Coram AI raises $35 million to rethink security cameras

Published: 2026-06-12T08:28:48.000Z

Most security cameras are dumb in the most literal sense. They record whatever passes in front of them and wait for a human to review the footage later. Coram AI wants to flip that, raising $35 million to turn ordinary cameras into something closer to an autonomous investigator.

Most security cameras are dumb in the most literal sense. They sit on a wall, record whatever passes in front of them, and wait for a human to scrub through the footage after something has already gone wrong. Coram AI wants to flip that script, and investors are backing the idea. The San Francisco company has raised $35 million in a Series B round to turn ordinary cameras into something closer to an autonomous investigator, bringing its total funding to $66 million. The round was co-led by Ansa Capital and Battery Ventures, with UP Partners, 8VC and Mosaic Ventures joining in. The pitch rests on a real pain point. When an incident happens at a large organisation, security staff often spend hours, sometimes days, stitching together video, access logs, visitor records and alarms scattered across disconnected systems to figure out what occurred. Coram's answer is a tool it calls Deep Investigation, an AI agent you can question in plain language. It searches months of footage and entry data across hundreds of cameras and multiple sites, then returns a report with the relevant evidence attached. Work that used to eat up hours, the company says, now takes minutes, which lets a single security professional cover far more ground without adding headcount. The founders bring an unusual pedigree to the problem. Ashesh Jain and Peter Ondruska both spent years building artificial intelligence for self driving cars, Jain leading autonomy work at Lyft's self driving division and at Zoox, and Ondruska heading AI research at Lyft and later Toyota's Woven unit. Having taught machines to perceive and make decisions about the chaotic physical world from a moving vehicle, they reckoned the same techniques could transform an industry that has barely been touched by modern AI. The traction suggests they are onto something, with revenue reportedly quadrupling since the Series A and the platform now running at more than 1,500 sites including Fortune 500 firms, school districts, hospitals and a Dallas megachurch watching over tens of thousands of worshippers. There is a sharper edge to all this worth naming. The same platform that speeds up investigations also offers facial recognition, licence plate reading and live gun detection, and it is being pointed at schools, churches and workplaces. Coram leans on a privacy friendly design, running its AI on local chips so footage need not leave the building, but an agent that can autonomously investigate across every camera and door is also a powerful surveillance machine. The efficiency is genuine, and so is the reach. The regional angle is hard to ignore. Across the Middle East and North Africa, governments are investing heavily in AI driven surveillance and smart city infrastructure, with the UAE and Saudi Arabia building some of the most camera dense urban environments anywhere as part of projects like NEOM. Technology that makes security systems autonomous fits squarely into that ambition, though it carries the same uncomfortable trade off between safety and monitoring that the Gulf's smart cities are already navigating.