County Public Safety Agency Cut Average Video Release Time

At a Glance

A county public safety agency with more than 200 sworn officers was taking about 150 public information requests a month. Two staff handled disclosure, and the 28 percent of requests involving video or audio consumed roughly three quarters of their hours.

Video requests were taking an average of 62 days to release, and written notices that a deadline could not be met went out about 50 times a month, one request in three.

Investigators had a second problem. Their systems could search words but never pictures. If a vehicle drove through the frame and nobody mentioned it on the recording or wrote it into a report, no search would surface it. The only way to know it was there was for somebody to watch the footage.

The agency deployed VIDIZMO Redactor and VIDIZMO AI Intelligence Hub on its own infrastructure, alongside the cloud evidence management system it already ran. Nothing was replaced.

In the first six months, average video release time fell from 62 days to seven. Those notices fell to about 12 a month, and the agency has met every date it has set since. Investigator effort on a case fell from about eight hours to 45 minutes, and prosecutor packages moved from three business days to the same day.

Agency Profile

The Challenge

Everything below describes how the agency was operating before it deployed VIDIZMO.

Two Staff, and a Clock That Did Not Stop

The agency's body-worn camera program started in 2022. By the time 175 body-worn and in-car units were in service, the disclosure unit was taking in about 150 public information requests a month. Two staff handled all of it, video, audio and documents alike.

Video and audio were the minority of the volume and most of the work, 28 percent of requests and 76 percent of redaction hours.

Video took that share of the hours because of how long it takes to redact by hand. A typical video request runs five to ten hours of work for each hour of footage, and a crowded scene with several objects to keep tracked runs fifteen to thirty hours. One request, a 30-minute foot pursuit recorded on a single body-worn camera, took eight to ten hours, because the pursuit crossed crowded areas and the footage caught license plates, house numbers, and bystanders throughout.

Each of those calls carried a legal basis. A bystander and a juvenile in frame are treated differently. A plate on an uninvolved vehicle identifies someone who was never part of the case. An address read out over dispatch never appears on screen at all. The reviewer decided each one and had to account for it if a release was questioned, and nothing recorded the reasoning.

With two people and that many hours of work behind each video request, the requests queued. Video took an average of 62 days to release, against nine days for documents. A hundred and thirty requests sat open at any given time, and 75 of those had passed the date the agency had committed to in writing. State law required the agency to produce records promptly, and to send the requester a written notice naming a new date whenever it could not meet the deadline. Those notices were going out about 50 times a month.

The agency had looked at automating this before. Its evidence management system included a bundled redaction capability, which the agency did not consider reliable enough to run without a reviewer checking the output. At renewal it chose not to continue with that capability, and looked instead for a redaction tool that would work alongside the system it already had.

A Case File Nobody Could Read

A single vehicular homicide file held over 300 photographs and 70 to 90 hours of video. The evidence was there. Getting an answer out of it was the problem.

Analysis was limited to search. If something was said on a recording, an investigator could find it in the transcript, and could search keywords and the section markers a long recording had been split into. If something was only visible, a vehicle in the background or a person walking past, there was nothing to search against.

So the work was done by watching. An incident with 30 cameras in the surrounding area meant reviewing all 30 to work out which one held the angle that mattered, before the investigation itself could begin. Checking whether a suspect's account after arrest matched an earlier one meant playing both interviews from the beginning. Establishing whether a vehicle had come up in another case meant remembering that it had.

Where a recording had no audio, and fixed cameras rarely have any, there was no transcript to review and no keywords to match. Those files could only be watched manually.

Building a case took about eight hours of investigator time, spent watching material to find out what was in it and querying the evidence system, records management, and dispatch one at a time. Assembling a package for the prosecutor took three business days.

The Solution

Redaction Inside the Workflow They Already Had

Redactor is integrated with the agency's evidence management system over its API, so evidence moves into the redaction queue and the released copy comes back without anyone exporting a file. Records staff work in a browser, with nothing to install.

Detection covers faces, license plates, and people in frame, and catches spoken personal information on the audio track. The unit built templates for the footage types that recur, among them traffic stops, foot pursuits, booking and custody, interview room recordings, and crowded scenes at public events, with several others, so a reviewer starts from settings that suit the material rather than configuring each request from scratch.

Each redaction sits on its own layer, annotated with the exemption it was made under and a note giving the reason. That was the part the old workflow could not produce, and it means the basis for a release is now in the file. Footage submitted at the end of the day is processed overnight and waiting for review the next morning.

One Question Across Every System

The agency's evidence management system already indexed the words. It transcribed speech and let staff search the transcript, the keywords, and the section markers. What it never did was look at the picture.

AI Intelligence Hub pulls media from that system through its API and analyzes it for what the system never captured, detecting the people, vehicles, weapons, and objects that appear in the footage. The evidence itself stays where it was, and the evidence management system remains the system of record.

An investigator can now ask a single question covering the evidence system, records management, and dispatch at once, and get an answer showing which file it came from and where in that file, even when what they are looking for was never spoken aloud or written down.

That changes where an investigator starts. Comparing a suspect's account after arrest against the one given earlier used to mean playing both interviews from the beginning. The Intelligence Hub now points to the passages in each where the subject comes up, so the investigator begins there, reads them side by side, and decides whether the accounts conflict. The full recordings stay available and nothing stops an investigator watching either one end to end. What changed is that finding the relevant minutes no longer takes hours.

Faced with the 30 clips from those cameras, an investigator can ask which of them show the person or the vehicle in question and start with those. The Intelligence Hub reports what it found and where; which angle carries the case is still the investigator's call.

A recording with no audio used to mean watching the whole thing, because there was no transcript to search. The Intelligence Hub analyzes the picture, so a silent camera feed can now be queried the same way a body-worn recording with speech on it can.

Asking whether a vehicle or a person has come up in another case returns the files and the points inside them, across the cases that investigator is cleared to see, rather than depending on somebody remembering that it had.

Evidence can now be reached two ways. Someone can open it in the evidence management system, or they can ask the Intelligence Hub. The question that matters is whether asking the Intelligence Hub could surface something the evidence management system would have kept from that person. It cannot. Each item carries the access and sealing status of the system that owns it, kept current, so sealing a record in the evidence system also takes it out of the Intelligence Hub's results. A search against a connected system runs under the credentials of the person asking, so that system applies its own rules. The index sits on the agency's servers under the agency's retention policy.

Where the Intelligence Hub generates a summary, it cites the timestamps it drew from and the investigator reviews those passages directly. Workflows carry human review and approval steps, and the finding of record is the investigator's.

What Changed

Figures below are reported by the agency and cover the first six months after deployment, measured against the six months before it.

The unit processed about 340 video and audio requests in that period, 252 new and roughly 90 drawn from the backlog, against about 250 processed by hand in the six months before.

Release times. Average time to release a video request fell from 62 days to seven. Documents fell from nine days to four. Across all requests the average went from 24 days to five.

Review time. Median review time for a video request fell from two and a half hours to 25 minutes. The mean fell from four hours to 40 minutes. The gap between the two figures is the handful of multi-camera incidents each month that still take twelve to fifteen hours. The 30-minute foot pursuit that had taken eight to ten hours now takes about an hour and a half.

The queue. Open requests fell from 130 to 40. Requests past the date the agency had committed to in writing fell from 75 to three.

Deadline notices. Written notices that a deadline could not be met fell from about 50 a month to roughly 12. The ones that remain are requests where the agency is waiting on a legal ruling about whether information may be withheld, and requests carrying an unusually large volume of records. Neither is backlog. The agency has met every date it has set since.

The same two people handle the work, and the late sittings have all but stopped.

Investigator effort on a case fell from about eight hours to 45 minutes, most of the saving coming out of the time that used to go on watching material to find out what was in it. Assembling a package for the prosecutor moved from three business days to the same day.

The agency is pursuing CALEA accreditation, and the exemption record and the named-user access log support the documentation its assessors ask for.

Why VIDIZMO

The agency needed two things. A redaction tool fast enough for the statutory clock, and one that would work with the evidence system it already had rather than replace it. And a way to ask one question that covered the evidence system, records management, and dispatch at once, including what appeared in footage and not only what was said, with every answer showing which file it came from.

VIDIZMO LLC is ISO/IEC 27001:2022 certified (certificate RA-2507091), and its processes and software are aligned with NIST SP 800-53 (self-attested). Platform controls including AES-256 encryption at rest, TLS in transit, role-based access control, MFA and SSO, and audit logging support the agency's CJIS compliance obligations. VIDIZMO's features align with FIPS 200 minimum security requirements and use FIPS 140-3 validated cryptographic modules on supported infrastructure; FIPS 140-2 validations remain recognized during the NIST transition.

Book a walkthrough to see these workflows applied to your own evidence and records.

August 2026. Document version 1.0. Figures are reported by the agency and cover the first six months of production use at a single deployment; outcomes vary with request volume, evidence type, and configuration. © 2026 VIDIZMO LLC.