In this article, we spoke with Jehan Wickramasuriya, Senior Vice President of Security and Resilience Software at Motorola Solutions, to further explore the technology.
Avigilon, a Motorola Solutions company, debuted their Visual Alerts technology last year and featured it prominently at
Intersec earlier this year. Leveraging generative AI and natural language capabilities, the technology seeks to make video security more intelligent and effective. In this article, we spoke with Jehan Wickramasuriya, Senior Vice President of Security and Resilience Software at Motorola Solutions, to further explore Avigilon Visual Alerts.
Visual Alerts is Avigilon’s on-premises, generative AI (GenAI)-based technology within the Avigilon Unity Video 8.7 platform. The solution expands the scope of object and visual event detection and helps organizations identify various safety, compliance, and logistical issues before they turn into highly critical events. Compared to previous versions of video management software and analytics, Visual Alerts include several new features including:
- Focus of attention: It activates Avigilon’s "Focus of Attention" interface, translating thousands of data points to present them as an intuitive, color-coded map that alerts security teams when they need to triage and respond;
- Automated action workflows: Detections are integrated into a robust rule engine, allowing users to configure automated responses that suit their distinctive needs;
- 100 percent on-premises processing: Unlike many cloud-based tools, the AI analysis happens entirely on-premises within the organization's infrastructure to ensure data privacy and compliance.
‘Making video security proactive rather than passive’: The use of Gen AI
According to Wickramasuriya, Visual Alerts represents a shift from static, rule-based analytics to adaptive, descriptive intelligence. A major element that enables this is generative AI, which is now trending highly in video security.
“A primary goal of Visual Alerts is to make video security proactive rather than passive. Traditional analytics are often limited to pre-defined objects like people or vehicles, but Visual Alerts uses generative AI to expand the system’s vocabulary, allowing it to detect complex scenes and specific activities that were previously undetectable,” Wickramasuriya said. “Generative AI is considered the fastest-growing segment in video security. It evolves video cameras from tools that just watch to real-time, operational intelligence that shifts the cognitive load from human operators to AI.”
Natural language interface
Another feature that makes Visual Alerts stand out is the use of a natural language interface, where the operator can simply use conversation-like input to set rules. Once an event is detected, the system can respond according to the predefined rule, for example triggering audible alarms, sending email snapshots, or initiating API calls to external security systems. All this supports a faster, orchestrated response.
“Unlike older versions where users had to navigate complex menus to set rules, Visual Alerts allows users to simply describe what to detect in a scene using natural language,” Wickramasuriya said. “It eliminates the need for specialized technical knowledge to create complex rules. Operators can type prompts such as ‘Alert me when there's smoke,’ ‘Alert me when a window is broken,’ or ‘Alert me when a person is lying on the floor.’"
Natural language can also play a key role in smart search where the user enters a query such as “find people wearing a red shirt,” and the system will retrieve all relevant footage. “While we do not currently support natural language searching, it is a use case we are actively exploring and expect to share more about in the future,” Wickramasuriya said.
Industries that can benefit
Visual Alerts is strong at workflow optimization and proactive incident prevention. With that, the technology well suits the needs of a variety of complex industries and vertical markets. These include:
Healthcare and hospitals: For large healthcare facilities, Visual Alerts helps security teams reconcile data from expansive camera networks to manage large volumes of people, understand different events simultaneously, and take immediate, verified action during incidents.
Oil and gas: In complex industrial sites, Visual Alerts can play a key role in upholding safety, meeting compliance and detecting hazards. Operators can create highly specific alerts, for example immediately flagging an "unauthorized vehicle near a shut-off valve".
Logistics and warehousing: Visual Alerts provides crucial, early hazard detection and workflow management. This can save vital time when the early signs of smoke or fire are detected in a warehouse, helping security teams to conduct a safe evacuation.
Retail and public venues: Large public venues can utilize the system’s generative AI for advanced crowd management, identifying "high crowd density,” helping operators to proactively manage pedestrian flow and identify potential safety bottlenecks.
Privacy first
Visual Alerts is all about data privacy. As mentioned earlier, the technology is designed with a privacy-first architecture that ensures generative AI video analysis is performed entirely on-premises.
“By utilizing the AI Appliance 2X with its Avigilon Hardened OS, sensitive video data is processed using local models without the need to send data to the cloud. This server-side approach allows organizations in highly regulated sectors, such as healthcare and critical infrastructure, to deploy advanced AI capabilities while maintaining strict data sovereignty and compliance,” Wickramasuriya said.
Furthermore, Visual Alerts provides privacy-aware AI analysis that focuses on events, not identity, allowing administrators to blur out individuals or specific, sensitive areas in the camera's field of view in real-time.