What's New in Ambient Platform: An AI Agent for Every Camera, a Connected Case Story, and Double the Stream Density

At a Glance

The Ambient Platform reasons alongside your security team. It goes beyond simple recording to actually help you understand what's happening. Today, we're introducing a set of new capabilities that push understanding further:

• Agentic Video Walls give you an AI agent that continuously monitors every camera and surfaces every 60 seconds what matters most.
• Double the stream density on the same hardware for our AI-native VMS environments.
• A whole set of camera management features gives operators and IT teams direct control over camera health, credentials, and connectivity.
• Case Management turns scattered clips into one connected incident story in minutes.
• Sharper results for natural language searches with Semantic Search and a simpler way to trigger a Similarity Search.

Here's what's new, and why it matters.

AI-Native VMS: More Intelligence, More Scale, More Control

Sites add cameras faster than teams can watch them, hardware budgets stretch further with every expansion, and routine maintenance competes for the same IT and engineering time as everything else. Solving one of these problems in isolation doesn't solve the others. An operator with more attention still hits a hardware ceiling, and more hardware doesn't help if maintaining it still means filing a ticket. This release addresses all three at once, bringing more intelligence to how every camera gets watched, more scale from the hardware already in place, and more control handed directly to the teams running it.

Agentic Video Walls: Focused, Always-On Situational Awareness

A traditional video wall is built to live in an operator's peripheral vision, hopefully catching their attention at the right moment. At scale, that wall holds maybe 20 streams at once. On a 1,000-camera campus, that's blindness to the other 980. The site feels covered because the streams are lit up. However, that coverage is mostly an illusion as there's likely nothing to catch on any given feed. And footage sitting in storage doesn't stop an incident if nobody's watching when it happens.

Agentic Video Walls, powered by Ambient Pulsar, is an AI agent that perceives every connected camera, reasons about what it's seeing, and assesses what's actually worth attention. Once every 60 seconds, it surfaces the single most relevant moment happening across the site, with a plain-language description attached. Operators stop scanning a fixed grid and start reviewing what the agent has already decided is worth a glance.

The agent infers each camera's context automatically and requires no manual setup. A perimeter camera reads differently than an interior corridor from the first day it's connected. Operators shape what the agent surfaces going forward through two controls, Pin and Dislike:

1. Pin fixes an event at the top of the wall and tells the agent to weigh similar activity at that camera or location more heavily.
2. Dislike deprioritizes a pattern. Because there's nothing to configure and no thresholds to tune, the agent delivers value from day one.

With Agentic Video Walls in place, security teams stop spending their attention scanning a wall of static feeds, hoping something catches their eye. They spend it instead on judgment and response, acting on the moment the platform has already found rather than watching and waiting for it to appear.

Double the Stream Density for AI-Native VMS

Camera fleets are growing faster than the hardware budgets to support them. That's true whether a security team is adding cameras to an existing site, standing up a new site, or trying to control the hardware costs of an expansion. What customers need is more density from the appliances they already have, not just more appliances.

Scaling up capacity shouldn't require scaling up hardware. Ambient Foundation now introduces a set of optimizations that support up to 2x the number of camera streams on a single Ambient Edge Appliance, with no new hardware required. That's possible because Ambient Foundation runs a smaller, more focused application, which frees up compute and memory headroom per stream.

Existing Ambient Foundation customers get up to 2x the streams on the appliances they already own. New customers, or new sites, need roughly half the number of appliances to cover the same stream count. Either way, camera fleets can grow faster than the hardware budget attached to them.

These optimizations ship as a software update to the Ambient Edge Appliance, instead of a hardware refresh. That means they apply automatically to appliances already deployed in the field, not only to new installs. A security team no longer has to choose between growing camera coverage and controlling hardware spend, since the appliance already on the wall simply does more.

Camera Management: Health, Degradation, PTZ, and Credentials

Scaling down overhead means fewer dependencies on other teams for routine work. As camera fleets scale into the thousands, routine tasks such as renaming streams, rotating credentials, fixing connectivity, or spotting a degraded view multiplies. Each of these has typically required a ticket to engineering or support, which slows down maintenance right when it's needed most.

Ambient Foundation now closes that gap with a set of camera management capabilities built for exactly this kind of work. Operators and IT teams get direct control over camera health, credentials, and connectivity right in the product:

• Camera Health Notifications: configurable daily or weekly digest emails per site, with unhealthy-duration thresholds.
• View Degradation Detection: a third health state, alongside healthy and unhealthy, that flags a stream connecting fine while detection is compromised by blur, low light, or an obstruction.
• PTZ Controls: remote pan, tilt, and zoom with configurable step size, plus a filterable indicator for which streams support it.
• Bulk Stream Renaming: rename streams in bulk via CSV upload.
• In-Product Network and Credential Updates: update a stream's network settings or credentials without losing its historical footage and alert history, so it never needs to be re-onboarded as new.

Speed Up Investigations and Time to Report by 20x

After an incident, reconstructing how someone moved across an entire site is a manual process. It usually means jumping from camera to camera, scrubbing footage by hand, and keeping track of which camera name maps to which location. A ten-minute incident can turn into several days of manual investigation and reporting, right when speed matters most.

Case Management: From Incident to Shareable Story in Minutes

Case Management is the Ambient Platform's capability that lets operators turn scattered clips from across every camera into one connected, shareable incident story in minutes. It follows the same arc as the investigation itself: track, record, report. Operators follow a person across cameras using natural language with Semantic Search, or manually with video scrubbing. As they find clips, they save them directly into a case, with no hand-tracking of timestamps. When it's time to report, Case Management auto-sorts every clip chronologically and captures metadata like time, site, and camera automatically. One click rolls every clip description into a single AI-generated case narrative, fully editable before it goes anywhere.

Sharing is built for how incidents actually get handed off. It includes expiring links, read-only external access that's revocable anytime, and a full audit trail of who viewed or received the case. Case creation and editing are role-gated, so this scales to teams and to formal chain-of-custody requirements without extra process.

Building a case well is easier when the tools underneath are sharper. Two updates in this release make that especially true: a streamlined Semantic Search experience, and faster, more targeted scrubbing.

Forensics Streamlining: Faster Paths to the Right Match

Semantic Search is the platform's natural language search engine, and this release sharpens the results experience on three fronts. Matching frames of the same person or vehicle across a five-minute span now roll up into a single grouped result instead of surfacing every frame individually, so what used to take browsing through dozens of near-identical results now takes a click or two. Results can also be filtered down to an individual camera or a group of cameras.

Similarity Search is the Ambient Advanced Forensics capability that takes a single confirmed sighting of a person and finds every other appearance of it across the site, and this release adds a faster way to launch it. From any frame already open in the regular video scrubbing view, operators can select the AI Magic Wand to highlight everyone the platform detects in that frame, then click directly on the person of interest to launch Similarity Search. What used to require finding that same person through a search result now starts from wherever the operator already is.

The recorded-video player itself also got a simplified, timeline-first redesign this release, letting operators fast-forward up to 5x and jump directly to scenes where people or vehicles were detected, so tracking a person or vehicle, or finding one specific moment, takes less time and less re-orienting across cameras.

One Platform, Built for Agentic Physical Security

What ties these capabilities together is that they all run on one integrated platform. An operator monitoring the video wall, an investigator building a case, and an IT admin managing camera health are all working from the same live data, the same footage, and the same system of record, rather than reconciling three tools that each see part of the picture. Everything in this release ties to the same Ambient Edge Appliance that connects to your cameras, the same Cloud SOC that allows you to operate, and the same underlying intelligence that powers all our capabilities, starting with Ambient Foundation as our AI-native VMS, and then moving to advanced capabilities available through Ambient Advanced Forensics, Ambient Access Intelligence, and Ambient Threat Detection. That means adopting more of the platform is never a migration, but a simple setting you turn on. You start where it makes sense today, and everything you add later works the way the rest already does, each capability making the next one stronger, until what you're running isn't a stack of tools but a single system for Agentic Physical Security.

If you think these capabilities would be valuable to your operation, get in touch with Ambient.ai for a live demo.

If you plan to attend GSX 2026, stop by our booth (#3923) to see these latest capabilities and the rest of our platform updates in person.