The platform

Everything Archivers.ai does.

AI drafts the record, an archivist accepts every field, and the result exports into the systems you already run.

Start free Talk to us

30 free credits a month · No card needed

A catalogue record in Archivers.ai for a printed index page of Coroners’ Inquests 1732–1836, showing the drafted title and description, subject tags, the scanned source page and an Accept button, with the record still marked Unreviewed

Works with the systems you already run

What it does

Draft, review, find, export.

Four jobs, and one reviewed record underneath all of them.

A drafted Archivers.ai record with a personal-data notice, an open AI evidence panel explaining how the date and location were inferred, and evidence marks beside the title, date, creator, location and description fields
Evidence

Every field shows its working.

Each drafted field cites its evidence from the source, or says plainly that it has none, and any personal data the AI spotted is flagged.

  • Evidence cited field by field, inside the record
  • Rationale you can open and read
  • Personal data flagged before review
How we use AI
An Archivers.ai accession of ninety board-minute images shown as a sortable table of ninety items, none yet reviewed, with columns for content warnings, evidence citations, AI-drafted title, date and creator
Review

Review a backlog, not a record.

An accession opens as a working table, so a reviewer sees where the warnings are before opening anything.

  • Edit any field, then accept the record
  • Free volunteer and reviewer seats
  • Nothing exports without a human accept
See the workflow
The Explore research chat in Archivers.ai answering a question about a football club’s community role before 1900, drawing on an 1899 board-minutes accession and naming the gaps and silences in the record
Find and ask

Search it, then question it.

Full-text search runs across records and transcripts. Explore answers research questions and cites the items behind the answer.

  • Search metadata, descriptions and transcripts
  • Explore cites the items it used
  • Audio and video transcribed on paid plans
Public access and Explore
The Results and Export screen in Archivers.ai offering a PDF report, a BagIt package and a catalogue exports menu, above a summary showing five items processed and none failed
Export

Leave in an open standard.

Reviewed records export as EAD3, BagIt with PREMIS, Dublin Core, and CSVs shaped for the system downstream.

  • EAD3 finding aids and Dublin Core
  • AtoM, ArchivesSpace and Spectrum CSV
  • PDF report for committees and funders
All export formats
Who it’s for

Four routes through one platform.

The same review gate and the same exports, sized to how your service works.

Archivists

Multi-level ISAD(G) drafts, EAD3 out, a review gate on every field.

For archivists

Museums

Spectrum-mapped object records for Axiell, Modes, Adlib and Mimsy.

For museums

Community archives

A free plan, volunteer seats and a public portal, with no IT department.

For community archives

Directors

Published prices, EU-resident AI, and a pilot before anything is signed.

For directors

Common questions

What people ask first.

What file types does Archivers.ai support?
Documents as PDF, Word and scanned images; photographs as JPEG, PNG and TIFF including multi-page; audio as MP3 and WAV; and video.
How does the AI classification work?
The AI reads each file, runs OCR or transcription, then drafts structured fields. Each one cites its evidence, or is flagged as having none, and the record waits for a human accept. Transcripts carry per-word OCR confidence.
What metadata standards do you support?
Drafts are mapped to ISAD(G) and Spectrum. Exports include EAD3, Dublin Core, BagIt with PREMIS, and import CSVs for the common repository systems.
Can I export to my existing collections management system?
Yes. Archivers.ai exports CSV, EAD3 and Dublin Core — formats most collections management systems can import, including Axiell, Modes and Adlib.
Is my data secure?
AI processing runs in EU data centres. Customer collection data is not used to train foundation AI models; we follow vendor no-training commitments — see Trust.
What does it cost to run?
One credit catalogues one photograph or object, one document page, one minute of video, or five minutes of audio. Every price is published.

Draft it. Review it. Publish it.

Start free with 30 credits a month, or talk to us about an institutional pilot.

Start free Talk to us