Modern AI models excel at language but are spatially blind: a language model reasons in words, not physical environments, and quickly hallucinates once it comes to current situations such as closed roads or relocated shops. TomTom aims to close that gap with its Orbis platform, which forms the basis for two product lines: the Orbis Road Model for hands-off driver assistance, already active in tens of millions of vehicles, and the Orbis Lane Model for eyes-off autonomous driving, fed with anonymised camera data from regular cars instead of expensive mapping vehicles.

For the enterprise market, TomTom introduced specific tools for AI agents, including the Maps MCP (Model Context Protocol) and the Agent Toolkit, which let language models query geospatial data directly to verify their answers against reality. In practice, US-based Flow Labs already uses the traffic data to adjust traffic lights in real time, cutting rush-hour delays by more than 24 percent. Through a partnership with market leader Esri, TomTom’s data now also reaches over 650,000 business users of the ArcGIS platform.

That shift from hardware to software and data has pushed the company’s gross margins from around 40 to roughly 90 percent, according to CFO Taco Titulaer.