Cerrion raises €16M to turn factory cameras into AI agents tackling €1.2T downtime crisis
Factory downtime costs the global manufacturing industry €1.21 trillion annually—a 319% increase since 2019 as supply chains grow more complex and energy prices rise. Zurich-based Cerrion raised €15.6 million ($18M) Series A led by Creandum to scale AI video agents that transform existing factory cameras into intelligent production monitors.
The funding comes amid escalating pressure on manufacturers to reduce unplanned downtime while managing rising operational costs. Traditional monitoring relies on human operators watching dozens of screens or reactive maintenance after problems occur—creating blind spots that lead to cascading failures.
The Manufacturing Monitoring Bottleneck
Manufacturing downtime has reached crisis levels. According to Cerrion’s data, the global cost increased 319% since 2019, driven by supply chain fragmentation, energy price volatility, and increasingly complex production environments. Traditional surveillance systems capture everything but understand nothing—requiring human operators to monitor endless video feeds for anomalies across vast factory floors.
The bottleneck lies in the gap between data capture and actionable intelligence. Factories have cameras everywhere but lack real-time analysis capabilities that can detect quality issues, safety risks, or process deviations before they cascade into costly shutdowns. Most monitoring remains reactive rather than predictive, with problems discovered only after significant damage occurs.
Video AI Agents as Industrial Infrastructure
Cerrion’s approach converts existing factory cameras into autonomous AI agents that understand production processes. Rather than requiring new hardware installations, the platform integrates with current surveillance infrastructure to provide real-time monitoring, alerting, and intervention capabilities.
The technical architecture spans three layers: computer vision models trained on industrial environments, natural language interfaces for human-AI collaboration, and integration APIs that connect to existing manufacturing execution systems (MES) and enterprise resource planning (ERP) platforms. This modular design enables deployment across diverse factory environments without requiring extensive system overhauls.
Video agents monitor what human operators physically cannot see—microscopic quality defects, early signs of equipment wear, or subtle process variations that indicate impending failures. The system triggers alerts, slows or stops machines autonomously, and updates relevant personnel instantly when anomalies are detected.
Enterprise Validation Across Industries
Cerrion is deployed in production environments at manufacturers including Unilever, Riedel, Schott Zwiesel, Stölzle Lausitz, Sisecam, and Verallia across glass, food, timber, and consumer packaged goods industries. These manufacturers supply products to major brands including Pepsi, Coca-Cola, Pfizer, and Novartis.
The company reports 50% faster problem detection and resolution, cutting both downtime duration and scrap losses in half. Revenue grew 10x since 2024, driven by strong adoption from US manufacturers alongside consistent demand across Europe and Latin America.
Enterprise validation demonstrates the platform’s ability to integrate with existing workflows while providing measurable operational improvements. The diversity of industries—from pharmaceutical glass manufacturing to food processing—indicates the broad applicability of video AI agents across production environments.
Infrastructure Category Emergence
Cerrion’s funding reflects broader recognition that manufacturing monitoring represents a distinct infrastructure category requiring specialized AI capabilities. Unlike generic computer vision applications, industrial monitoring demands domain-specific models that understand production processes, safety protocols, and quality standards.
The Series A included participation from Y Combinator, Goat Capital (Justin Kan, Founder of Twitch), 10x Founders, and Session VC, with notable angels including Harry Stebbings (20VC), Thomas Wolf (Founder of Hugging Face), Garret Langley (Founder of Flock Safety), and Filip Kaliszan (Founder of Verkada).
This investor composition signals market recognition that video AI agents represent critical infrastructure for industrial digitization. Rather than point solutions addressing specific manufacturing problems, the technology enables comprehensive factory intelligence that scales across production environments.
Looking Forward: From Monitoring to Orchestration
Cerrion plans to expand beyond vision-based monitoring to create comprehensive factory orchestration platforms. The company aims to double headcount across Europe and the US while extending capabilities beyond video analysis to include sensor fusion, predictive maintenance, and autonomous production optimization.
The technical roadmap includes integration with robotic systems, IoT sensor networks, and industrial automation platforms—potentially enabling factories that self-optimize production parameters in real-time. As manufacturing complexity continues increasing, video AI agents may evolve from monitoring tools into orchestration platforms that coordinate entire production ecosystems.
Founded by Karim Saleh (former Captain of Egypt’s national Water Polo team turned startup builder) and Nikolay Kobyshev (founder of Assaia, deployed across 40+ airports globally), Cerrion assembled its engineering team from ETH Zurich, Google, and EPFL. The combination of domain expertise and technical talent positions the company to capture the emerging category of industrial AI orchestration.
Manufacturing represents one of the largest infrastructure transformation opportunities for AI agents, with video intelligence serving as the foundation for factory-wide orchestration systems. For organizations building agent infrastructure platforms, tools like Overclock provide orchestration capabilities that enable seamless coordination between monitoring agents, production systems, and human operators—essential for scaling industrial AI deployments across enterprise environments.