Request a pilot

Your cameras, finally paying attention

SurveiLens watches the cameras you already have — tracking everyone in frame and flagging real threats the moment they happen.

Runs on-site  ·  Your footage never leaves the building  ·  Works with your existing cameras
SurveiLens Monitor · real detection output LIVE

Real SurveiLens output on real store CCTV — boxes and alert generated by the model, not mocked. Robbery / theft detection is in active pilot validation.

How it works

Three steps, running quietly on your feeds.

01 — WATCHES

It watches every camera at once.

SurveiLens reads your live feeds the way a focused guard would — tracking people and movement across the frame, on every camera, without a break and without getting tired.

02 — UNDERSTANDS

It judges what's actually happening.

Not just "something moved" — SurveiLens reads the scene and tells a fight from a hug, a threat from normal activity. When this waiting-room scuffle turned physical, it flagged it as a real threat within a second. On a calm store, it stays silent. (Audio cues are being added.)

03 — ALERTS AT ONSET

It flags the moment it crosses the line.

Not a report the next morning — a real-time alert the instant something becomes a threat, so your team can act while it still matters. We detect at onset; we don't claim to predict the future.

One engine, two very different problems

A gas station and an ER share almost nothing — except a camera and a threat.

The same detection engine, pointed at two problems. What changes is what you tell it to watch for, and how sensitive it is — tuned on your cameras.

REAL-TIME ALERT · LOSS PREVENTION
Robbery / theft in progress REVIEW
Person reaching over the counter toward the register, away from a normal transaction.
cam 04conf 0.90status pilot
Retail · loss prevention

Someone moves on the register.

SurveiLens flags the robbery as it happens and pings the owner at the register. On real store footage it fires on the theft and stays quiet on paying customers — though we're still measuring that rate. (Robbery/theft detection is in active pilot validation.)

REAL-TIME ALERT · STAFF SAFETY
Physical aggression detected URGENT
Physical altercation in the waiting area. Security notified in real time.
cam 11conf 0.90onset +1s
Healthcare · staff safety

A confrontation turns physical.

Aggression toward staff pings security the moment it turns — someone's on the way before it escalates. This is the capability we can stand behind today: in our testing it fired only on real physical violence, and stayed silent on everything else.

Loitering in a stairwell. A fall in an empty corridor. A crowd forming. You choose what SurveiLens watches for — and how sensitive it is — on your cameras.

Built to earn a security team's trust.

Hardware, privacy, false alarms — the things most AI security gets wrong, built right from the start.

Runs on the cameras you already own

It's software, not another camera to buy. Point it at your existing IP cameras and NVRs — nothing to rip out, no capital project before you see a single alert.

Judges what's actually happening

Reads the whole scene in context — a fight from a hug, a real threat from normal activity — not just "something moved."

Your footage never leaves the building

Every frame is processed on-site. Nothing ships to anyone's cloud to become a breach waiting to happen.

Alerts your team will actually act on

A pre-violence detector that fires on every shadow gets unplugged in a week. We tune every zone to your space and measure our false-alarm rate on your own footage before we trust it.

Proof in progress.

Real pilots on real footage — measuring catch rate and false alarms in the buildings that need it most.

Pilot · in progress

Hospital

University police running SurveiLens against real emergency-department footage on-site — measuring catch rate and false-alarm rate in their own ER, footage never leaving the building.

Pilot · starting

University campus

A large open campus with cameras already everywhere — testing detection of fights, crowds forming, and after-hours activity across quads, walkways, and parking.

Pilot · starting

Convenience store

An owner-operator with a real shoplifting problem, testing whether SurveiLens catches concealment in time to act at the register.

Frequently asked questions.

Yes. SurveiLens runs on hardware inside your building. Video is processed on-site and never uploaded — there's no cloud copy of your footage to leak. The only thing that can leave, if you want it to, is a lightweight alert.

No. SurveiLens is software — it runs on the IP cameras, NVRs, and feeds you already have. There's nothing to rip out and no new cameras to install. Most pilots start on your highest-risk cameras with zero new hardware on the wall.

Days, not months. Because there's no hardware to install, we point SurveiLens at a sample of your existing footage on-site and you're seeing real detections quickly — no rewiring and no capital project before the first alert.

Yes — it's a monthly software subscription, and you only pay for the cameras you choose to monitor, so you can start with just your highest-risk ones. The rate depends on what those cameras are protecting — a hospital ER and a corner store aren't priced the same — so we scope it with you rather than post a flat number. And the pilot is free, so you see exactly what it catches in your space before you pay anything.

We bring a machine on-site and run SurveiLens against a sample of your own footage, in your building. You get a report — catch rate and false-alarm rate on your cameras — and we get the data to tune it. Design-partner pilots run at no cost.

Put SurveiLens on your cameras.

Tell us about your space and we'll set up a pilot on your own footage.

No cost for design partners. We reply within a day.