Ambient AI Inc., developer of physical security software that uses image recognition to identify threats captured on security cameras, today announced that it has added 25 new behaviors to its threat detection library.

That brings the total number of visual scenarios it can now analyze to more than 150. The new behaviors include the ability to identify weapons, detect crowd commotion and spot unauthorized entry attempts.

The company, which exited stealth mode with $52 million in venture funding early this year, says its technology can sharply reduce false alerts while ensuring that security cameras are constantly scrutinized. Its technology analyzes video feeds from existing security cameras to detect threats without using facial recognition or other tactics that can compromise privacy.

Its signature library looks for such anomalies as cars parked overnight in a business parking lot or a person “tailgating” closely behind another to bypass electronic security protections. Founders Vikesh Khanna and Shikhar Shrestha both hold master’s degrees from Stanford University in smart imaging and deep learning. Its investors include Andreessen Horowitz LLC and Okta Inc. co-founder Frederic Kerrest.

Physical security incidents continue to plague organizations; 88% of security leaders surveyed by Ontic Technologies Inc.’s Center for Protective Intelligence early this year reported a “dramatic increase” in physical threat activity from last year.

Conventional surveillance systems are prone to missing incidents because human monitors can’t keep up with every video feed at the same time. Ambient AI says its technology can deliver alerts in real-time to enable responses up to 10 times faster than is possible with human guards.

Among the behaviors the 25 new signatures cover are a person brandishing a firearm or knife, doors being held or forced open, a person jumping a fence or loitering after hours, unauthorized vehicles in a secure area, people falling, large crowds forming, sudden egress and a person removing assets from a secure area.

The signatures will become part of the core product. Founded in 2017, Ambient AI says it now counts five of the most highly capitalized U.S. technology companies as customers as well as multiple Fortune 500 companies.

“We have experienced an amazing amount of demand [and] have closed more new customers and bookings in the last six months than we did since founding the company, including multiple new seven-figure annualized recurring revenue wins,” said Shrestha in an email interview. “With security threats on the rise, we believe the security industry is at a real inflection point right now.”

Future enhancements to the product will be aimed at improving early threat identification, reducing door alarm noise, new threat signatures for perimeter security and real-time search capabilities aimed at reducing investigation times, he said.

Image: Ambient AI