San Francisco startup Conntour has secured $7 million in seed funding to tackle a persistent problem: the sheer uselessness of most recorded surveillance video. The company argues that while organizations globally have installed over a billion cameras, the footage is often a digital black hole—recorded but never reviewed until a painful, manual search is required.

Conntour’s solution is software that layers onto existing camera systems, allowing users to search video with natural language. A query like “white truck backing into dock after 10 PM” would return specific clips in seconds. The round was led by General Catalyst, with participation from Y Combinator, where founders Jared Bhatt and Keshav Dulal met in early 2026.

The technology relies on modern vision-language AI models to index video in real time, analyzing objects, actions, and context. This approach, now viable due to recent advances in multimodal AI, requires no hardware replacement. This hardware-agnostic stance is a key differentiator in a market where rivals like Verkada tie analytics to their own cameras.

Conntour is avoiding facial recognition, focusing instead on attributes like clothing or vehicle types to sidestep privacy controversies. Initial customers are in retail and logistics, aiming to reduce theft and speed up investigations. The company charges a monthly SaaS fee per camera.

While competitors from startups to legacy providers offer video analytics, Conntour is betting that the flexibility of its search engine and the timing of the underlying AI will convince enterprises to finally make their camera investments pay off.

Source: Webpronews