Conntour Nabs $7M to Make Security Video Searchable with AI

The surveillance technology industry currently finds itself under intense scrutiny, not for advancements, but due to mounting controversies. Recent events, such as the U.S. Immigration and Customs Enforcement's alleged access to Flock’s camera network for monitoring individuals and home camera manufacturer Ring facing criticism for features enabling law enforcement to request neighborhood footage from homeowners, have ignited a widespread public debate concerning safety, privacy, and the appropriate boundaries of observation.

Despite these controversies, the market for surveillance technology remains robust, fueled by continuous advancements in vision-language models. These innovations are empowering companies to develop increasingly sophisticated solutions for premises monitoring.

Matan Goldner, co-founder and CEO of the video surveillance startup Conntour, emphasizes the critical ethical considerations within this industry, stating that his company maintains a highly selective approach regarding client acquisition. While this might seem unconventional for a startup just two years old, Goldner explains that Conntour's existing portfolio of significant clients, including various government entities and publicly-listed corporations such as Singapore’s Central Narcotics Bureau, provides the financial stability to exercise such discretion.

In an exclusive interview with TechCrunch, Goldner elaborated on this strategy, stating, "The fact that we have such big customers allows us to select them and to stay in control […] We’re really in control of who is using it, what is the use case, and we can select what we think is moral and, of course, legal. We use all our judgment, and we make decisions based on specific customers that we’re okay [to work with] because we know how they will use it."

This demonstrated traction has yielded benefits beyond client selectivity. Investors have recognized Conntour’s potential, leading to a recently secured $7 million seed funding round with contributions from prominent firms including General Catalyst, Y Combinator, SV Angel, and Liquid 2 Ventures.

Goldner revealed that the funding round concluded remarkably quickly, within just 72 hours. He recounted, "I think I scheduled around 90 meetings in like eight days, and just after three days — we started on Monday and by Wednesday afternoon, we were done."

Conntour’s discerning approach appears well-founded, particularly considering the advanced capabilities of current AI tools in the surveillance sector. The company's proprietary video platform leverages AI models, enabling security personnel to conduct natural language queries across camera feeds. This functions as a specialized "Google-like" search engine for security video, allowing real-time identification of specific objects, individuals, or situations within footage. Furthermore, the system can autonomously monitor for and detect threats based on predefined rules, automatically generating alerts.

In contrast to older surveillance systems that rely on predefined parameters for detecting objects, motion patterns, or behaviors, Conntour asserts its system's superior flexibility and usability stem from its integration of natural and vision language models. For instance, a user can pose a query like, “Find instances of someone in sneakers passing a bag in the lobby,” and Conntour’s platform will rapidly scan all recorded or live video feeds to present pertinent results.

The platform's integrated AI models also empower users to ask direct questions about the video footage, receiving textual answers alongside the corresponding video segments, and to automatically generate comprehensive incident reports.

However, Conntour’s primary differentiating factor is its exceptional scalability. Goldner clarified that the platform stands apart from other AI video search services due to its design for efficient expansion, capable of managing systems with thousands of camera feeds. He specifically noted that Conntour’s system can monitor as many as 50 camera feeds using just a single consumer-grade GPU, such as an Nvidia RTX 4090.

This efficiency is achieved by employing a sophisticated approach involving multiple models and logic systems. The algorithm intelligently identifies the most suitable combination of these models and systems for each specific query, thereby minimizing computational demands while delivering optimal results to users.

Conntour further states that its system offers versatile deployment options, capable of operating entirely on-premises, completely in the cloud, or through a hybrid configuration. It is designed to integrate seamlessly with most existing security infrastructures or to serve as a comprehensive, standalone surveillance platform.

A longstanding challenge within the video surveillance industry, however, remains the fundamental limitation that the effectiveness of surveillance is directly tied to the quality of the captured footage. For instance, discerning crucial details from video recorded in a dimly lit parking lot by a low-resolution camera with an obscured lens often proves difficult.

Goldner explained that Conntour mitigates this inherent challenge by assigning a confidence score to its search results. If the input from a camera feed is deemed to be of insufficient quality, the system will accordingly present results with lower confidence levels.

Looking ahead, Goldner identifies the paramount technical hurdle as integrating the full spectrum of Large Language Model (LLM) capabilities into their system without compromising its current efficiency.

He elaborated, stating, "We have two things that we want to do at the same time, and they contradict each other. One one hand, we want to provide full natural language flexibility, LLM-style, to let you ask anything. And on the other hand there’s efficiency, so we want to make it use very few resources, because again, processing [thousands] of feeds is just insane. This contradiction is the biggest technical barrier and technical problem in our space, and what we’re working really, really hard to solve."