General operational and educational information for corrections professionals. Not legal, medical, or compliance advice, and not a certification of compliance with any law or standard. Policies and standards vary by agency and jurisdiction; follow your facility's policy and your own legal, medical, and professional advisors. The library is reviewed and updated monthly; see the changelog.
Safety Intelligence runs on the camera system a facility already has. What a deployment needs from the facility is mostly access, people, and a few hours of honest conversation: written authorization for specific feeds, a secure outbound-only connection, a walk of the building, a Day-0 baseline, and a jailer, an IT contact, and shift supervisors who know how the facility really runs. This page lays out each piece for the jailer and the IT administrator.
In most cases no new cameras are required. VPT works on the facility's existing camera feeds, and the access authorization the facility signs states that VPT adds no cameras. There is no hardware to mount, no wearables, and nothing attached to the people in custody. See why passive cameras are not enough for the reasoning.
During technical onboarding, the IT contact and VPT confirm the camera makes and models are compatible, identify the video management system or NVR, and identify the camera-system vendor of record with a direct contact and escalation path.
Where the site survey finds a view that is unusable or a zone with no camera at all, that is recorded as a finding for the facility to decide on. It is not a condition of starting.
The connection is outbound only. VPT does not ask the facility to open inbound ports on its firewall.
The IT administrator and VPT agree the secure access method, confirm that bandwidth and uplink are sufficient, and confirm a CJIS-aligned data path. Credentials are issued on least privilege with multi-factor authentication.
Access is limited to the feeds listed in the signed Network and Camera Access Authorization. Zones are mapped to authorized feeds only, PREA-excluded feeds are confirmed blocked, and the facility can revoke access in writing at any time.
An outage and failover notification path is set before go-live, so the facility knows who is told, and how, if a feed or the connection drops. IT teams can find more in the IT and security start page.
Before anything goes live, VPT walks the facility with the jailer. The walk covers the high-risk zones (intake and booking, isolation and segregation, medical and detox, high-risk housing, holding, movement corridors, and recreation or common areas), noting for each one which cameras are present and where the blind spots are.
The walk also lists the areas excluded by design: strip-search areas, toilets, attorney rooms, and other PREA-prohibited or private spaces. Those are written down so it is on paper that they are not covered.
The result feeds the deployment scope, which locks the cameras and zones the Audit will cover, starting with the highest-risk zones first.
The Day-0 baseline is captured at install, before tuning. It is the facility's starting point: the honest picture of how safety runs today, which every later Proof Report is measured against.
It is built from a short facility profile, an interview with the jailer and shift supervisors about how events are discovered and how rounds are documented today, a review of recent incident records the facility already keeps, and the coverage findings from the walk.
Where the facility's records do not contain exact timing, the baseline says so and labels anything reconstructed from interviews as reconstructed. Nothing is invented to make the before look worse.
The jailer or officer in charge: walks the facility, sets the zone priorities, joins the baseline interview, and approves go-live.
An IT contact: works through the technical onboarding checklist with VPT's technical lead and signs it off.
Shift supervisors: describe how things really work on each shift, including overnight, and help confirm how observations reach staff and get closed out.
An authorizing official: signs the access authorization naming the feeds and zones VPT may access.
Where the facility's CJIS authority requires it, VPT staff with access complete background check authorizations before access is granted.
Before go-live, VPT confirms the event-review workflow with the staff who will use it: how an observation arrives, who acknowledges it, and how it is closed with a note. A person reviews and decides before anything is acted on. See human-in-the-loop.
The system does not replace rounds, posts, or staff judgment, and training is framed that way.
No facial recognition setup. VPT does not use facial recognition.
No inmate biometric data, and nothing worn by or attached to people in custody.
No access to PREA-prohibited or private areas.
No inbound ports opened on the facility firewall.
No new cameras in most cases, and no rip-and-replace of the existing system.
For how a deployment proceeds after these pieces are in place, see how deployment on existing cameras works.
In most cases, no. VPT works on the facility's existing camera feeds and adds no cameras. The site survey records any unusable view as a finding for the facility to decide on.
No. The connection is outbound only, with a secure access method agreed with your IT administrator and least-privilege credentials.
PREA-prohibited and private areas such as strip-search areas, toilets, and attorney rooms. They are listed on the site survey and excluded from the access authorization.
The facility's starting picture, captured at install before tuning, from a facility profile, staff interviews, recent records, and the coverage walk. Later Proof Reports are measured against it.
The jailer, an IT contact, shift supervisors, and the official who signs the access authorization.
No. VPT does not use facial recognition and does not need inmate biometric data.
