Introducing Deep Investigation, a new capability that transforms video search and analysis into an operational intelligence tool.

With Deep Investigation, organizations can ask complex questions across activity captured throughout the Coram unified physical security platform, and Coram's autonomous agents analyze video and connected data across cameras, locations, and time periods to investigate what happened, identify patterns, quantify findings, and deliver a synthesized, evidence backed answer.

This expands the value of Coram from finding incidents to understanding operations. Security teams can investigate faster, while operations, facilities, safety, and loss prevention leaders can uncover patterns, quantify business impact, and make better decisions from the visual data they already capture.

Go to Coram Assistant to try it now, or click here to learn more.
​

Maintain multi-session conversations and go hands-free with voice mode.

Your conversations now also persist across sessions. Return to previous chats, see which have new messages, re-run past video searches with a single click, rename conversations, and share them with colleagues via deep link. Chat history works across web and mobile, so you can start a search at your desk and pick it up on your phone.

The mobile app also adds a full voice mode. Tap the microphone, speak your question, and the Assistant responds with audio; ideal for security operators on patrol or facility managers doing walkthroughs.

Click here to learn more.
​

Protect access to sensitive footage by controlling which displays can stream your kiosks.

Kiosks gain a new Secure Mode. When enabled, a display opening the kiosk URL sees a six-digit access code instead of live video. An administrator enters that code in kiosk settings to authorize the display, and the feed only begins streaming after approval.

A new Paired Devices section shows every authorized display with the ability to revoke access at any time. The kiosk admin page has also been redesigned with a cleaner table layout, and management actions are now restricted to kiosk owners and admins.

​Click here to create a secure Kiosk now, or learn more here.
​

A clearer, faster way to scrub, browse, and investigate detections on your phone.

The mobile app gets a completely rebuilt camera timeline this month. A new horizontal layout makes scrubbing more intuitive, with color-coded detection bars showing exactly when people, vehicles, faces, and motion events were captured.

Tap View Detections to open a full-screen detection browser, where you can filter by category and narrow results by date and time. Tap any detection thumbnail to jump straight to that moment in the video. This brings the full detection browsing experience from the desktop app to mobile for the first time.

From a detection, you can also launch View Journey to trace a person's path across multiple cameras, select the cameras you want to include, and create a journey right from your phone.

The clip player has also been redesigned with fullscreen support and playback speed options, and PTZ control on mobile is now near-instant.
​

Videos load faster. Scrubbing is smoother. Walls perform better.

Coram's video streaming engine has been completely rebuilt. The result: videos start playing almost instantly when you click on a camera, scrubbing the timeline feels seamless with no loading gaps between segments, and Walls load faster with a more efficient connection model for multi-camera views. These improvements apply across Live View, Timeline, Walls, and Kiosks on both web and mobile.
​

More focused firearm detection, smarter face matching, and more accurate license plate reads.

Firearm detection now checks multiple frames before confirming an alert, instead of triggering on a single frame. This significantly reduces false positives from objects that momentarily resemble a weapon, one of the most common sources of unnecessary alerts.

Face matching is more precise: only high-confidence matches now link a face to a surveillance profile, and ambiguous cases are evaluated more carefully instead of being forced into a yes-or-no decision.

License plate recognition now uses real confidence scoring, so low-quality reads are properly filtered instead of surfacing as false positives. Plates that were previously missing from AI Analytics results now appear correctly.
​