Ambient.ai Launches Pulsar, a New Vision-Language Model for Physical Security
Ambient.ai has released Ambient Pulsar, a new AI engine the company describes as its most advanced model to date for physical security applications. The system is now generally available and is designed to deliver agentic monitoring and investigation capabilities for enterprise environments.
According to and announcement, Pulsar is built on Ambient.ai’s edge-optimized reasoning Vision-Language Model architecture and is trained on more than one million hours of enterprise video. The company states that the model processes more than 500,000 hours of video each day and is engineered to reason like a security operator at machine speed and scale.
Cary Monbarren, Senior Director of Corporate Security at SentinelOne, said in the release that the platform is reducing manual triage and response times for GSOC teams. Ambient.ai CEO and co-founder Shikhar Shrestha said the new model enables cameras, sensors and SOC environments to operate as intelligent agents capable of prioritizing and responding to activity in real time.
Platform capabilities
With the launch, Ambient.ai introduced several new platform features powered by Pulsar. These include the Agentic Video Wall, which highlights video streams with notable activity, Activity Notifications that use natural-language inputs to define events or scenes, and Semantic Search, which allows operators to query video archives through questions.
Two capabilities are also available in early preview. Agentic Investigations is designed to assemble incident timelines using tools such as Semantic Search, Similarity Search, and License Plate Recognition. Custom Threat Assessment enables security teams to adapt alert severity through natural-language input.
The company describes Pulsar as an open-set, fully agentic model developed for physical security. Key features include always-on edge reasoning, open-set detection, contextual intent recognition, scalable autonomy for multi-site deployments, and an edge-optimized architecture that uses Ambient.ai’s appliance powered by NVIDIA AI infrastructure.
Broader vision
Ambient.ai states that Pulsar strengthens its approach to Agentic Physical Security by unifying monitoring, threat assessment, access intelligence, investigations and response under a single AI system. According to the announcement, enterprise customers across several sectors are reporting reductions in false alarms and faster alert resolution times.
The company positions Pulsar as the core of its Ambient Intelligence engine and asserts that the technology is intended to identify precursors to risks, support proactive prevention, and coordinate autonomous response and investigation.