How to read this page.
Demonstration workflows describe functions shown in scoped development workflows. Availability and limits are confirmed for each configured build; this is not a claim of universal acceptance on arbitrary collections.
Pilot completion goals are included in the agreed scope but do not have a complete accepted workflow established here. They are not advertised as delivered features.
Visual discovery — demonstration workflows.
Browse an indexed collection; search images by description; find visually related images; review exact and near-duplicate candidates; return to an original file. Collection summaries can show file types and quality or coverage information available in the index.
Similarity is a research aid. It does not verify a person, object, maker or provenance.
Document review — demonstration workflows.
Read supported documents, search extracted text, locate source quotations and inspect document/page evidence. Native text and OCR can support retrieval depending on the file and configured workflow. AI explanations remain reviewable drafts rather than authoritative interpretations.
Audio review — demonstration workflows.
Work with locally produced transcripts, inspect likely speaker changes, search words or topics and replay source passages by timestamp. Speaker labels are anonymous clusters until a person confirms an identity. Answers and speaker attribution remain subject to source review and build-specific acceptance.
Video review — demonstration workflows.
Navigate supported recordings through transcript passages, sampled frames and original playback. Review what is said and what is visible without confusing nearby context with exact speech at a selected moment. Clip limits and coverage are build-specific.
Connected discovery and Ask — integration work.
The Pilot goal is a coherent path from a mixed collection to useful search, reviewable answers and original-source navigation. Individual source-linked workflows have been demonstrated; complete current-build cross-media acceptance remains a release gate. Ask for the specific workflow you need to see.
Collection intelligence — Pilot completion goals.
Persistent Image Families would connect related versions for review. Purpose-Specific Best Copy would help explain which copy suits OCR, preservation review or display. Existing similarity and duplicate results are foundations, not proof of every planned lineage or scoring function.
Other completion goals include portable Evidence Receipts / Evidence Packs, a Collection Assessment Report, consistent human-review states, a System & AI Fact Sheet, Time-to-Evidence measurements and a reusable Gold Question test set. Completed export and acceptance results must be checked before these are offered as delivered capabilities.
Audiobook and basic video production — bounded Studio goals.
The agreed personal Pilot 1 scope also includes local narration, chapter output and listening review; simple trim/join/titles; and a shared builder for approved audio plus one image or a small image set, with an intro/title, simple fades and MP4 output.
These are separate creation workflows, not features to infer from transcript analysis or video playback. Their full accepted implementation is not established here. No custom cloned-voice training, professional editor replacement or completion date is promised.
Deployment and permissions are separate decisions.
Initial sessions are developer-operated on a configured VAIMS workstation. Supported formats, processing limits, sample permissions and enabled capabilities are agreed first. A demonstration is not a downloadable installer, a certification or approval to deploy the entire development environment elsewhere.