Conntour, a surveillance technology startup, has successfully raised $7 million in a seed funding round led by General Catalyst and Y Combinator. The capital injection, which also saw participation from SV Angel and Liquid 2 Ventures, will accelerate the development of an AI-powered search engine designed to transform how security teams monitor and analyze video feeds in real-time.
Redefining Surveillance with Natural Language AI
As the surveillance industry faces intense scrutiny regarding privacy and the ethics of AI, Conntour is positioning itself as a platform built for both power and selectivity. Unlike legacy systems that rely on rigid, preset motion parameters, Conntour utilizes advanced vision-language models. This allows security personnel to query live and recorded feeds using natural language, effectively functioning as a search engine for physical premises.
Users can perform complex searches, such as asking the system to “find instances of someone in sneakers passing a bag in the lobby.” The platform then sifts through vast amounts of footage to retrieve the specific event, automatically generating incident reports and surfacing alerts based on the findings.
Efficiency at Scale
A primary differentiator for Conntour is its architectural efficiency. CEO Matan Goldner claims the system is optimized to monitor up to 50 camera feeds using a single consumer-grade GPU, such as the Nvidia RTX 4090. By employing a dynamic logic system that selects the most efficient model for each specific query, the platform minimizes computational overhead while maintaining high performance across thousands of cameras.
The platform offers flexible deployment options, including fully on-premises, cloud-based, or hybrid configurations, ensuring compatibility with existing security infrastructure.
Addressing Ethics and Technical Hurdles
Despite the rapid growth—the funding round closed in just 72 hours following a flurry of 90 investor meetings—Conntour maintains a strict vetting process for its clients. The company currently counts major government entities, including Singapore’s Central Narcotics Bureau, among its users. Goldner emphasizes that this strong customer base allows the company to remain selective about use cases to ensure moral and legal alignment.
To combat the perennial issue of poor image quality in surveillance, Conntour provides a “confidence score” with every search result. If the source footage is low-resolution or compromised by poor lighting, the system flags the results accordingly, ensuring users are aware of the data’s reliability.
The Road Ahead
Looking forward, the company faces the significant technical challenge of balancing full Large Language Model (LLM) flexibility with operational efficiency. Goldner describes this as the core contradiction of the sector: “We want to provide full natural language flexibility… and on the other hand, there’s efficiency. Processing thousands of feeds is just insane. This contradiction is the biggest technical barrier in our space.”
As Conntour continues to scale, its ability to marry high-level AI reasoning with resource-efficient processing will likely define its trajectory in the competitive security software market.