Conntour Raises $7M Seed to Bring Real-Time AI Understanding to Video Surveillance
Most systems watch. Conntour pays attention, and that difference is where things start to get uncomfortable for anyone relying on yesterday’s definition of “visibility.” Out of Miami with serious muscle in Tel Aviv, Matan Goldner and Tomer Kulla built this after seeing firsthand what happens when “smart” surveillance misses the obvious because nobody pre-labeled the threat. Turns out reality does not wait for your dropdown menu. So they ditched the rigid rulebook and gave operators something closer to instinct, except it runs on silicon and does not blink.
Now the market is paying attention. $7M in seed, led by General Catalyst with Yuri Sagalov in the mix, alongside Y Combinator, SV Angel, and Liquid 2 Ventures. Not bad for a company that only just stepped on stage publicly on March 26, 2026. Even better, the round closed in 72 hours after roughly 90 investor conversations. That is not hype. That is velocity meeting conviction.
The product reads like science fiction until you realize it is already in play. Natural language queries across live and recorded feeds. “Show me the guy with the tattoo.” Done. “Alert me if someone climbs that wall.” Done. No presets, no babysitting the system like it is a toddler with crayons. It plugs into existing cameras, runs on premises, and scales without asking you to rip and replace your entire stack. One RTX 4090 handling up to 50 feeds like it is a light workout.
And the results are not subtle. Up to 90% reduction in manual review time. Missed events down by as much as 80%. False alarms cut by 70%. Operators go from watching hundreds of cameras to thousands, which is less about working harder and more about finally working like the system should have allowed all along.
The proof is already walking. A paying deployment with Singapore’s Central Narcotics Bureau within months of founding. A nod from The Wall Street Journal. A seat in Y Combinator’s W25 batch and Palantir’s first startup fellowship cohort. When government operators trust you early, it usually means you solved something they could not afford to keep getting wrong.
The takeaway is not just about better surveillance. It is about removing the assumption that the world can be pre-programmed. Conntour leans into the chaos and says, ask better questions in plain language and let the system figure it out. That is a different posture entirely, and it travels well across security, logistics, retail, and anywhere cameras outnumber humans.
