Smart Eye has announced that it is to host a technology demonstration next month of a new multimodal in-cabin AI agent that can interpret context, decide what to do next and take action
The agent draws on multiple sources of input, including voice, driver identity, gaze, attention, emotion, driver state and occupancy, alongside information from the vehicle’s interior and exterior sensing systems.
Agentic AI can require significant computing resources, particularly for complex tasks. Adding multimodal context to each request can provide agents with more complete and targeted information, potentially reducing the processing required to reach a useful outcome. This could help make automotive agentic AI practical for a wider range of applications, the company Eye says.
In Smart Eye’s demonstration, the agent will carry out tasks through connected vehicle functions and services and create dashboard widgets using live driver and cabin data. It will also combine gaze with exterior sensing to understand what a driver is referring to outside the vehicle.
Built on years of research into human attention and behavior, the platform combines human insights with data from ADAS, vehicle systems and other sensors. It supports different foundation models, voice systems and agents, operating either at the edge or in the cloud.
“Automotive agents become economically viable when the value they create outweighs the cost and complexity of running them,” said Martin Krantz, CEO and founder of Smart Eye.
“Our agnostic platform is the first to combine human, cabin, vehicle and environmental context for automotive agentic AI. We believe this multimodal approach can bring the technology into production much sooner. At Smart Eye we are proud of this world-first release.”
The technology will be presented publicly for the first time at InCabin Europe in October 2026.
Recent news, Stoneridge debuts EVO ECU platform for commercial vehicle ADAS and connectivity