Participated · Series A · Oct 2021

Launched in 2017 and led by CEO Maya Pindeus, Humanising Autonomy develops ethical computer-vision software that understands, predicts, and interprets human behaviour. Its Behaviour AI connects to existing camera platforms or video streams to provide context between people and machines and delivers actionable data and insights. The technology is positioned for deployment across mobility and smart spaces sectors, and the company plans expansion into new verticals. A recent debt financing from Silicon Valley Bank UK is intended to support continued deployment and growth. The financing amount was not disclosed; it follows an $11m Series A in October 2021. Humanising Autonomy aims to help businesses make operational decisions, reduce risk, and enhance human experiences. Humanising Autonomy provides a Behaviour AI platform that uses modular, low-power computer vision to ethically understand and predict human behaviour for real-time and historical analysis. The GDPR-compliant technology is used to inform business decision-making and integrates with chip manufacturers such as Ambarella. The company launched new cloud-based products in 2020 and intends to accelerate cloud deployment into additional industries and geographic markets. It plans to scale use of its platform to inform business processes and to expand its offerings in advertising effectiveness, smartphone applications, construction and warehouse safety in 2022. Humanising Autonomy also intends to expand existing automotive and dashcam capabilities that can alert drivers and inform fleet management. The company will expand its diverse team across Europe, Asia and North America as it grows. Humanising Autonomy builds modular machine-learning models that predict pedestrian behaviours from RGB camera footage to help vehicles and autonomous machines read and respond to human intentions. The company trains models on diverse data sources — CCTV, dash cams and autonomous-vehicle sensors — and specifically analyzes accident footage to improve robustness across cities, camera types and conditions. Its models use only RGB data (forgoing lidar depth) and are modular so components can be tuned for different environments and processing constraints, enabling deployments from high-power Level 4/5 stacks to lower-power aftermarket retrofits. HA is working with mobility providers in Europe, the U.S. and Japan and has case studies and partnerships with Daimler Mercedes Benz and Airbus. It sees adjacent opportunities in warehouses and manufacturing where partial retraining of modules can adapt the system to worker-monitoring use cases. The company says its system outperforms humans on the task and aims to be licensed into sensing stacks rather than having manufacturers build bespoke solutions. The recent seed financing is intended to accelerate commercialization and integration into tier-1 stacks.