What Smart Sentry AI's own documentation says its monitored service does not cover, written for security and facilities teams running due diligence. Every constraint carries its documentation, its operational impact, and its workaround.
Every constraint below comes from Smart Sentry AI's own current documentation and FAQ, checked on August 17, 2026. Where the company documents nothing, this page says so rather than guessing, because an absence in a document is a question to ask rather than a finding to claim. Spot AI sells a competing platform, so nothing here rests on an anonymous source or an aggregated user rating. Spot AI has published its own page about Sentry AI, and this check found nothing in it that Smart Sentry's current documentation contradicts, so this banner carries no correction.
The sizing rule, the install time and the response chain are all published as numbers. What each camera is asked to look for, and what happens to the video afterwards, are not on a page.
A constraint list is only useful next to an honest account of the product.
Smart Sentry's own FAQ states the system works with all cameras and NVRs that can be connected to the internet, wired or wireless, and adds that almost every camera able to connect has good enough resolution for the AI to detect incidents effectively. Installation is published as about an hour when the cameras are already online, which turns a pilot into an afternoon rather than a project plan.
One camera covers an area of roughly 100 feet by 50 feet, an average property runs 8 to 12 cameras, a smaller lot manages with 6 and a larger one needs 16 or more. Detection latency is published at 5 seconds and deployment at 500 or more enterprise sites. A buyer can estimate their own coverage and challenge a quote before anybody walks the site.
Live talk-down, a roaming guard dispatched to the site for intervention and law enforcement called are all published, with a security expert reviewing the incident in real time before escalation and the monitoring team coordinating between agents, patrol guards and the customer. Much of this market stops at an alert to a phone, so a documented human verification step ahead of a documented response is a real strength.
Each one is a consequence of how the service is designed rather than a defect. What matters is whether it collides with your starting point.
Smart Sentry publishes people inside the product: security experts review the incident in real time, roaming guards are dispatched to the site for intervention, and the monitoring team coordinates between agents, patrol guards and the customer. The published sizing rule sits alongside it, one camera to roughly 100 feet by 50 feet, with 8 to 12 cameras on an average property and 16 or more on a larger one.
The recurring line is monitored hours and dispatch capacity, not licenses, so putting this next to a per-camera software subscription compares two different things. A 12-acre equipment yard needing 16 or more cameras is buying continuous human attention across those feeds, and the honest benchmark is the guard shift or the patrol contract it replaces rather than another platform's license.
Model it against your current guard and patrol spend, not against a software line, and use the published sizing rule to estimate camera count per site before anybody walks it. Ask which hours are monitored, what happens outside them, and how dispatch is charged when a guard is actually sent, because those three answers are what the recurring number is made of.
The service is described in terms of people, vehicles and unusual behavioral patterns rather than as a named catalog with conditions attached, and which detections run on which camera is not publicly specified. The documentation is precise about sizing, install time and what happens after a detection, and it does not state what each camera is being asked to look for.
The coverage a specific site gets is settled in the sales conversation rather than on the website, so two quotes for the same yard can mean different things and neither is checkable in advance. It also makes an acceptance test hard to write, because there is no published list to test against.
Ask for the detection list per camera type in writing and get it attached to the contract rather than the proposal. Because installation is published at about an hour when the cameras are already online, a two-week trial on the specific cameras you care about is cheap, and it answers the question the documentation does not.
The named verticals are enterprise, commercial, multi-family, auto and parking lots, and construction, and the published detections are security shaped. No safety or operations set appears anywhere: personal protective equipment compliance, forklift proximity, dock dwell, spills and procedure adherence are not part of what the service describes itself as covering.
On a construction site or a distribution yard the same cameras are usually being asked safety questions by a different budget holder, and this service answers the overnight question only. An operations director trying to understand why a gate queue builds at shift change is looking at a second platform on the same feeds, or at a camera program that never gets past security.
Split the requirement by function and by hour. Keep the monitored service for the overnight perimeter job it is built for, and price the safety and operations coverage separately, either as a second product or as a platform that ships both sets on the cameras already installed.
The published requirement is that the cameras or the NVR can reach the internet, wired or wireless, which means video leaves the site. What is not published is a storage location, a retention period, an export path or a statement of what continues when connectivity is lost. The documentation covers what has to be reachable, not what happens when it is not.
A construction site on a temporary link and a multi-family property on a shared uplink are exactly the environments this service targets, and both have outages. Without a documented degraded-mode behavior, a security manager cannot tell a client or an insurer what coverage looked like during the hour the connection was down, and evidence retention is unanswerable from public pages.
Ask in writing what continues to function without connectivity and for how long, by site type, and request the retention schedule, the export path for evidence and the legal-hold behavior. Confirm the bandwidth needed per site in the same conversation, because the whole model rests on the uplink.
No SOC 2, ISO or NDAA statement appears beside a service whose staff watch live video from customer premises, and no published standard covers how monitoring agents or dispatched guards are vetted, trained or licensed. This is an absence in published evidence rather than a finding about the service, and it is exactly the kind of absence a security review turns into a blocking question.
A multi-family operator or an enterprise campus running a vendor review will need documents that are not on the website, and if they surface late the procurement slips. Guard dispatch also carries state-level licensing and insurance requirements that vary, so a national rollout has a per-state question attached to it.
Request the certification pack, the agent vetting and training standard, and the licensing and insurance position for guard dispatch in each state you operate in. Ask early, treat it as a documentation exercise rather than a disqualification, and put the answers in the evaluation record next to your own policy.
The same five constraints in one view, sized to paste into an evaluation document.
Swipe the table sideways to see every column.
Sentry AI data comes from Smart Sentry AI's own product pages and FAQ, checked on August 17, 2026. Gaps are marked as not publicly specified.
If the job is covering sites overnight and the budget being displaced is guard hours, most of the constraints above stop applying. A construction site, an auto lot, a parking structure or a multi-family property with no security team of its own is buying continuous human attention on cameras that already exist, with a documented response that ends in somebody arriving. The published sizing rule, the roughly one-hour installation and the 5 second detection latency make that easy to trial on one site before committing to a portfolio.
Spot AI answers the same overnight problem with software rather than staffing, which is why the constraint lists differ. 15+ pre-trained Video AI Agents are named and shipped, covering vehicle break-in, fire, cash register theft, after-hours intrusion, personal protective equipment, forklift near-miss, falls and crowding in hazard zones, and Iris builds anything else in natural conversation in about eight minutes. There is a published list to test a pilot against, and it spans safety and operations as well as security.
The architecture also answers the connectivity question directly. Full-resolution video stays on the Intelligent Video Recorder in the building and only event metadata crosses the network, so recording does not depend on the uplink and the residency conversation stays short. Any ONVIF or RTSP IP camera works at full functionality and legacy analog comes in through the IVR, and an event at 03:00 is answered automatically with talk down, strobes and horns through standard speakers.
The useful question is not which option has fewer constraints, but whether you are buying attention you do not have or automation you can run yourself.
None of that makes Spot AI the right answer here, and for some sites it is not the same purchase at all. Spot AI is software, so it does not staff a monitoring center, review an incident on your behalf, dispatch a roaming guard or call law enforcement for you, and a property whose real gap is that nobody is available at 02:00 is buying something Spot AI does not sell. Smart Sentry also publishes an installation time and a detection latency figure, and Spot AI publishes neither, so on those two numbers there is more to hold Sentry AI to than there is to hold Spot AI to. A constraint list earns its keep by lining each option's shape up against your starting point.
A live pilot on your cameras answers in a week what a spec sheet cannot.
Customer-reported outcomes from named Spot AI customers.
A top five North American EV charging network reports an 80% reduction in incidents with autonomous deterrence and no human monitoring.
Don Franklin Family of Dealerships had incident footage on responding officers' phones within four minutes of an alarm, and reported five recoveries within the hour.
Tidewater Fleet Supply unified three distribution centers and 14 retail locations from Florida to Virginia on one platform.
"With Spot AI, these facilities are more secure and better watched than the ones we staff traditionally, where we pay for a full-time person on-site."
Five, all documented. It is a staffed monitored service, so the recurring cost is monitored hours rather than licenses and it does not compare like for like with software. The detections are described as people, vehicles and unusual behavioral patterns rather than named, and which run on which camera is not specified. The published scope is perimeter security only. Where video is stored, for how long, and what happens when connectivity drops are not publicly specified. And no certification or agent vetting standard appears on the published pages.
Yes, and unusually plainly. Smart Sentry's own FAQ states the system works with all cameras and NVRs that can be connected to the internet, wired or wireless, and adds that almost every camera able to connect has good enough resolution for the AI to detect incidents effectively. Installation is published as about an hour when the cameras are already online. What is not specified is which detections run on which camera, so that belongs in the contract rather than the proposal.
Smart Sentry publishes the rule rather than leaving it to a site walk: one camera covers roughly 100 feet by 50 feet, an average property runs 8 to 12 cameras, a smaller lot manages with 6 and a larger one needs 16 or more. That makes it one of the few services in this category where a buyer can estimate coverage and challenge a quote in advance, and it is also what shapes the recurring cost, because the monitored feed count follows the camera count.
Four, because none of them are on the public pages: the certification pack covering SOC 2, ISO or NDAA if it exists, the retention schedule and export path for video, the vetting and training standard for monitoring agents, and the licensing and insurance position for dispatched guards in each state you operate in. Ask early rather than late, and treat the answers as a documentation exercise rather than a disqualification.
It depends on whether the gap is attention or automation. If nobody is available at 02:00 and the budget being displaced is guard hours, a monitored service is the right shape and Sentry AI publishes more of its own numbers than most. If the requirement is a named detection set across security, safety and operations that a team can run itself, a platform such as Spot AI fits: 15+ pre-trained Video AI Agents run on any ONVIF or RTSP camera plus legacy analog through the Intelligent Video Recorder, video stays on site, and talk down, strobes and horns fire automatically.