Sightline launches real-time video understanding SDK and nets $28M Series A

AI · 5 min read

Sightline’s SDK can detect store shelf events, count people with anonymized tracking, and classify actions with timestamped rationale outputs suitable for audits. The models are optimized to run efficiently on edge devices and include tools for domain adaptation without uploading raw video.

The $28 million Series A was led by EdgeOrbit and will fund additional edge optimizations, regulatory compliance tooling, and partnerships with camera OEMs. Sightline positions itself for use cases where video cannot leave the premises due to privacy or bandwidth constraints.

Pilot customers in retail reported measurable improvements in out-of-stock detection and shrink reduction, while robotics teams valued the SDK’s low-latency affordances for reactive behaviors. Sightline plans to add multimodal sensor fusion (lidar + video) in the next major release.