Conntour Raises $7M From General Catalyst, YC To Build An AI Search Engine For Security Video Systems

AI Security Camera Search Engine Raises $7M — Turning Clients Away

A two-year-old startup just closed a $7 million seed round in under 72 hours. Its AI platform lets security teams search thousands of camera feeds using plain English — and its founders are quietly choosing who gets access.

Credit: Conntour

An AI-powered search engine for security camera footage just raised $7 million in seed funding — and the entire round wrapped up in less than three days. The startup behind it, Conntour, lets security teams search live and recorded camera feeds using plain, everyday language, almost like typing into a search engine. Major government agencies and publicly listed companies are already using it. And yet, the founders are saying no to certain clients.

$7MSeed raised

72hRound closed

50×Feeds per GPU

What does an AI search engine for security video actually do?

Most surveillance systems work on rigid rules — motion zones, object sizes, preset triggers. Conntour throws all of that out. A security operator can type something like, "Find anyone in sneakers passing a bag in the lobby," and the platform searches all footage — live or recorded — and returns relevant clips in seconds. This is made possible by vision-language AI models, which understand the content of video the way a human analyst would, just much faster and at enormous scale. It's a fundamentally different way of thinking about physical security monitoring.

A $7 million round that closed in three days flat

The seed round came together with unusual speed, backed by General Catalyst, Y Combinator, SV Angel, and Liquid 2 Ventures. The CEO scheduled roughly 90 investor meetings across eight days — and had commitments locked in by Wednesday afternoon of the very first week. That kind of momentum doesn't happen without genuine traction to show. Conntour already counts Singapore's Central Narcotics Bureau among its active clients, alongside several large government bodies and publicly listed corporations. Those names carry weight in investor conversations.

Why this AI surveillance startup is turning customers away

Here's the part that makes Conntour's story genuinely unusual. Despite being just two years old, the company is deliberately screening which clients it will and won't work with. The surveillance industry is under real scrutiny right now — debates around law enforcement access to commercial camera networks and neighborhood-level video monitoring are intensifying. Conntour's leadership says that pressure has only strengthened their resolve to vet every customer carefully. Their existing high-profile contracts give them the financial room to walk away from business they're not comfortable with — something most early-stage startups simply can't afford to do.

"We're really in control of who is using it, what is the use case — we select what we think is moral and, of course, legal."

Matan Goldner · Co-founder & CEO, Conntour

The technical edge: processing thousands of camera feeds efficiently

Most AI video platforms buckle at enterprise scale. Conntour was built with that ceiling in mind from the beginning. A single consumer GPU — like the Nvidia RTX 4090 — can handle up to 50 live camera feeds simultaneously on the platform. The system achieves this by routing each query through a mix of specialized AI models and logic layers, always picking the most efficient combination for the task. Deployment is flexible too: fully on-premises, entirely cloud-based, or a hybrid, and it plugs into most existing security infrastructure without a full rip-and-replace.

Built-in honesty: confidence scores over empty promises

One of the quieter but more meaningful design choices Conntour has made is what happens when the system isn't confident. Poor lighting, low-resolution cameras, dirty lenses — these are real-world problems that have limited surveillance quality for decades. Rather than hiding uncertainty behind polished output, Conntour returns a clear confidence score alongside every search result. If the footage quality is poor, users see that reflected immediately. It's a small detail, but it signals a maturity you don't often see in surveillance tech this early in its lifecycle.

The big unsolved problem still ahead of them

The CEO is refreshingly open about the central tension the team is racing to crack. Full natural language AI flexibility demands serious computing power. Monitoring thousands of feeds efficiently demands the opposite. Getting both right at the same time — without forcing customers to choose between capability and cost — is the defining technical challenge in AI security video right now. For Conntour, that contradiction isn't something to hide. It's the roadmap. And it's exactly the kind of hard problem that tends to define which startups become platforms and which ones fade.