A modern platform for libraries, universities, and government archives. Upload anything, let AI draft the description, and publish to international standards — without leaving your workflow.
Not a wiki, not a spreadsheet. Every item is a live document with AI-drafted fields on one side, the source on the other, and a persistent track-changes bar so nothing ships without curator sign-off.
Ask
items I edited this week
A floating prompt bubble lets curators query the catalog in their own words. Items I edited this week. Fonds from the 1920s. Drafts ready for review. Each result jumps straight to the matching record. Recent prompts are remembered so routine queries are one click away.
Most archives run on a stack: a DAM, an OCR vendor, cataloging software, a transcription service, an image editor, and whatever publishes the finding aid. Archively.AI replaces the lot — one platform, one standards-compliant data model, one place curators work.
Images, PDFs, audio, video, spreadsheets, Office — each gets the right preview and the right pipeline. No second tool to license, no separate ingest path.
Native text extraction for born-digital PDFs, AI OCR for scanned pages — both in the same workflow, with every word track-changed before it ships to the catalog.
ISAD(G), Dublin Core, MARC21, PREMIS — mapped to the data model, not bolted on as an export. Eight catalog modules — Items, Fonds, People, Organizations, Families, Accessions, Subjects, Events — every record search-faceted out of the box.
Word-level AI transcription with per-word confidence. A proofreader view to accept, edit, or reject — sentence by sentence — without ever leaving the file.
Archival crop presets, deskew, AI object detection, and archival captions — without leaving the file detail. Every edit lands as a child file with its own provenance.
Every tenant gets a public portal at its own subdomain, with six visual templates and four feature flags. EAD2/EAD3 + MODS + EAC-CPF + BagIt exports, OAI-PMH for harvesters, IIIF manifests for external viewers — automatic.
Search any entity, let AI fill in the fields, review every change with full diffs, and publish a citable, immutable record — in one continuous workflow.
Upload, AI describe, review with diffs, publish. Four stages, one audit trail. This demo cycles the states; you can tap a stage to jump.
Three layers — AI output, curator review, and the immutable published snapshot — sit side by side. Inline diffs show exactly what changed; one click reverts or publishes.
An immutable snapshot of what the model produced.
Your working copy, with pristine/dirty tracking.
Signed, dated, and locked. Citable for decades.
All tenant-scoped, track-changes enabled, and server-searchable.
ISAD(G)-shaped records.
Hierarchical arrangement.
Creator authorities.
Corporate bodies.
Kinship-shaped creators.
Acquisition tracking.
Topical access.
Contextual timelines.
More modules on the roadmap: exhibitions, requests, and reading-room workflow.
See all features →The data model is the same. The workflows and portals adapt to how your institution actually works.
Libraries, universities, and schools — special collections, oral history centers, and yearbook runs back to founding.
Public, academic, and specialist libraries catalog special collections, publish finding aids, and sync with their existing ILS — without forcing manuscripts into bibliographic records.
ExploreInstitutional archives, oral history centers, and departmental research collections — all on one deployment, each with its own tenancy, branding, and SSO-backed access.
ExploreIndependent schools, boarding schools, and K-12 academies use Archively.AI to bring yearbooks, alumni records, head-of-school correspondence, athletics history, and FERPA-bound student records into one searchable catalog.
ExploreFour flagship demo tenants — one per anchor category — running on the same deployment as everyone else. Ten more live alongside.
1891 – 1935
Manuscript archive of a fictional provincial medical society, including 1918 pandemic correspondence.
See it →1975 – 2015
Audio and video interviews with alumni, faculty, and townspeople — one under embargo.
See it →1918 – 1925
Public-health records with chain-of-custody, redaction, and PREMIS preservation metadata.
See it →1961 – present
Exhibition records, donor histories, and curatorial research files with events as first-class records.
See it →Fragmented tooling, AI you can't fully trust, metadata locked in a proprietary format, and a catalog no one outside the building can reach. Here's how Archively.AI is built to answer each one.
Upload, OCR, transcription, cataloging, image work, and a public portal live in one place — one data model, one login, one audit trail from intake to publication.
Every AI-generated field carries a three-tier trail — AI draft, curator review, published snapshot — with word-level diffs. Nothing becomes public without a human accepting it.
Records map to ISAD(G), Dublin Core, MARC21, and PREMIS in the data model. Export EAD 2002/EAD3, METS, MODS, EAC-CPF, and BagIt; expose OAI-PMH and IIIF; mint ARK and DOI identifiers.
Each institution gets a branded discovery portal with full-text and semantic search, deep-zoom image viewing, and media transcripts — published to the standards external harvesters and viewers already speak.
Spin up a tenant, import a small collection, and see a standards-compliant, AI-described finding aid in an afternoon.
