Excitement versus structure.
For anyone who leads or manages designers, researchers, or a wider design and research team, whose organisation is being told to use AI without a plan for what that means. Built and proven inside a live e-commerce business, not written as a theoretical framework.
Enthusiasm without structure does not scale.
Most AI programmes fail for one reason. The excitement is real, but nobody has built the scaffolding to turn it into something a business can act on. Leadership asks a team to use AI. Nobody defines what good looks like, how quality gets checked, or what happens to the ideas that come out the other end.
This programme is the structure I built in response to exactly that question, raised directly by my own manager after repeated pressure from above with no plan behind it.
The Five Step Method
Once the team had a shared capability baseline, I put it to a real test, not a hackathon, something with actual stakes and a fixed shape. Here's what happened when I ran it for real.
The business signed a partnership agreement with Google and Gemini in the months that followed. I won't claim the challenge was the sole cause, enterprise procurement has too many moving parts for any one programme to take full credit, but it ran through the same evaluation window and covered the same ground.
Where does your team sit?
Experimenting
Individuals use AI opportunistically, with no shared approach and no visibility for anyone else on the team.
Assisted
AI gets used repeatedly inside individual workflows, but nobody's evaluated it, and nothing is shared or repeatable.
Structured
The team has an agreed baseline, consistent evaluation, and a shared understanding of where AI genuinely helps.
Integrated
AI is embedded into the actual product development process, not bolted onto the side of it.
Strategic
AI capability starts influencing team structure, tooling decisions, and product strategy itself.
Most teams asking how to use AI are somewhere between Experimenting and Assisted. The Five Step Method moves a team from Assisted to Structured. An assessment starts by giving you a straight answer on where you are now, not an aspirational one.
You are not at Level 3 unless all three are true: someone other than the person who wrote a prompt has evaluated its output against an agreed standard, a new joiner could find and reuse the team's prompts without asking anyone, and you could name, today, one task where you concluded AI was the wrong tool, and why. A team that cannot name a rejection has not been evaluating, only adopting.
Everything, shown upfront
Understand: The AI Workflow Playbook
The method, the proof, and what's involved at each stage. Enough to know whether this is worth pursuing.
Diagnose: The AI Workflow Assessment
Book directlyA working session that leaves you with five things: a capability map, an opportunity map, a readiness review, a recommended pilot, and an executive recommendation.
Prove: The AI Workflow Pilot
A working engagement that leaves you with five things: a trained team, a completed sprint against a real problem, peer-reviewed submissions, a facilitated senior judging panel, and roadmap-ready ideas with scope and effort defined.
Scale: The AI Workflow Advisory
For organisations wanting the parallel, multi team version, or ongoing involvement embedding the capability permanently.
Talk about your team
If this already sounds like a gap on your team, the fastest next step is a conversation, not another download.