
/Product3 min read
Europe's agents need web search that stays in the EU
See how Tavily’s EU endpoint keeps agent search queries in the EU, with local processing and optional zero query retention.
Europe is putting AI to work. The European Commission's Joint Research Centre found AI use more than doubled in almost every strategic sector between 2021 and 2025, and added that "in many cases AI still needs to be embedded into core business processes". Many of these processes depend on current information from the web. An agent that checks a supplier, enriches a CRM record before a sales call, or collects evidence for a case has to go out and search for it.
Depending on the use case, a search query may contain the name of a supplier, a customer, or a case, and that information travels with the request. For a European team, that raises a question it has to answer before shipping: where does that query get processed? If the team has EU processing requirements, the answer cannot be anything other than "in the EU".
That is what we built the Tavily EU endpoint for. Queries are processed in the EU, and with zero query retention enabled, no query content is kept after the response is returned.
Where the query is processed
The EU endpoint is a separate deployment of Tavily search that processes your queries in the EU. It uses only the web indexes and search runtime we built and own, and every step, including work handled by subprocessors, happens inside the EU.
What stays in the EU
The API sits in Frankfurt. Our index and its retrieval runtime are hosted on Nebius infrastructure in France. Reranking and model calls happen on EU infrastructure, and logs and monitoring stay in Europe. With zero query retention enabled in your account settings, no query content is kept after the response is returned.
Everything you already have with Tavily remains unchanged: your organization, API keys, usage, and billing remain on Tavily’s shared systems, which are hosted in the US. This is EU query residency.
Built around what European teams need
Owning the web index means we can build coverage around the sources our customers need: company registers, regulatory notices, and local reporting. We work with each team on the sources and languages their agents depend on, and those shape what we add next.
For example, at a plant run by thyssenkrupp, the German industrial group, an engineer who programs and tests a battery inspection machine with a Siemens copilot had been typing to it in English. One day, he tried "Moin", a casual German greeting, to see if it would recognize it. It answered in German, and he spent the rest of the session in German. I want European teams to have that same choice in the sources their agents search.
Bring us your use case
European teams are increasingly building agents, and the search call inside them needs to stay in the EU. The EU endpoint is live, and we are already onboarding teams. If you are building agents whose queries have to stay in the EU, join the waitlist, bring your own queries, and be among the first to run them from home.