Transparency

This page describes how we produce what you read, and what we refuse to do. It is kept up to date with the site.

Our European path

We want to build a travel app that is as European as possible. Here is where we stand, without embellishment.

Already in placeIn progressPlanned
  • The company, Tom Explore SAS, and its team are based in France.
  • Tom AI, the app's AI agent, runs on models from Mistral AI, a French company.
  • Account verification emails are sent by Brevo, a French provider.
  • Moving the hosting of our data, currently with Google Cloud, to European providers.
  • Hosting our data 100% in Europe.

The details of the data we collect and how we process it are in our privacy policy.

How our articles are built

Our place data comes from public sources and from our own records, which we then process ourselves. On top of that we add our own expertise: our opinions on the places, our recommendations, what is worth the detour and when to go. Articles are written from that database, with occasional help from artificial intelligence for structure and transitions. Every article is then read through and checked manually before publication.

Where the places come from

Our places have two origins: public sources (business listings, reviews and practical information published by the venues themselves) and our own data, collected in-house. We then process them ourselves: categorisation, location, consistency and quality filters. Human review covers cities and selections rather than each listing taken in isolation.

The ratings we display

The rating shown on this site is the real average of the reviews left on the App Store and Google Play, weighted by their number, and we publish those reviews without discarding any based on its rating. We do not publish an aggregate rating in our structured data: these are reviews from third-party platforms, not reviews we collect ourselves. We neither buy nor reward reviews.

What crawlers read

Search engines and AI assistants have access to the whole site, articles included: we want to be cited. Crawlers that collect text to train models are, however, kept out of our articles. This policy is stated plainly in our robots.txt file and in an HTTP header served on every page: reading and answering allowed, training refused.

What we don't do

We don't serve different content to crawlers and to visitors. We don't make figures up: our usage numbers carry the date they were measured, and our company details are those of the public register. We publish no rating and no review that we haven't genuinely collected.