A talk and an AI build, in parallel — where the next decade of marketplaces gets decided.
The talk is about how AI and LLMs are reshaping automotive marketplaces — where demand starts, where value accrues, and what marketplaces should do about it. Carsnip is just one marker of how far things have moved in ten years: a product layer that cost £3.89m and a team of thirty in 2016, rebuilt in an afternoon in 2026 — one person and an LLM, live on stage. The talk runs in three parts.
A working product, built in minutes — live on stage. By the end you see exactly what it produced.
The numbers on AI velocity, shifting demand, and the economic stakes for marketplaces.
What history — Disney in 1957, Lego in 2003 — tells us about navigating this shape of disruption.
of all public GitHub commits are now written by Claude Code alone — doubled in a single month.
SemiAnalysis · Feb 2026growth in AI-driven referral traffic to retail sites, in seven months. Organic search fell 9%.
Adobe Digital Insights · 2026AI-influenced car-buying journeys per year by 2030 — the US figure alone.
BCG / OpenAI · 2026the average UK buyer spends deciding, across four car-search sites. You own one stop.
Motors Digital Touchpoints · 2025Distributed channels, centralised value.
Disney drew this diagram in 1957 — TV, films, parks, merchandise, all separate businesses feeding one centre. Lego lived it in 2003, absorbing the disruptors instead of fighting them.
The same shape now applies to marketplaces: open the data, route the leads, share the revenue, keep the centre. The buyer leaks to a thousand places at once — not to a competing marketplace, but to everywhere else. The question is which posture you take before someone else sets the default.
The full deck — eighteen slides on the disruption, the data, and the precedent.
Open / download the full deck (PDF)Building in this space, or just want to talk through how AI is reshaping automotive marketplaces? Drop me a line.
The question isn't whether someone builds the future. It's whether you build it.
