your constraint, not to a template. You see progress weekly. You sign off on every consequential decision.
- You run it (ongoing). The system ships. Your team operates it — with or without us. The governance runs underneath. You don’t have to think about it, until the moment you need it. Then it’s there.
We bias toward small, governed deployments that compound. Start with one workflow. Prove it. Then expand.
FAQ
Common questions
- What do you mean by “governed”?
- It means you don’t have to worry about what the AI did while you weren’t watching. Someone already thought about what happens when it gets it wrong — the boundaries, the checks, the escalation path, the record of who approved what. You get the speed without the exposure. That’s the engineering. You won’t have to become an expert in any of it.
- Do you build on our stack or yours?
- We start with what you have — Claude, ChatGPT, Copilot, your own models. You don’t need to buy anything new to begin. The governance layer wraps around your existing tools, not replacing them. If we recommend something, it’s because your operation needs it — not because we sell it.
- Can we start small?
- Yes — and most teams do. A 3–5 day Architecture Sprint gives you a blueprint you can build from, without committing to a build. You’ll know what it would take, what it would cost, and what “done” looks like — before you decide whether to go further. No obligation, no pressure. The sprint is useful even if you never build with us.
- Will you work under NDA?
- Yes. Standard practice — we can sign yours, or provide a mutual NDA. We work with regulated operations and sensitive data. Discretion is built in, not bolted on.
- What does it cost?
- Architecture Sprints from £8,500. Governance Setup from £24,000. Build & Run from £16,000/month (initial deploy), then from £4,500/month to run. Sprint fees are credited in full against a build. Prices exclude VAT. We scope to your constraint, not to a template. If you’re not sure which tier fits, that’s exactly what the discovery call is for — we’ll tell you in 30 minutes.
- Who owns the code and the system?
- You do. Code, configs, playbooks, governance specs — all yours. No lock-in, no proprietary runtime, no “platform fee” to keep your own system running. We work on your infrastructure where possible. The engineering is yours to keep.
- Can we bring it in-house after the build?
- Yes — that’s the design intent. The operating model, governance specs, and playbooks are documentation your team can run with or without us. We hand over with a runbook and a 30-day support window. If you want us to stay, great. If you want to take the keys, that’s fine too. The engineering doesn’t need us to keep working.
- Will you work with our existing AI tools?
- Yes. We start with what you have — Claude.ai, ChatGPT, Copilot, your own models. The governance layer wraps around your existing tools, not replacing them. If you’re already using something that works, we don’t break it. We make it safe to trust at scale.
No form to pass. Just a conversation.
Not sure what the right first project is yet?
That's the normal place to start — and finding the one move with real business impact is part of what we do before we build anything. Tell us as much or as little as you like. Andy reads every one.
or
The calendar isn't wired up yet — leave a note below instead and Andy will reply with times.
Got it. Andy will read this himself and reply — usually within a day.