AI + a16z Building the “See Something, Say Something” AI for Every Camera
50 snips
Dec 16, 2025 
Shikhar Shrestha, CEO and cofounder of Ambient, discusses how a childhood trauma pushed him to revolutionize security with AI. He shares how Ambient's technology empowers cameras to detect and prevent threats in real-time. Shikhar explains the development of their unique reasoning vision-language model, Pulsar, and how it efficiently analyzes footage while respecting privacy. The conversation touches on real-world incidents that Ambient has averted, the challenges of operating during COVID, and the critical need for an AI-driven future in physical security.
AI Snips
Chapters
Books
Transcript
Episode notes
Childhood Robbery Sparked The Mission
- Shikhar Shrestha describes being held at gunpoint at age 12 while staring at a camera and hoping someone watched.
- That memory inspired him to build Ambient to make every camera a real-time "see something, say something" node.
Make Every Camera A Smart Node
- Cameras vastly outnumber human watchers, making manual monitoring ineffective.
- Turning each camera into an intelligent node lets security teams respond to precursor behaviors and prevent incidents.
VLMs Are The Game Changer
- Vision-language models now link image backbones to language reasoning, delivering human-level scene understanding.
- That leap lets Ambient plug in a more powerful engine and immediately improve detections and automation.
