How to take an AI pilot into production in a regulated business
Build the evidence for a production decision.
Bring risk and compliance into pilot design, agree decision ownership, and keep evaluation and human-review records as you build. Tenhaw’s 90-day programme takes one agreed workflow into production and use. This is not legal advice.
30 minutes with James · how this would work in your organisation
The basis for this approach
This is our approach to production readiness. Tenhaw’s published regulated-sector examples are proofs of concept: PPL and the AI Voice Insights proof of concept James Rooney led in his own role inside HSBC. Neither is evidence of a Tenhaw production deployment. This is delivery guidance, not legal advice. Your legal owner determines the duties and classification for each system.
The demand signal
The published PPL proof of concept worked on real data in the client’s estate. Nothing moved into production and no regulatory or audit position changed. That distinction explains why a working pilot and a production decision need different evidence.
Bring the reason this is on your agenda to a 30-minute call with James, and we can explore what it would take in your organisation.
Book a 30-minute callWhy it stalls
2 failure modes we keep meeting
The pilot has no agreed release test
A demonstration does not establish a baseline, acceptable errors or the consequences of a wrong action. Agree those with the people who will carry the production decision.
The reviewers arrive after the build
Risk and compliance need to help design ownership and control points while choices can still change. Evidence reconstructed after the build leaves the production decision waiting.
If one of these sounds familiar, bring it to a 30-minute call with James and we can explore a way past it.
Book a 30-minute callHow we approach it
3 moves, in order
- 01
Name the system and the decision owner
Record the purpose, source access and accountable person in a proportionate AI inventory. Work with your legal and risk owners on classification and the human-approval boundary. This is delivery guidance, not legal advice. Your legal owner determines the duties and classification for each system.
- 02
Test the whole workflow
Use agreed examples, record failures as well as passes, and inspect the actions and tools behind the final answer. Your own assurance owners should hold the release gate.
- 03
Price production separately
The 90-day programme agrees one workflow and its pound value first. Weeks 1–2 cover design, weeks 3–10 build, and weeks 11–13 adoption, measurement and handover. It is £120,000 or £180,000 fixed; the production decision remains evidence-based.
Your legal owner determines the duties and classification for each system.
Prefer to talk it through? Ask us on a discovery call →
Sources
Every source below was opened and read before it was attached. Where nothing survived that check, the claim on this page was softened rather than given a plausible-looking link.
Tenhaw, AI governance and regulatory evidencePPL, the two-week working proof of conceptOne greenfield build paired with one client engineer. Nothing moved into production and no regulatory or audit position changed.Tenhaw, 90-day outcome programmeEuropean Commission, AI Act framework and application timelinePrimary overview checked 5 November 2026, with a link to the final AI Omnibus text and current high-risk dates.The closest guides to this one
Nearest first, then the rest. Each one carries the same label: written from delivery, or the method we would bring. All 34 are on the hub.
How do we verify an AI agent beyond a vendor demo?
Give your release decision evidence the business can inspect.
Our approachWhat should an AI governance framework include?
Keep the AI inventory, accountable decisions and evidence connected.
Our approachAgent identity and access
Give each workflow the authority it needs, with a way to take it away.
DeliveredWhat does an AI-native software delivery lifecycle look like?
Connect requirements, AI engineering and human release decisions.
DeliveredTarget operating model for an AI-native organisation
Design what agents do, what people decide and how the work improves.
DeliveredEnd-to-end agentic workflow implementation
Design the whole process, from arrival to a recorded result.
Or bring the problem to a call instead of reading three more of these.
Book a 30-minute callWork out the next step for your team.
A 30-minute call with James Rooney to discuss your workflow, people and budget. Leave with a rough scope and a useful next step.
90-day outcome programme · £120k or £180k fixed · 90 days · excluding VAT
No time that suits? Email hello@tenhaw.com. Calendar not loading? Open it on cal.com.
Describe the work
Describe the work in a paragraph. James replies with a rough scope and budget range within two working days.
Questions this guide answers
Why do regulated AI pilots stall?
A pilot may work while leaving the production owner, evaluation threshold, review boundary or run budget unresolved. Identify which gap is blocking the decision before commissioning another demonstration.
What evidence does production need?
Start with an inventory entry, named decision ownership, evaluation results, approval records, change history and the design for human review. Your legal and risk owners determine the duties for the specific system. This is not legal advice.
When should risk and compliance get involved?
During workflow and operating-model design, so they help define the control points and evidence. Bringing them in only to review a finished system makes changes harder.
Does the EU AI Act apply?
It can apply to UK firms placing systems on the EU market. Ask your legal owner to establish the system’s classification, your organisation’s role and the duties that follow using the current official sources. This is not legal advice.
How long does pilot-to-production take?
Tenhaw’s published production engagement runs for 90 days against an agreed workflow and value, at £120,000 or £180,000 fixed. Scope, access, evidence and responsibilities must be agreed; the duration is not a guarantee that every existing pilot can meet a release gate.
What if our pilot used synthetic data?
Treat that as a limit on what the pilot proves. Test the agreed workflow against real examples and permissions in your tenancy before relying on the result for a production decision. Tenhaw’s proof of concept uses the client’s data and includes a separately scoped assessment of production readiness.