Coram AI raises $35M to deploy autonomous agents that investigate security footage in minutes

Coram raises $35M to turn cameras into AI detectives

The startup's agents can search months of footage across hundreds of cameras in minutes. It sells that as privacy-preserving, even as it pushes facial recognition and gun detection into schools and churches. Coram AI has raised $35m to turn the security cameras already bolted to walls into something closer to an autonomous detective. The Series B is co-led by the new investor Ansa Capital and Battery Ventures, with UP Partners, 8VC and Mosaic Ventures joining. It takes the San Francisco company's total funding to $66m. Coram's pitch is that physical security is stuck in the past. When something goes wrong, staff spend hours scrubbing through footage, access logs and alarms to piece together what happened. Its answer is software it calls 'Deep Investigation', an AI agent you query in plain language. It searches months of video, entry records and visitor data across hundreds of cameras and sites, then hands back a report. Work that took hours, the company says, now takes minutes. Founded four years ago by Ashesh Jain and Peter Ondruska, Coram now runs at more than 1,500 locations, from schools to factories. Privacy pitch, surveillance reality Coram leans hard on privacy. Its boxes run AI models on local NVIDIA chips at the edge, it says, so sensitive video never has to leave the building for the cloud. It also works with any existing IP camera, avoiding a costly rip-and-replace. But the same platform sells facial recognition, licence-plate reading, 'tailgating' detection and live gun detection, and it is being pointed at schools, churches and workplaces. One customer, a Dallas megachurch, watches over 30,000 worshippers across eight campuses. A high school swapped old cameras for real-time weapon detection. The efficiency is real; so is the reach. That trade-off, safety bought with more monitoring, is not new to AI security. But autonomous agents sharpen it. A system that can investigate on its own, across every camera and door, is also a system that is always watching, and now draws its own conclusions. The 'operating system' land grab Coram is part of a wave of startups trying to become the 'operating system' for a single industry by wrapping AI agents around it. Its bet is that every building will eventually run hundreds of agents in the background. The money is chasing a real gap. 'Physical security is one of the largest industries yet to be transformed by modern AI,' said Allan Jean-Baptiste of Ansa Capital, and the incumbents largely sell cameras and dashboards, not autonomy, even as firms pour record sums into AI elsewhere. For now, the headline numbers, '10x more effective', 'hundreds of agents per space', are Coram's projections, not proof. But with $66m in the bank and 1,500 sites live, it has the runway to test whether the building of the future really does watch itself.

Physical security startup Coram AI raises $35 million co-led by Ansa Capital, Battery Ventures

The San Francisco-based company was cofounded by Ashesh Jain, formerly head of autonomy at Lyft's self-driving division, and Peter Ondruska, who led AI research at Lyft before moving to Toyota's Woven division after it acquired Lyft's self-driving unit in 2021. Physical security startup Coram AI has raised $35 million in a round co-led by Ansa Capital and Battery Ventures, with participation from UP Partners, 8VC, and Mosaic Ventures. This brings its total funding to $66 million. The San Francisco-based company was cofounded by Ashesh Jain, formerly head of autonomy at Lyft's self-driving division, and Peter Ondruska, who led AI research at Lyft before moving to Toyota's Woven division after it acquired Lyft's self-driving unit in 2021. Coram operates an AI platform that integrates video surveillance, access control, visitor logs, and emergency response into a single interface. The company's software is deployed on-premise and is built to work with cameras and infrastructure organisations already own, sidestepping costly hardware overhauls, Jain told ET. "We spent years building AI that helps cars read a scene and act before someone gets hurt. The same approach protects places where people live and work: catch risks earlier, and keep schools, hospitals, and workplaces safer instead of just documenting what went wrong," Jain, who's also the CEO, said. The new funds will be used to accelerate product development, expand the go-to-market and customer success teams, and grow its engineering presence in Bengaluru, which the company describes as a key hub for building its global platform. Since its $13.8 million Series A round last year, the company claims to have quadrupled revenues and tripled its customer base. It now covers over 1,500 sites across the US and Canada, including Fortune 500 companies, school districts, healthcare providers, and municipalities. According to Jain, most physical security systems in place now are reactive, i.e., they record incidents rather than flagging them, and that the same advances in AI perception used in autonomous vehicles can be applied to make security operations more proactive. "There is a monetary benefit of integrating all the security aspects on one platform through Coram, but the more important return on investment (ROI) for customers is increased proactive security, compliance with standards like Occupational Safety and Health Administration (OSHA), and reduced insurance costs," Jain said. The platform is deployed on-premise at the customer site to ensure data privacy, while maintaining full cloud accessibility for mobile and desktop users. The current customer base, predominantly located in the United States and Canada, includes government facilities, schools, hospitals, retail, and manufacturing sectors. The startup is also launching a new capability called Deep Investigation, which deploys autonomous agents to analyse videos, access events, and visitor activity across cameras, locations, and time to answer complex security queries with the evidence attached. These agents can conclude investigations in minutes as opposed to the hours it took previously. The company runs offices in the Bay Area, London, and Bengaluru.