Free AI training for business owners·Every Thursday at 12pm, Paris time Free AI training·Thursdays 12pm Sign up

Case study · Industry case · E-commerce

An e-commerce site produced 10,000 detailed product pages, with a significant impact on SEO and GEO, at a fraction of the cost.

A group that fits electronics on offshore racing and cruising boats sells thousands of technical parts online. Its founder built a harness of about twenty AI agents that turns the technicians' knowledge and supplier documentation into complete product pages, for customers, for search engines and for AI assistants.

Company
Marine electronics group with a technical online store
Function
E-commerce and search
Built by
The company itself, with Claude agents
Case type
Industry case, shared on our podcast
The founder explaining the system on our podcast
The founder walks through the system on our podcast.
10,000product pages generated
€5per page, against €30 to €60 by hand
~20agents working in parallel

Figures given by the founder on our podcast, October 2026.

The challenge

The knowledge was in the workshop, not on the website.

Technical parts sell on detail: compatibility, installation, the questions a skipper will ask before buying. That knowledge lived in the technicians' heads and in supplier PDFs. A good page written by hand cost €30 to €60, so most of the catalogue had thin pages, or none.

At that price, writing the whole catalogue properly was never going to happen.

What we built

A harness of agents, each with one job.

The founder built what the team calls a harness: about twenty agents running in parallel, each with a role. The product enricher reads supplier documentation and the technicians' notes, then writes a complete page, structured for a customer, for search engines and for the AI assistants that now answer buying questions.

The same system drafts customer-service answers. Every one of them is checked by a person before it goes out.

  1. Collect

    Supplier documentation and the technicians' know-how are gathered for each product.

  2. Enrich

    An agent writes the full page: use, compatibility, installation, frequent questions.

  3. Structure

    Each page is built to be read by a customer, a search engine and an AI assistant.

  4. Check

    The same agents draft customer-service answers, each one reviewed by a person.

Before and after.

Before How it worked

  • Thin or missing pages for most of the catalogue
  • €30 to €60 to write one page by hand
  • Know-how locked in the technicians' heads
  • Customer questions answered one by one

After How it works now

  • 10,000 complete product pages
  • About €5 per page in AI costs
  • Technicians' knowledge published on every page
  • Customer-service drafts prepared, checked by a person
The shift

The catalogue stopped being a cost to write. It became the company's knowledge, published.

The business impact.

  • 10,000product pages enriched and published
  • €5per page in AI costs, against €30 to €60 by hand
  • ~20%of the group's activity now run through the agent system
  • 100%of customer-service drafts checked by a person before they go out
Built with Claude Multi-agent harness Supplier documentation E-commerce platform

This is an industry case: the system was built by the company itself, and its founder shared it on our podcast. We did not build it. The figures are the founder's own; traffic and conversion figures have not been shared.

You can keep adding tools.
Or rebuild how your company works.

Book a discovery call 30 minutes With a founder

Not ready yet? Join the free live on how to create your company brain, every Thursday at 12pm, Paris time.

Sign up for the live