There’s a particular kind of dread that comes with a discovery packet containing six hours of CCTV footage. No audio. No timestamps you can trust. Just a timestamp counter and the quiet resignation that someone has to scrub every second of it. In 2024, Colorado’s Office of the State Public Defender found digital evidence in state public defender offices has surged 4,500% since 2016. Silent surveillance video is a significant driver of that growth. Meanwhile, felony cases can now include up to 200 hours of video footage, making exhaustive manual review not just inefficient but practically impossible.
The hours lost to mechanical video review aren’t just an inconvenience. They’re advocacy time that clients never get back.
The Hidden Cost of the Digital Tsunami
When we talk about the digital tsunami overwhelming public defender offices, we usually think about body-worn camera footage or jail calls. But silent video (CCTV, parking lot surveillance, retail cameras) carries its own specific burden. Unlike body-cam footage, there’s often no accompanying report telling you where to look. Attorneys and investigators watch from the beginning, hoping to find a 90-second window buried in hours of nothing.
Defenders spend hours per case dealing with audiovisual evidence, including locating relevant moments and documenting what they found. In offices already managing caseloads well above recommended limits, that math doesn’t work. Every hour spent scrubbing footage is an hour not spent on motions, client contact, or trial prep. That’s administrative drag in its most concrete form.
From Raw Footage to Relevant Frames, With AI Doing the Heavy Lifting
ZLS.app’s new AI Vision Analysis engine changes the equation for silent video. Rather than passively playing a file, it actively watches it, analyzing frames to identify people, clothing, objects, and actions, then generating a factual, chronological narrative with timestamps:
“04:12 – Subject A enters vehicle. 04:14 – Vehicle departs northbound.”
This creates a searchable visual transcript, the same concept as audio transcription applied to what the camera sees. Defenders can jump directly to moments that matter, review an AI-generated summary of the full recording, and annotate relevant frames, all within the case file.
It’s also compatible with audio transcription, so a single video file with both audio and silent sections can be processed end-to-end without switching tools or workflows.
Data as a Byproduct of Defense Work
There’s a secondary value here that’s easy to overlook: structured data. Every AI-generated narrative becomes a piece of documented case work, timestamped, searchable, and reviewable. For office leaders trying to demonstrate workload to county commissions or justify budget requests with evidence volume data, this kind of documentation matters.
The goal behind ZLS.app’s approach to AI tools is to unleash the data that already exists inside everyday defense work. Attorneys are already doing this review. AI Visual Analysis simply captures it in a form that’s useful beyond the individual case file. That’s the difference between data as a burden (yet another thing to log) and data as a byproduct of work defenders are already doing.
Reclaiming the Work That Actually Matters
AI tools in public defense succeed when they respect the complexity of the work and fail when they try to replace human judgment with shortcuts. AI Visual Analysis doesn’t tell defenders what to think about a video. It removes the mechanical hours standing between an attorney and the moments that deserve their attention.
Built by public defenders for public defenders, ZLS.app’s AI features are designed around one question: what takes time away from advocacy, and how do we give it back? Silent video review is one of the clearest answers to that question, and now there’s a practical solution that fits inside your existing workflow.
Please reach out if you’d like to talk through how it fits your office’s current evidence workflow.
