AssuranceGovernance frameworks and audits
ISO 42001-aligned frameworks, EU AI Act readiness, automated decision-making audits and AI use policies. Written so a board can actually approve them and an auditor can actually test them.
Nimble AI, AI governance and service management
Your AI is already making decisions. We make them defensible.
We are a European AI governance and assurance practice that also ships AI products, so the people writing down how your systems work have built the same kind of thing themselves.
Run the free instant AI exposure assessment See how engagements run
The case file · Ashworth Group Ltd
She left in February. Her account approved £18,650 in May.
Samantha Jane Archer left on Friday 27 February 2026. Seventy-four days later her surviving approver right in the ERP passed invoice INV-208441, £18,650 to Kestrel Fleet Services Ltd. Three systems held three keys for her, and not one of them ever joined up.
Walk through the case of Sam Archer Seven questions, asked of one leaver’s records
PAY-2000 payroll 0047812 
PeopleFirst HR SA-1178 
Ledgerline ERP s.archer2 Three keys, one person, no join.
Sam Archer is a fictional composite; the portraits are AI-generated images of a person who does not exist.
What it read, and the phrases that fired rules
Nothing running yet. Describe your use of AI on the left, or pick one of the examples, and this panel shows every step of the classification as it happens.
Ambermoderate exposure
Read this as a starting point, not an answer. This is a first-pass indication produced by a keyword classifier running in your browser. It has read one or two sentences. A real assessment looks at the system, the data it touches and the decisions it affects, and only that can tell you where you actually stand. It is not legal advice.
Obligations that bite
Recommended next step See what this involves
This ran entirely in your browser. Nothing was sent anywhere, nothing was stored, and no model was called.
The productivity is real. So is the exposure gap.
Boards are already relying on it. Most haven't written down how.
Four numbers that explain why governance is now the boring, necessary part of getting value from AI.
75%
of UK AI adopters report improved workforce productivity
DSIT AI Adoption Research, 2026
48%
of AI-adopting firms report higher profits in the past year
Lloyds Business Barometer, 2026
61%
of businesses have no fully implemented internal AI usage policy
techUK / Shoosmiths, 2025
10
governance dimensions we score you against, free
Nimble AI Governance Framework
What we actually do
The same team that audits AI also ships it.
Most governance firms have never put a model into production. We run consumer AI products of our own, with real users, real costs and real failure modes. That is why our audits ask the awkward questions.
EngineeringAgentic AI, with oversight built in
We build agents that take real actions in real systems, with the scope limits, human approval points, logging and audit trail designed in from the first commit rather than bolted on after the pilot goes wrong.
Our own productsLive AI products we run ourselves
Show Me My Book turns a manuscript into an explorable canon bible. Bring My Book To Life builds annotated visual walkthroughs for authors. Both are live, both are ours, both cost us money every time a model runs. That is a very effective teacher.
We ship AI products with real users and real running costs. That is what makes the audits worth having.
How engagements run
A ladder, not a cliff.
Every step is proportionate and priced up front. Start free, stop whenever it stops being useful, and never buy the big framework before you know whether you need it.
Step 01AI Readiness Scorecard
Ten questions, scored against ten governance dimensions, with a written result you can take to a board meeting.
Step 02AI Exposure and Scope Check
A live conversation that establishes which of your AI uses are actually regulated and which are not, so you stop worrying about the wrong ones.
Step 03AI Governance Health Check
A structured maturity assessment across ten governance dimensions, grounded in ISO/IEC 42001, the NIST AI Risk Management Framework and the EU AI Act, with a prioritised action plan.
Step 04EU AI Act Rapid Audit
Classification of each system against the Act, the obligations that follow, and the evidence you will need if a customer or regulator asks.
Step 05Automated Decision-Making Audit
For systems that make or materially shape decisions about people: lawful basis, meaningful human review, explanation and route to challenge.
Step 06ISO 42001-Aligned Framework
The full management system: policy, roles, risk assessment, controls and review cycle, aligned to the standard and sized to your organisation.
Add onAI Use Policy Pack
A tailored AI use policy, a one-page staff dos-and-don'ts reference and a short risk note, handed over on a 30-minute call.
BuildAgentic AI Implementation
We design and build the agent, wire in the oversight and audit trail, and hand over something your own team can run.
Reach
European rules follow your customers, not your postcode.
The EU AI Act can bind any business that places an AI system on the EU market, and UK and EU GDPR can reach any controller or processor that targets or monitors people in the UK or the EU, wherever that business is established. Sell into Europe, and this can be your compliance problem too.
Delivery
Remote-first, because it costs you less.
No travel, no expenses, no days lost to transit: engagements run by video call and shared document as standard, at the same fixed price. On site is available too, and we source a local specialist whenever a job genuinely needs one.
Find out where you stand before somebody else asks.
Start with the free scorecard Ten minutes, no sales call attached