An Evergreen Search Box for Physical Security

Just like Google is a search box for the Web, Camio is a search box for Physical Security. Now that Artificial Intelligence understands what’s happening in the real world, search is the simplest and fastest way to deliver real-time insights.

The Search Box gives AI its power to optimize every stage of security event processing.

AI shines in video analytics — often as a slick forensic search or alert like “vehicles at loading dock over 20 seconds” or “people tailgating in lobby.” But when AI is applied throughout the entire pipeline — from video and sensor event processing through the resulting workflows and data — it reduces costs 10x and improves security efficiency by 50x. The simplicity of the search box is central to delivering these benefits, because it shifts the burden of managing security systems to AI that’s focused on the desired outcomes instead.

1. Natural language descriptions of real-world events are very effective in determining the optimal system behaviors and responses at every stage of the pipeline (e.g. retain video 365 days for slip-and-fall, encode empty rooms as lo-res timelapse, dispatch guards when breach of IDF, etc…)
2. New scenarios are simple to incorporate because new activity classifiers simply add their vocabulary to the search engine query parser (e.g. adding a detector for a gas nozzle in a vehicle comes free with the new query terms “nozzle” and “nozzle inserted”)
3. The search box is evergreen in accommodating new technologies, communication protocols, workflows, reporting, and multi-modal inputs (voice, text, machines, languages,…)

Search is evergreen in easily expanding its vocabulary to incorporate technology advances.

CFOs pay for results (not systems management)

It’s the protection and business insights that drive budgeting for physical security. Yet our budgets go to video management systems (VMS), door alarms, and stationed guards. The technical insight behind Camio is that real-time video search, when baked into every processing decision, moves us from managing systems to putting them to work for us.

Take a simple example of protecting IDFs. The desired outcome is to intervene immediately if a person enters without authorization. Historically, a VMS records video either continuously or upon motion detection. An access control system sends alarms for door held open, door forced open, and access granted. And guards are dispatched only after a SOC analyst wades through 93% false positive alarms. Instead, with real-time video search applying AI throughout the process, we get:

1. Over 80% reduction in storage costs. Recording in HD only when people are present in the IDF, even while keeping a 24x7 continuous video history via low resolution time lapse, reduces storage far beyond simple pixel based video encoding techniques like H.265.
2. Over 97% reduction in bandwidth. When AI summarizes each event with a compact metadata representation, a 2Mbps video stream becomes a 30kbps trickle with just the essential information required to make decisions about subsequent steps.
3. 50x efficiency gain in SOC triage. By reducing false motion alerts 10x and review time by 5x, we’re able to protect 50 times the number of locations with the same staff.
4. 4x faster response times. When mobile guards can receive alerts on their phones only for the events that warrant their intervention (e.g. their alert trigger query “unauthorized entry in IDF”), they act on relevant events delivered immediately to them.
5. 100% more hardware capacity. When search is foundational, it makes hardware more efficient. e.g. “vehicle approaching entrance” is a trigger query that avoids burning GPUs on license plate recognition on passing cars. Kubernetes workloads balanced by query conditions enable the same hardware to cover twice as many streams.
6. Over 300% reliability gain in compliance. Human attention span ranges between 15 and 20 minutes. A wall of live streaming video, interrupted by phone calls, conversations, and bathroom breaks inevitably misses events. Machines don’t get fatigued or distracted. They don’t miss a thing.
7. Information — not just data. Seeing a door held open alarm is a lot different than knowing that there were also inbound people associated with that alarm. So searches like “IDF tailgating” provide actionable insights and measurement of security policy compliance and vulnerabilities.

Embedding search throughout the pipeline drives cost savings, efficiencies, and reliability.

Don’t miss a thing

Beyond the efficiency and reliability gains above, the search box is evergreen in its ability to leverage new innovations continuously and quickly. As the pace of innovation in Artificial Intelligence accelerates, it’s critical not to be painted into a corner.

The extensibility of search engine query language, when applied at every stage of the platform rather than as a forensic analytics layer on top, means that every AI advance has an immediate path to delivery. Unleash the machines!