AI is the biggest shift in how we design and build products in my career. I'm not watching it from the sidelines — I use it daily, build with it, and form my view by doing. My take, in short: AI doesn't replace design judgement, it amplifies the people who have it.
How a 25-person design team went from first experiments to 90 initiatives in 90 days — and helped shape an agentic operating model for a ~400-person Product, Design and Engineering organisation. Shared by invitation.
I built our long-term supplier product vision hands-on in Claude Code as a working prototype, with AI embedded in the experience and powered by MCP connectors. It aligned our teams and the C-suite, and it's now the key artifact behind rebuilding our supplier platform — in partnership with an agency to accelerate delivery.
A personal agent with a memory vault that learns me, my team, my stakeholders and my projects — reading from tools like Productboard, Figma, Slack and my calendar to write my morning brief, prep 1:1s and flag where attention is needed. Inspired by Dex, the AI chief of staff from Pendo's CPO. Built deliberately as proof of the operating model I was proposing: agents handle the routine so people can focus on judgement.
Principles I've tested in practice — leading a design team and a product organisation through an AI transformation.
AI amplifies the people with taste and judgement — it doesn't substitute for them. The bar for craft goes up, not down. And it has to be demonstrated, not just stated.
The same user obsession applies to AI features as anything else. Useful beats novel; trust, clarity and control matter more than ever.
Leaders can't delegate understanding AI. I form my view hands-on — building with the same tools I ask my team to adopt, and learning visibly alongside them.
Adoption doesn't come from instructions. It comes from people seeing their colleagues' work celebrated — and finding their own reasons to jump in.
Low-risk decisions can sit with agents and deterministic guardrails. The higher the stakes, the more human judgement belongs in the path.
Adopting tools is the easy part. The real work is changing how a product organisation builds — while it's still shipping.
At EcoVadis I was one of the leaders of a 90-day mandate to make AI part of how products are designed, built and shipped. I co-authored a 13-activity framework redefining our product development lifecycle (PDLC) for an AI-native way of working, took my 25-person design team from first experiments to 90 initiatives — with every designer contributing — and helped facilitate the planning that produced an agentic operating model for a ~400-person Product, Design and Engineering organisation.
The standout wasn't planned: designers began shipping code — branching the live codebase, making front-end changes with AI and opening pull requests, with engineers as reviewers rather than intermediaries.
Before that, at Constellation I drove AI and ML initiatives across the group — including a vehicle damage-detection product that created new revenue streams and increased confidence in both consumer and trade buying. The pattern I keep coming back to: start from a real customer problem, use AI where it genuinely moves the needle, and hold the same quality bar you'd hold anywhere else.
Always happy to talk where AI is heading in design and product.