COMPREDICT is showcasing the Headlight Leveling Virtual Sensor at ELIV - Electronics in Vehicles in Baden-Baden.

COMPREDICT is a German deep-tech company and pioneer in AI-powered Virtual Sensor technology. Founded as a spin-off of TU Darmstadt, the company builds intelligent software algorithms that extract precise, actionable measurements from vehicle signals that are already on board — no additional hardware required. Where traditional approaches demand physical sensors wired into a vehicle, COMPREDICT's Virtual Sensors compute the same outputs in software: tire pressure, headlight leveling, suspension displacement, brake and tire wear, drivetrain loads, and more. The result is a leaner bill of materials, faster development cycles, and continuous vehicle health monitoring across the entire product lifecycle — from prototype to series production to aftersales. COMPREDICT's technology is deployable in the cloud or embedded directly into vehicle systems, and scales across the full mobility spectrum: passenger cars, commercial vehicles, buses, two-wheelers, and rail. Customers include global vehicle OEMs, Tier-1 suppliers, and fleet solution providers. Solutions & Products Virtual Sensors for Vehicle Development — Replace physical test sensors with AI-based alternatives that deliver up to 95% accuracy, cutting development time by up to 50% and testing costs by up to 30%. Software-Defined Vehicle Functions — Convert previously unmeasurable physical parameters (forces, torques, load cycles) into digital signals, enabling new vehicle capabilities without hardware additions. Health & Usage Monitoring — Real-time and predictive monitoring of tire wear, brake wear, powertrain ageing, and battery state across individual vehicles and entire fleets. DataPilot (SaaS, launched 2026) — COMPREDICT's cloud-based platform for scalable Virtual Sensor deployment. DataPilot automates the quality checking, signal matching, event labelling, and data export pipeline for vehicle time-series data — making the full power of Virtual Sensing accessible without deep engineering integration effort.

Founded 2016