Building an AI Solutions practice
Leading the development of an AI Solutions practice that connects product strategy, transformation and delivery, starting from organisational problems rather than models.
- Type of work
- Practice deliverables
- Organisation
- WPP
- Sector
- Enterprise services
- Role
- Head of Product Management and AI Solutions
- Period
- Nov 2024–present
- Disciplines
- AI transformation, Operating model, Governance, Propositions
- clients served in my WPP role
- 25+
Problem
Most organisations do not lack access to AI. They lack a reliable way to turn it into changed work: which problems to pick, how to redesign the workflow, how to govern it, and how to get from a convincing demo to something people use every day.
Context
As Head of Product Management and AI Solutions at WPP, I own product strategy across commerce, data and AI. I lead the product management capability, setting portfolio priorities, standards, staffing and coaching for product managers, product owners and business analysts.
I built the AI Solutions practice to connect opportunity selection, workflow redesign and adoption with investment decisions, organisational capability and governance. The aim is a repeatable way to turn AI investment into products and workflows that people use, with a clear path to scaling delivery.
Why it mattered
AI work tends to stall in the gaps between disciplines. Strategy identifies opportunities without a delivery path. Engineering builds capable prototypes with no workflow to land in. Commercial teams struggle to estimate work that is new to everyone.
A practice has to connect all three. Otherwise every engagement starts from zero.
Approach
I treated the practice itself as a product: a clear customer problem, a repeatable path, and a commercial model that works for both sides.
The path runs from AI readiness and opportunity selection to workflow redesign, proof-of-concept design, adoption and scale-up. Each stage asks a different question. Readiness asks whether the organisation, its data and its governance can support AI. Opportunity selection asks where the value is concentrated. Workflow redesign asks how the work should change. The proof of concept asks whether that change holds up with real users. Adoption and scale-up ask whether it lasts.
Around that path sit the commercial pieces (propositions, estimates, staffing models and statements of work) and governance, including transparency considerations under the EU AI Act.
System model
Governance Human oversight, evaluation and EU AI Act transparency, from the first stage to the last
-
AI readiness
Can the organisation, its data and its governance support AI?
-
Opportunity selection
Where is the value concentrated?
-
Workflow redesign
How should the work itself change?
Where value is designed -
Proof of concept
Does the change hold up with real users?
-
Adoption
Is it used, and is it trusted?
-
Scale-up
Does it last, and can it be repeated?
Commercial model Propositions · estimates · staffing models · statements of work
Decisions
-
Start from workflows, not models
Opportunity selection looks for valuable organisational problems and the workflows around them. Model and vendor choices come after the problem is understood, not before.
-
Design proofs of concept for adoption
A proof of concept should test whether people will use it and whether it moves a measure, not only whether the technology works.
-
Make governance part of the offer
Governance, human oversight and transparency obligations under the EU AI Act are scoped from the first conversation instead of being added at the end.
-
Productise the commercial side
Repeatable propositions, estimates, staffing models and statements of work make AI work easier to buy, plan and deliver consistently.
Practice deliverables
An AI Solutions capability that connects product strategy, AI transformation and delivery.
- 25+ clients served in my WPP role
- A defined path from AI readiness to scale-up
- Commercial propositions, estimates, staffing models and statements of work
- Governance built into delivery, including EU AI Act transparency considerations
What I learned
AI transformation is mostly organisational design. The model is usually the easiest part to change.
Client names, commercial terms and internal details are confidential and deliberately left out.