Contributing editor
In a Security Info Watch Q&A, Vikesh Khanna describes what the company means by agentic physical security, puts numbers on a ServiceNow deployment, and draws a line between automating alarm triage and automating a door lock.
The argument Rish Gupta made to loss-prevention audiences this summer: retail asked its security teams to cover more stores, more footage and more kinds of incident, and gave them cameras that only watch.
Video agents have to be tuned to each site, and a trade-show booth cannot show that. Spot AI has taken the Retail Zone at the SVIP Experience Center in Dallas-Fort Worth so buyers can watch the tuning happen on a floor that looks like theirs.
A new class of agents that watch for a specific behaviour and its consequence, shipped in July 2026. What each one does, and the small interface change that matters more than it looks.
Surveillance records people so that someone can assign blame later. The safety directors running video agents describe something else: a camera that notices a forklift near-miss now, so the next one never happens.
June to September brought a dozen agent features from seven vendors. Sorted by what each one does, and whether it ships, is in beta, or exists only in a press release.
In October 2024 Spot AI launched the first Video AI Agents and closed a $31 million round with Qualcomm Ventures. What the chipmaker was actually buying.
The cloud was supposed to eat everything. For video agents it lost on arithmetic: a facility with 64 cameras produces 30 terabytes a month, and a lock has to fire in under a second.
SiliconANGLE's report on the Menlo Park startup's Series A describes an agent that detects anything you can describe in words, warns intruders over a speaker and runs standard operating procedures with a human on call.
Spot AI put a video agent on a mast with lights and a loudspeaker. Surveillance towers have looked like this for twenty years. What changed is that this one can tell a delivery driver from a thief, and speaks to each differently.
Custom video detection used to mean an ML team and six to eight weeks of work. Iris made it a ten-minute conversation. What that changes.
Ninety million automated actions, one hundred thousand cases, two thousand agents built. What one platform's 2025 tally says about where the work actually is.
The Copenhagen VMS maker says its summarization and search tools were tuned on a 300,000-item library of customer video, gathered under fair-exchange agreements and anonymized before use. Summaries ship now; search arrives at the end of 2026.
Rish Gupta says every big call at Spot AI runs through one line of Sherlock Holmes. The same four steps are a sensible way to decide what a video agent is allowed to do on its own.
The Eagle Eye VMS now takes a plain-English description of what to watch for, such as a worker without a safety vest, and starts sending notifications. No new hardware, and one integrator has it guarding EV charging cables.
Three LinkedIn posts from the CEO of Spot AI, thirteen months apart, track video agents from a hashtag on a record-quarter note to the reason the company rebuilt how it sells.
Ambient.ai, Coram and Milestone have each put a report-writing step inside the video system. What they automate, and where the operator still signs, differs more than the press releases suggest.
Whether the model runs in the building or in someone else's data center is now a procurement gate. In the sales calls it was raised by a meat plant, a clinic, a laundromat chain, an integrator in Nigeria and a telecom reading the EU AI Act.
Read in order, Spot AI's customer mailers from 2021 to July 2026 show a product that shipped search first, analytics second, an assistant third and agents last. The page where people look at cameras got rebuilt the whole way through.
The inbound email has changed shape. It now arrives as a numbered list: false-positive governance, human-in-the-loop, auditability, on-prem options, validation paperwork, and a pilot whose length the buyer has already decided.
In March 2026 Spot AI stopped selling direct and put $2 million of equity on the table for partners. It is betting that the integrators who pull cable today will be the ones tuning agents on site.
A hyperscale site is a construction project for a year and a half, then a locked box for twenty. Cameras now measure the poured concrete, catch the missing hard hat, spot the loiterer at the fence and watch the cage door.
A July 2026 post by Rhombus's Jeremy Gulley describes a command-line tool that exports a live snapshot of every site and device, and a Claude plugin that loads it before each query. The premise is that an agent that does not know where the cameras are cannot be trusted with much.
Ask an advisor why monitoring customers leave and the answer is a camera that stayed dark for a week. The support queue and one 82-second clip say the same thing.
AI News reported on the April 2025 launch of a no-code builder that lets a plant manager describe a detection, feed it a handful of examples and attach an action such as shutting down a line. The claim was eight minutes, against the eight weeks a custom detector used to take.