AI + automation · Consultancy · 2026 · In-house

AI guard rails

Governance that every use of AI runs through

Before anyone could use AI on real work, the organisation needed platforms it trusted and controls nobody could route around. We turned its AI guard rails into working technical controls.

In-house work: our founder leads AI engineering at a multi-site UK organisation, which isn’t named here.

The brief.

People wanted to use AI, and some already were. The organisation handles sensitive personal data, so it needed approved platforms, clear rules and controls in place before AI went anywhere near it.

What we built.

Platforms chosen on evidence: An evaluation matrix covering security, data residency, cost and fit, with a recommended platform and purpose for each.
Secured before anyone signs in: Each platform set up with single sign-on, role-based access, audit logging and no training on company data, and signed off by Information Security.
Controls on sensitive data: Personal data redacted or blocked before it reaches a model, usage logged and access tiered by role. Live AI use routes through these controls, designed with the security and governance teams.
A route from idea to approval: A technical risk assessment every AI use case goes through, a register of AI use cases kept current, and the technical sections of the AI usage policy.

Where it stands.

The approved platforms are live behind these controls, and new AI use cases go through the risk assessment before they start.

How we worked

Guard rails came first because everything else depends on them: the call agent, the automations and the specialist tools all run on the platforms secured here.

Want something like this?

Tell us what you’re trying to build or fix. A few sentences is plenty; we’ll ask about the rest.

What happens next

1We read your message and reply by email.
2We talk it through on a short call.
3If we’re a good fit, you get a written plan.