of burglars said cameras or surveillance would push them to pick a different target.
University of North Carolina at Charlotte · interviews with 422 incarcerated burglarsThe person you would be backing.
I know which doors get tried, at what hour, and why one shop on a strip gets hit and the one beside it does not.
I am seventeen and I have lived in Scarborough my whole life. That sentence above is not colour for a pitch. It is the reason Ocula stays quiet almost all night, speaks only when someone stops or reaches for a handle, and never calls the police on anyone's behalf.
Each of those is a constraint most teams learn from a complaint. A product built somewhere else would have shipped a system that shouts at everything that moves, because that demos well and nobody on the team has watched what it does to a street.
I built the whole system myself. Detection, the judgement layer that decides whether anything is worth saying, the escalation ladder, the voice, the phone alerts, the morning reports, the incident records for insurers, and the unglamorous parts that keep it alive when a camera drops or the process dies. 44 automated tests. An earlier version was lost, and I rebuilt it from written product decisions, tests before features, to a working confrontation on real hardware.
I install every unit personally, and that is the point rather than a gap in the plan. It is how the field data gets collected: what was seen, what was said, whether they left and how fast. Nobody publishes that number, because guard companies bill hours instead of results. It is the input to the judgement and eventually the only claim in this category anyone can actually verify.
The honest sheet
Age17
BasedScarborough, Toronto
TeamSolo, no cofounder
SystemRuns end to end
Automated tests44 passing
Paying customersNone yet
Revenue$0
RaisedNothing
The thesis
The evidence and the price point say the same thing.
Every claim rests on one finding: cameras stop crime when somebody is watching and acting, and do close to nothing when they are not.
Zeromeasurable effect from cameras that only record
Actively monitored cameras cut crime significantly. Record-only systems showed no statistically significant effect at all.
40 year systematic review, 161 studies, 1978 to 2017 · Piza, Welsh, Farrington, Thomas
60%
$300 to $800
a month for live guard monitoring, because the service is priced in human hours.
Published pricing, Deep Sentinel and comparable servicesShops already pay for a voice at 3 AM. The cost floor is labour, so the category can only serve properties that absorb hundreds a month. Automating the watcher moves the same outcome to about $99 and opens the low end, which is where nearly all of the doors are.
Defensibility
Why the obvious copy does not happen quickly.
The people who could copy it fastest lose the most by doing it
Monitoring companies bill by the hour. Automating the watcher deletes the line item their revenue sits on. That is the standard reason incumbents arrive late to their own disruption.
The platforms with distribution will not confront anyone
A wrong confrontation is a support ticket for us and a news cycle for a company with millions of installed cameras. They ship conservative because the blast radius forces them to. That is structural, not a lack of engineers.
The asset compounds and cannot be bought
Every confrontation produces a labelled outcome: what was seen, what was said, whether they left and how fast. Nobody publishes that. It is the input to the judgement, and to the only credible claim in the category.
The low end is a segment incumbents cannot price into
Not unwilling. Unable, for as long as a person has to be paid to watch a screen.
Said plainly, because pretending otherwise would be the bad sign: none of this is a permanent barrier. It is a window, plausibly two to three years after the category is publicly proven. The work is to spend that window accumulating judgement data and published field results, so what exists at the end is a record nobody can replicate by shipping a feature.
Where this actually is
There is no traction to dress up.
What exists is a system that runs on real hardware, a category that is already paid for, and the speed the first version was built at. Nothing else yet.
What is built
Detection, judgement with armed hours and zones, a four rung escalation ladder, the voice, evidence clips, phone alerts written in plain language, morning reports, incident records for insurers, offline watchdog, startup self test, crash supervisor and retention. 44 tests passing.
What is not
No paying customers, no revenue, nothing raised, and zero armed nights on a real storefront. The next milestone is deliberately small and falsifiable: pilots on Toronto shops, and the first field numbers nobody else publishes.
If any of the above changes, this page changes with it. I would rather be told no on accurate information than yes on a story.
Get in touch
Reach me directly. It is one person.
No call centre, no ticket number, no assistant screening for me. Whichever of these you use, it reaches my own phone.
Text or call437 757 2052
Emailoculaai@outlook.com
WhereScarborough, Toronto. Happy to meet in person, which is also how every install happens.
Most useful right nowAn introduction to a shop worth piloting. Field nights are the thing standing between this page and real numbers.