04.1 · Broadcast and news
Forty years of bulletins, findable from the desk.
Newsrooms do not have a footage problem; they have a finding problem. The rushes exist, the tape restores exist, the Media Asset Management (MAM) system knows a title and a date — and the producer needs the clip where the minister said the word, by four o'clock. Manual logging has never kept pace with acquisition and never will.
We index in place from ingest and central storage, write scene, speech and on-screen-text fields back to your MAM as sidecar metadata, and let the desk search by description. The Non-Linear Editing (NLE) suite receives the exact frame. The Master Control Room (MCR), the Central Apparatus Room (CAR), signal processing and the video server sit outside anything we touch: we are never in the transmission path, and we say so in writing.
The archive stops being a storage bill and starts behaving like a licensing desk.
- Reads from
- Ingest chain, central storage, LTO restores, the MAM
- Hands to
- The NLE, as the exact frame plus logged clip and still exports
- Never touches
- MCR, CAR, signal processing, video server and playout
04.2 · Government archives and heritage
After digitisation, the finding aid.
A funded digitisation programme turns carriers into files. It does not turn files into finding aids. An archivist knows that a preservation master is not a working copy, that a checksum is a promise, that a collection-level description is honest but useless to a researcher who needs the speech, and that machine-generated metadata must never masquerade as archival description.
So the index is built from access copies, carries the identifier and checksum of the preservation object it points to, and labels every machine-derived field as machine-derived. Researchers search by what was said and shown; the catalogue remains the catalogue; the repository remains the repository. Retention and disposition run on your schedule — the system never disposes of anything on its own.
- Reads from
- The catalogue and finding aids, the preservation repository, digitisation output
- Returns
- Labelled discovery fields, timecoded, linked by identifier
- Never becomes
- The preservation master or the authoritative description
04.3 · Courts and digital evidence
Find the moment. Prove it is the same file.
Evidence is scattered across body-worn cameras, closed-circuit television, interview rooms, submitted phones — and, for older cases, tape and disc that no central search has ever reached. The test is not whether a reviewer can find the moment; it is whether finding it leaves the evidential original untouched, the chain of custody intact and the disclosure defensible.
The original is never modified. The index is a derivative beside your Digital Evidence Management System (DEMS) and Records Management System (RMS), which keep the primary relationship — we are the legacy-archive, federated and sovereign search layer they do not have. Every access and export is logged; a production pack carries the digest before and after, so the recipient can verify the file for themselves. Nothing here decides, scores or adjudicates; it finds, and it proves. Identification of people is off unless a lawful, audited government mode is agreed in writing.
- Reads from
- DEMS and RMS identifiers, the evidence store, restored legacy media
- Produces
- Hash-verified production packs with access history
- Never becomes
- The evidential original, the adjudication, or a redaction tool
04.4 · Education and public media
The lecture, searchable to the slide.
Students remember what the lecturer said and what was on the slide, not the file name in the Video Content Management System (VCMS). Librarians inherit special collections whose descriptions were written for a card catalogue. Both want the same thing — jump to the moment — without replacing the lecture-capture pipeline, the Learning Management System (LMS) or the streaming service that already work.
We index the recordings the institution already owns, speech and slides together, and return timecoded results and captions that respect course permissions. Migrations stop being a discovery loss: index before, during and after the move. Availability windows and approval workflows follow your content-governance policy. There is no identity inference on students or staff.
- Reads from
- VCMS and LMS course structure, lecture-capture output, digital collections, SSO
- Returns
- Timecoded results, captions and transcripts in the course context
- Doesn't replace
- The capture system, the LMS, the streaming platform
04.5 · Streaming and sport
Every instance of a moment is a sentence, not a project.
A producer knows the moment — the save, the line, the sponsor's board in shot — and not the asset name, and the content supply chain answers questions about assets. Across a season of full matches, "find every time this happened" is a project unless something has already watched the footage. We have.
Moments come back semantically across the whole catalogue, with the rights window and licensing fields that already exist surfaced beside each one, so the second question — may we use it — is answered with the first. Logging conventions are yours; we supply time-coded descriptors to shrink the backlog. We feed the highlight; we do not cut it. Playout, transmission and distribution stay exactly as they are.
- Reads from
- The content supply chain, MAM proxies and storage, rights data
- Hands to
- Editors, as deep links into the NLE; exports logged
- Never touches
- Playout, transmission, the edit, or rights adjudication
04.6 · Banking and regulated enterprise
Books and records, answered by search.
A regulator asks for the communications, not for a description of the system that holds them: readily accessible, protected from alteration, promptly produced in a usable format, kept for the schedule and not a day less. Recording is solved. Retrieval across years of calls in several languages is what is answered by reconstruction and sampling.
We index recorded calls and meetings from the recorder you already run and the Write Once, Read Many (WORM) archive you already keep — recorder-agnostic, both left untouched. Compliance searches by what was said; a request becomes a batch selection and a production pack with a time-stamped trail and digests. Retention schedules and legal holds apply; personally identifiable information (PII) stays behind roles. It is post-recording search and information governance, not market-abuse detection and not live monitoring.
- Reads from
- The call and meeting recorder, the WORM archive, surveillance and quality-assurance (QA) case identifiers
- Produces
- Production packs with time-stamped audit trail and digests
- Never becomes
- The book of record, a trade-surveillance engine, or real-time interception