A structured review of the on-premises video management system category for IT, security and facilities teams running their own recording servers: what it does well, where the constraints sit and who should look elsewhere. This page is about a category rather than a company, so every line describes traits common to on-premises VMS products.
A traditional video management system is the most dependable thing in this whole comparison at the job it was designed for. It records every camera to storage in the building on infrastructure you own, holds retention for as long as the policy says, and does it with no cloud dependency at all. The category's other quiet strength is device support, and two products on this hub show the scale: Milestone lists more than 16,500 devices on its Supported Device List for XProtect, and Genetec publishes its own, with analog reachable through encoders and licensing usually per channel. It scores lowest on what the recording is for, because analytics are typically separate modules or partner add-ons rather than detections that ship, and because response at the moment of an incident depends on somebody being in front of a monitor. Choose a traditional VMS when dependable recording under your own control is the requirement. Look elsewhere when the footage has to raise the events worth seeing rather than wait to be searched.
Five criteria, each scored against what on-premises video management products commonly document, reviewed on August 24, 2026. Traditional VMS is a category rather than a company, so there is no single vendor's documentation behind these scores: where a line rests on a published document it names one, such as Milestone's Supported Device List for XProtect or Genetec's supported device list for Security Center. Spot AI competes with products in this category, so no score here rests on an anonymous source, an aggregated user rating or a private benchmark. Check each line against your own product's release notes and licence terms before you plan around it. This is not a paid placement.
Every box here describes traits common to on-premises video management products rather than one vendor's documentation. Capture and retention are the category's strength. What happens after capture is where the gap sits.
A traditional video management system is a recorder with a client on top. Cameras stream to recording servers or an NVR in each building, video is written to local storage sized to the retention policy, and an operator opens a client application to play it back, scrub a timeline and export a clip. Licensing is usually per channel, analog reaches the system through encoders or a hybrid recorder, and administration happens per site.
That architecture is not a compromise. It is the reason a large share of hospitals, schools, utilities and manufacturers chose it: video never leaves the building, storage is auditable, the estate keeps working when the internet does not, and an internal team can inspect the whole system without a vendor in the room. An evaluation that treats on-premises recording as a legacy default has misread why it is there.
The category's other genuine strength is breadth of hardware support, and the documents are public. Milestone publishes a Supported Device List for XProtect carrying more than 16,500 devices, updated as new devices and firmware versions are evaluated, and Genetec publishes a supported device list for Security Center, so procurement can check a real inventory against a real document before anything is bought. What these products generally do not ship is a set of named detections: analytics arrive as separate modules or partner add-ons, priced and deployed on their own, and on-site deterrence is not part of the category at all.
A traditional VMS is strongest at capturing and keeping video under your own control, and thinnest at telling anybody which of that video was worth watching.
Both columns are documented. Nothing here comes from an anonymous review.
The Spot AI column describes documented behavior. The category column describes traits common to on-premises VMS products rather than any single vendor's documentation, so check it against the product you actually run.
Swipe the table sideways to see every column.
The category column summarises traits common to on-premises video management products, reviewed on August 24, 2026, rather than one vendor's public documentation. Where a line rests on a named document, the document is named.
The honest split, stated the way a shortlist call would state it.
If the requirement is that every camera records reliably, that retention meets a policy somebody will audit and that video never leaves the building, this category answers it and has answered it for years. Estates with a data-residency condition, a poor uplink or an internal team that wants to inspect the storage themselves have a genuinely strong reason to stay exactly where they are.
If the problem is that nobody has time to watch, that investigations take half a day each or that safety and operations now want answers from the same cameras, the recorder is not the constraint and buying a newer one will not move it. That is a case for detections that ship with the platform and a search that starts from a sentence rather than a timestamp.
Spot AI ships 15+ pre-trained Video AI Agents spanning vehicle break-in, fire, intrusion, personal protective equipment, forklift near-miss, falls and hazard-zone crowding, so a mixed estate gets shipped coverage on both sides. Iris builds anything else in natural conversation in about eight minutes, full-resolution video stays on the Intelligent Video Recorder in the building and the platform is SOC 2 Type II, NDAA-compliant and HIPAA-aligned.
The two are less opposed than a shortlist makes them look. Spot AI keeps the same on-site recording model, with full-resolution video on the Intelligent Video Recorder and only event metadata leaving the building, so the data-residency argument that anchors a traditional VMS survives intact. What changes is what the footage is asked to do: pre-trained agents watching the same feeds, plain-English search over the archive, and something that fires on site rather than waiting for a person to read an alert.
The cheapest way to settle it is a site running both at once for a fortnight. The existing servers keep recording throughout, so nothing is at risk and the comparison happens on identical footage. Pick the site with the messiest camera mix rather than the newest, because a fleet assembled over ten years is the real test.
Model three years both ways against the same camera count and the same retention obligation, and put the server refresh and storage growth in the on-premises column explicitly. A perpetual licence looks cheapest right up until the hardware cycle lands.
See what the AI catches on your live feeds before any platform decision.
Customer-reported outcomes from named Spot AI customers.
Liberty-Perry School District resolves an incident in about five minutes, after evaluating ten systems before choosing Spot AI.
Cambridge City cut footage search from two hours to 30 seconds after consolidating seven municipal locations on one platform.
Silver Bay Seafoods replaced fragmented legacy camera systems across 22 locations, including remote Alaska facilities, and lifted operational efficiency 15%.
"The biggest benefit of Spot AI is how easy it is to look up incidents, see the footage, and then share it and collaborate."
Yes, for the job it was designed for. The category scores 3.2 out of 5 here, carried by dependable local recording, retention control on storage you own, supported device lists in the thousands and no cloud dependency at all. It fits an estate whose requirement is reliable capture under its own control. Teams whose problem is that nobody has time to watch the footage should look at what ships with the platform rather than at the recorder.
Yes, and it is the category's strongest suit. Milestone publishes more than 16,500 devices on its Supported Device List for XProtect and Genetec publishes a supported device list for Security Center, licensing is usually per channel, and analog channels reach the system through encoders or a hybrid recorder. Check your own product's list for your specific version rather than the category description, because breadth varies between products more than the summary suggests.
Motion alerting is standard across the category, and anything deeper is typically a separate module or a partner add-on with its own licensing, its own server requirement and its own integration project. Some products publish a license plate recognition module of their own. What the category does not generally do is ship a set of named detections with the recorder, and on-site deterrence is not part of it at all.
The shape is consistent: perpetual licences per channel plus recording servers, storage and an annual maintenance agreement, with a capital refresh when the servers reach end of support. Model three to five years rather than one, because the maintenance renewal and the hardware cycle are what actually move the number, and both land outside the licence line.
It depends on which part is failing. If retention and residency are working and only the searching is painful, an alternative has to keep video in the building to be a real replacement. A camera-agnostic platform such as Spot AI does that: full-resolution video on the Intelligent Video Recorder on site, any ONVIF or RTSP camera connecting as it is, analog through the recorder, and 15+ pre-trained Video AI Agents across security, safety and operations. The alternatives roundup compares five platforms side by side.