Cobalt AI launches CMI SaaS for customer-managed security operations

Cobalt AI on Sept. 15 unveiled CMI SaaS at GSX 2026 in Atlanta, expanding its Monitoring Intelligence into customer-run security operations centers. The platform aims to cut alarm noise, speed verified responses and let enterprises keep control of their cameras, data, workflows and governance.

Why it matters: - CMI SaaS gives enterprises an AI judgment layer before alerts reach operators, which can cut time spent reviewing low-value alarms. - The offering is aimed at security teams that want AI assistance without handing over infrastructure, data or response control. - Cobalt AI says its monitoring intelligence has reduced unwanted alarm noise by up to 90% in real-world deployments.

What happened: - Cobalt AI announced CMI SaaS on Sept. 15, 2026. - The company introduced the product at GSX 2026 in Atlanta. - CMI SaaS brings Cobalt Monitoring Intelligence into customer-managed security operations. - The platform is available today.

The details: - CMI SaaS is designed for enterprises that run their own security operations centers. - The system evaluates events before they reach an operator. - It compares camera activity, access-control information and company security policies to decide whether an event can be cleared, needs corrective action or should be escalated. - Security teams receive verified events with context, root-cause analysis and documented reasoning instead of raw alerts. - Customer teams keep control of infrastructure, camera feeds, monitoring workflows, governance requirements and response procedures. - The platform supports multiple sites, business units, access-control environments and security policies under one operating model. - CMI SaaS integrates with Lenel, C-CURE, Genetec, Milestone and other enterprise systems through bidirectional APIs. - Companies can use the platform without replacing existing cameras or access-control systems. - CMI SaaS can be managed by an internal GSOC or the Cobalt AI Command Center. - Edge processing is available when needed for bandwidth, network architecture, data policy or internal security standards. - Cobalt AI says CMI can automatically clear 90% to 95% of alarms while documenting the reasoning behind each disposition. - The company says verified events can reach operators in less than 75 seconds, compared with an industry benchmark of about 6 to 8.5 minutes per event.

Between the lines: - The launch pushes physical security software beyond detection and toward AI-assisted judgment. - The pitch is aimed at organizations that already have mature security operations and want better triage, not a replacement for their current stack. - Preserving reasoning and disposition records may matter as much as faster response, since it supports review, governance and continuous improvement.

What's next: - Cobalt AI is demoing CMI SaaS at GSX 2026 from Sept. 14 to Sept. 16 at the Georgia World Congress Center. - The company is at Booth 3519. - Amrit Depaulo, head of product, is scheduled to present “Alarm Ops” on Sept. 15 at 2:45 p.m. in Room B405/B406A. - Ken Wolff, CEO of Cobalt AI, is scheduled to present “From Guarding to Orchestration: The New Operating Model for Physical Security” on Sept. 16 from 9:45 to 10:45 a.m. EDT in Room B308/B309. - Interested buyers can schedule a demonstration.

The bottom line: - Cobalt AI is betting that security teams want AI that decides, explains and routes alarms inside their existing operations, not just software that detects them.