What does AI search say your product is for?

A company makes claims for its product — to investors, to the market, to itself. What counts is the experience people actually have of it: what it let them do, what they learned, what they now trust themselves to manage. The two rarely match. The outcome a person is after, in their own life, is often not the one the product was built to claim — and the work is to make people good at the first, not the second.

Engineer and builder, twenty years freelance. Products built for AB InBev, Sony PlayStation, Renault and Thales; an agile, product and automation transformation of customer service at Booking.com. I work inside a team on product, on building with AI, and on AI search — for makers of connected products, products for movement and bodywork, and hospitality and travel.


Three practices

01

Product work inside the team

Not strategy from above. I look at what people use, what they get done, and where they get sent back — then work on what is real.

An engagement looks like: A seat in the team, a product to own, decisions made from use.

02

Building with AI inside the team

Forward-deployed. Agents, harnesses and working code, built in your codebase and rituals — not handed over.

An engagement looks like: I build it, instrument it, and leave it running with the team.

03

AI search — what the engines say you are for

I read what ChatGPT, Perplexity and Google AI Overview tell buyers a product is for. I name the representation it has, the one it wants, and the path between them.

An engagement looks like: A measured representation, an intended one, the work between.


The fit

All three practices come from building products, not reading about them.

Work →
  • Connected products: Connected products from zero for Renault, Vantiva, Baracoda and Thales; at AB InBev, the PerfectDraft programme to global launch — app, recommendation model and firmware as one product.
  • Hospitality and travel: Travel and hospitality at Swisscom, Avis Budget and Wizz Air, and customer-service product at Booking.com.
  • In my own practice: Gyrotonic, Airofit, PowerDot — the products I use, read and check against what they claim.

Since July 2026: more than 160 questions monitored, more than 4,600 AI answers read, more than 42,000 cited sources logged, across 8 brands and 5 markets, on ChatGPT, Perplexity and Google AI Overview.


Twenty minutes

No prep. AI search: three questions your buyers ask. Product work: the part that isn't working yet.

Book a call

Pass this on

Patrice Gorin is an engineer and builder, twenty years freelance.

He works inside a team on product, on building with AI, and on AI search — what ChatGPT, Perplexity and Google AI Overview say a product is for — for makers of connected products, products for movement and bodywork, and hospitality and travel.

Send by emailLinkedIn profile

Looking for something specific