[MEDIA] The Icetana AI Agents (There are currently two called...

The Icetana AI Agents (There are currently two called 'Watch Agent' and 'Triage Agent') present what it thinks is unusual to the camera operators.

But here's the cool part. Right now Icetana are testing (in a lab environment) the Chat GPT multimodal model - so that images and videos can be sent directly to and from different AI agents - and those AI agents have the cognitive power to assess a situation and make appropriate recommendations / take action.

e.g. the Triage Agent can currently make assessments like "Is this person on the camera in danger? Is the property on camera in danger? Does this incident need escalating?"

e.g. the Watch Agent can simply search for things like fire and smoke on camera, or someone holding a gun or a knife. If you are running a shopping centre, you can actually upload the rules of the shopping centre (no loitering/no smoking/don’t block fire escapes) to the Watch Agent, and it will scan all the cameras and look for any breaches that are detected.

The reason why this is interesting in security terms, is that a critical part of the process can be offloaded to AI for security teams that typically have to monitor thousands of cameras on their own. The Holy Grail of autonomous security is obviously full automation. Right now, people are needed in the loop to manage these Icetana AI Agents. But what Icetana think is coming soon is their AI managing the lion’s share of the CCTV process.

This will only get better over time.