Session 1 — The Problem (Discovery & the Value Gate)

Why most AI initiatives fail, and why the reason is strategic rather than technical. Participants meet their team, their roles, and the shared case study, then read a deliberately contradictory evidence wall before diverging on candidate opportunities with AI support.

– The two ways AI work dies: the POC graveyard and demo-as-product

– The double diamond, and where the Value Gate and Excellence Gate sit within it

– Why AI shifts desirability left but not feasibility or viability — and why that asymmetry matters

– Writing a one-sentence value hypothesis, and making a confident kill

– Team breakouts: diverging on opportunities, then clearing the Value Gate on one

Session 2 — The Solution (AI-Assisted Discovery Loops)

Teams generate three genuinely different solution approaches, then run fast build-test loops — first testing against an AI persona, then swapping teams to test the same slice against a real human reaction. The gap between the two is the heart of the session.

– Generating and choosing between three different solution bets, not three flavours of one idea

– The Loop Card: what this loop proves, what it deliberately won't build, and what changed

– Building a thin, clickable slice with an AI build tool — looping, not assembling

– Testing the same slice against an AI persona and a real human, side by side

– Why a synthetic persona is a hypothesis, never evidence

Session 3 — Ship or Stop(The Excellence Gate & the Decision)

Teams take their working slice and put it under real production pressure — old devices, no signal, no clean data, a customer who doesn't trust it — then write the honest three-line handoff that makes a fast prototype safe to walk away from or build on.

– Why demo scale hides technical debt that production scale reveals

– Demo versus product: designing for one happy path versus for 1,200 real users

– The Excellence Gate: naming what the demo makes look solved that isn't

– The honest handoff, and the persevere/pivot/kill decision

– Transfer exercise: running a real backlog item from your own organisation through both gates

The AI-Accelerated Product Discovery workshops are built for people who make or influence build decisions on AI-augmented product work — not just individual AI tool use.

• Product Managers who want a disciplined way to use AI in discovery without skipping the evidence
• Designers and Engineers working in cross-functional product teams alongside them
• Product leaders who want their teams building fewer, better-validated things, faster
• Organisations that have felt the AI "POC graveyard" problem and want a repeatable fix

No coding background is required. Participants should have access to a standard LLM chat tool and, ideally, a browser-based AI build tool for the hands-on loops — full details are confirmed with each organisation ahead of the session. All exercises run on fictional case-study data, so no client or company data is ever required.

• Write a defensible one-sentence value hypothesis before opening any build tool
• Kill a weak idea confidently — including one that belongs to someone powerful
• Run fast, AI-assisted discovery loops that end in a changed mind, not just more features
• Tell the difference between an AI persona's reaction and real human evidence
• Spot when a demo's speed is hiding real production risk
• Write an honest three-line handoff an engineer would thank you for
• Make a confident persevere/pivot / kill call, and defend it with evidence
• Apply the same two gates to a real item from your own backlog