A detective calls before a warrant deadline: what formats, how to package, where to deliver. The AI answers from your submission guidelines — correctly, at 4:45 PM or midnight.

“Where is my case?” answered exactly as your policy allows — stage and estimate for the authorized requester, escalation to the examiner for anything more.

Every call and chat produces a transcript tied to the case reference — who asked what, when, and what was disclosed. The intake desk becomes part of the record.

A suspect photo, a voice exemplar, a phrase of interest: the AI sweeps every file in the case for candidates and returns matches with timestamps and confidence context, documented for review.

The jacket from the CCTV still against the seized clothing photos; the tattoo across booking images; the vehicle across intersection cameras — matched with reviewable, documented results.

What was received, what was searched, what matched and where: the AI drafts the examination notes and report skeleton from the session itself — the examiner edits and signs.

CCTV exports, body-cam files, intercept audio, phone extractions, scanned custody forms — the ingestion pipeline turns the evidence room’s reality into queryable knowledge.

No source in the case libraries, no claim: the AI cites the file, the timestamp, the custody entry — or says it cannot answer. Nothing is invented into a case.

Examiners see their assigned cases; the lab director sees the unit; the intern sees nothing sensitive — enforced server-side, down to the field.

Media sweep, face-candidate review, voice comparison prep, transcript-and-phrase search, report draft: work types molded to your bench, with default protocols per type.

Rush warrants and court dates drive the queue: one-click views show what is urgent, late, today and tomorrow — a trial date never sneaks up on an unworked case.

Every work is a full session: reasoning streamed, findings timestamped and cited, milestones tracked, linked to the case and the calendar.

Media-handling SOPs, examination protocols, QA manuals, instrument logs: one library, controlled versions on top, superseded revisions outranked but archived for audit.

“What is the protocol for handling an encrypted device?” — answered in seconds with the SOP attached, to any authorized examiner, mid-examination or mid-testimony prep.

Work types carry their governing protocols: the AI running a media sweep reads from the media-examination SOP — by construction, not by memory.

New analysts learn your intake rules, media-handling SOPs and documentation standards through guided training — contributing in weeks, supervised properly throughout.

“How do I document a partial match?” — asked to the assistant, answered from the QA manual with the source attached, without pulling the senior examiner out of testimony prep.

Private chats work on any device with a login — the how-to knowledge follows the analyst from the bench to the evidence room to the courthouse hallway.

A submission arrives: case record opened, media inventoried, custody entry drafted, acknowledgment sent to the requesting agency with a reference number — before the custodian finishes the physical check-in.

New media triggers the sweep: targets searched, candidate segments marked, the triage summary filed — the examiner starts the morning with a map, not a haystack.

Court dates and rush requests drive reminders and status drafts automatically — the prosecutor’s “where are we?” email is answered before it is sent.

The report draft pulls exclusively from examiner-confirmed results: methodology, media inventory, findings with file-and-timestamp citations, limitations noted.

Weekly status digests to each requesting agency, drafted from the case queue by your disclosure rules — the detective stops calling, the examiner stops repeating.

Every outbound report and update lands in the approval queue. The examiner signs; the system records who approved what, when, and to whom it was delivered.

Cases, submissions and dispositions mapped field by field, with a dry-run preview before anything writes — push, pull or two-way, your choice per system.

Custody entries, storage locations and transfer records referenced by the AI when it drafts — the evidence system stays master of the chain.

A full API for whatever else the lab runs: digital-forensics tools, agency portals, court e-filing — data in, results out, under scoped keys.

Every access to case media and every AI operation on it is logged and searchable — the custody question about the digital copy has a documented answer.

Workgroups and field-level permissions segregate cases and units; the server redacts what a role cannot see — an examiner off the case sees nothing of it.

The owned lab server runs air-gapped with USB updates: chain-of-custody-grade custody where the media physically cannot leave — and no recurring fees on the hardware.

Ask the chat: “what are the submission requirements for CCTV media?” — watch it answer from sample procedures and offer the intake form.

Ask: “how would you sweep sixty hours of footage for one face?” — it walks the workflow it would run in your tenant: sweep, timestamp, cite, examiner sign-off included.

The full demo loads a forensic tenant with sanitized sample media: recognition sweeps, match review, report drafting — your bench, working.