AI Audit

Already running AI. Is it actually working?

Most organisations investing in AI do not have a clear view of what each initiative is actually delivering, what it is costing, and whether the trajectory is improving or declining. We give you that view, independently, across every initiative in scope.

The audit lenses

12 lenses. Every initiative.

Every AI initiative in scope is assessed across 12 lenses, from the strategic case all the way down to model health and codebase quality. Nothing is assumed. Nothing is taken on faith.

01

Vision & purpose

Why does this initiative exist? Is that still the right problem?

02

Intended outcomes vs reality

What was it supposed to achieve? What is it actually delivering?

03

Value to date & burn rate

What has been spent? What value has been realised? Is the trajectory improving?

04

Future value potential

If continued, what is the realistic ceiling? Is it worth the investment?

05

Tech stack & architecture

Sound infrastructure? Vendor lock-in risks? Scalability constraints?

06

AI codebase & model health

Are models drifting? Is there test coverage and monitoring?

07

Data quality & pipeline integrity

Is the initiative receiving clean, trustworthy data?

08

Adoption & frontline trust

Are the people who should use this actually using it?

09

Team capability & ownership

Who owns this? Can they evolve it? Single point of failure?

10

Dependencies & integration health

What does this depend on? How fragile are those dependencies?

11

Governance & compliance posture

Audit trail, explainability, oversight in place where required?

12

Strategic fit

Does this still align with where the business is going?

Why there is no fixed price

Every audit is scoped from scratch.

No two audits are the same. The scope depends on how many initiatives need to be reviewed, how deep the technical assessment needs to go, and what decision the audit is designed to inform. A board rationalising a 20-initiative portfolio needs something different from a product team trying to understand why one system is underperforming. The first conversation is about scoping, not selling.

Scope

Company-wide portfolioBusiness unitSingle productSingle initiative

Number of initiatives

12–56–1515+

Technical depth

Strategy and outcomes onlyStrategy + product layerFull stack

Urgency

Board or investor deadlineBudget cycle decisionTeam capacity reviewOngoing optimisation

We scope every audit after a 30-minute discovery call. No assumptions, no generic packages.

What you get

A decision. Not a report.

An initiative-by-initiative scorecard

Every initiative assessed across all 12 lenses. Clear verdict: accelerate, fix, pause, or stop.

A ranked portfolio view

All initiatives stacked by current value, future potential, and cost to continue.

A concrete recommendation set

Specific actions: what to do, in what order, why, and what it unlocks.

An ROI recovery plan

For underperforming initiatives worth saving: a scoped remediation path with a realistic value trajectory.

Start with a 30-minute scoping call.

Tell us how many initiatives are in scope, what the audit needs to inform, and how much technical depth is required. We scope it from there. No obligation to proceed.

What we need to know

  • How many AI initiatives need to be audited
  • Whether this is company-wide, a business unit, a product, or a single initiative
  • What decision the audit is designed to inform, board, budget, team, or operational
  • How deep the technical layer needs to go
  • Any hard deadlines driving the timeline