Every business now runs on AI.
Few are in control of it.
GenWA™ gives you visibility, cost control and evidence over every AI tool your people use. Then it puts AI to work, safely: as advice, as an all-in-one solution, or as an agent organisation we incubate and oversee for you.
of employees have used AI tools their employer never approved.
of employers have an AI policy at all.
of leaders can only partly see what AI costs them.
of businesses using AI have security practices that cover it.
BoardSeat. A trusted AI in the room.
A physical AI board member you place in the room where you want extra intelligence and governance. It listens and talks, challenges where it adds value, projects the discussion onto the wall and keeps the meeting on track. You choose when it is on, and what it records and leaves out. Your team makes the decisions.
- Sits locally and is completely configurable: tuned to your business, market and regulators.
- Designed to speed up decision making, with a design target of ten times faster.
- Not a black box: a glass window shows what is inside the crafted case.
Start at the right place for your business.
See what good looks like, what to avoid, and three ways to begin, from a training session to a full plan.
Start hereAlready using AISee it, control it, evidence itFind every tool and pound, enforce your rules, and keep a record an auditor can read.
See the productsReady to build with AIThe Incubator and the AcceleratorAn agent organisation we run for you, or senior AI leadership time for equity if you are a founder-led business.
Explore the IncubatorWhat AI control means for you.
We work with businesses of every size — as long as there’s a mandate to look at AI, or the will to pilot it.
From a board that’s asked for answers to a founder who wants to test the water. Here’s roughly where businesses like yours tend to start.
No IT department. Plenty of AI. One policy, company accounts, one number a month.
See the planMid-market & scale-up250–1,000 peopleEvery department bought its own AI. One view, one control point, a smaller bill.
See the planEnterprise1,000+ peopleAgents in production. Evidence for auditors, insurers and regulators.
See the planAdvice, products, or an organisation we run for you.
Most businesses start with one and add the others. All three sit on the same control layer, so nothing you do with us has to be undone. Most of the businesses we talk to are 50 to 500 people with no in-house AI team — that is who this is built for.
Independent AI audit, policy and controls, training, strategy and planning. Fixed fees, credited against year one if you go on to the products.
See advisory services02ProductsAnalyse · Guard · Evidence · Automate · GrowOne gateway, your rules, any model. Then agents on top of it.The control layer between your business and every AI lab, and a trained agent team deployed on it. Modular, priced as a range configured to your requirements, run in your cloud or ours. Built for a 50-person business as much as a 5,000-person one.
Explore the products03IncubatorOpen to any qualifying business · Paid per outcomeWe build and run an agent-first function, team or organisation for you. Open to any qualifying business, paid per outcome.A function, a team or a whole operating unit, run agent-first behind the gateway with a small human core, live in twelve weeks. You pay for verified outcomes, with a cap you set.
See the IncubatorThe right model for every job. Stop paying premium for routine work.
We know which model does each job well at the lowest cost, and we set it once. Cut your token burn. Keep your momentum.
How model selection worksThese are averages. Your business is not. The health check measures your own usage and gives you the exact figure.
Calculate your exact savingSpend: people × AI spend per person. The default of £92 a month comes from 15% of software budget going to AI (Ramp, 2026) on £7,400 software spend per employee (Zylo, 2026). Savings are applied in turn, so they never double count. Wrong model: 56% of AI spend is on premium models (Ramp, 2026), three-quarters of it routable to standard models at 85% lower cost (RouteLLM, 2024). The other three are GenWA working assumptions, not published statistics: large-context sessions, 20% of what remains, half of it avoidable; repeated documents and context, 25% of what remains, cut by 75% through caching, which providers bill at a fraction of the standard input price; duplicate tasks, 8% of what remains, 90% reusable. The total saving is capped at 70% of spend. Your health check replaces every assumption with your numbers.
AI at work, inside harnesses you set.
Every request your people and agents make runs through GenWA’s harnesses: your rules, applied at the gateway before anything reaches a model. What breaks a rule is held, recorded and reported to a named owner. The rules map to the regulation your market answers to, such as FCA Consumer Duty and SM&CR in financial services, UK GDPR for personal data, and the EU AI Act if you serve customers in Europe.
One number for the board. Four things behind it.
The AI Effectiveness Index scores adoption, control, efficiency and return, against UK benchmarks. It is the first thing the health check gives you, and Usage Insight keeps it current every month.
Get your indexExample figures. Yours are calculated from your own answers and data.
The news, from our side of the gateway.
Amodei, Altman, Hassabis and Musk called for a slower AI frontier this week. What it means for a UK business that already runs on AI, and what to do before the next model lands.
ReadWhite paperThe AI Control GapWhy policy without enforcement fails, and what a control layer looks like for a UK mid-market business.
DownloadWhite paperThe Hidden Cost of AIToken consumption, premium-by-default, unused seats and duplicate tools: the evidence, and the arithmetic of getting it back.
DownloadThe questions boards ask us.
Why do I need GenWA? I can do this myself with AI.
You can use AI without us. What you cannot do with ChatGPT or Copilot is see every tool your people use, enforce your policy on each request, choose the cheapest adequate model for each task, and produce a record an auditor can read. Those are control functions, and vendors have no incentive to provide them: their revenue is your token consumption. GenWA sits between you and every vendor, paid to reduce what you spend and to prove what you did.
Can't our IT team build this?
Some of it, over a year, with people you would rather have on other things. Our founders built these controls inside regulated banks; the Gateway, Ledger and evidence packs are the version we now set up for clients in their own cloud account. We agree the timeline with you before we start.
What happens when a better model launches?
Your policies, prompts, agents and records are attached to GenWA, not to a vendor. We test the new model in our sandbox, stage the switch under Change Control, and your people keep the same interface.
Will this slow my people down?
No. People keep the tools they already use, and nothing gets slower. 28% of unapproved use happens because there was no approved alternative (Microsoft UK, 2025, vendor survey); managed accounts and an approved model for each task remove that reason.
Where does our data go?
Nowhere new. Everything runs inside your own cloud account; nothing routes through GenWA's infrastructure. Sensitive data is removed before a request leaves your environment.
Are you regulated or certified?
No, and we do not claim to be. Our founders built and ran FCA-regulated businesses. GenWA's controls are mapped to ISO/IEC 42001, UK GDPR, SOC 2 and Cyber Essentials, and we prepare your evidence for certification; certification is issued by accredited bodies.
Find out where your AI stands. Then decide.
First findings within 48 hours of the health check · Audit fee credited against year one