For repository and preservation teams

Clean feeds into the repository you already run.

The hard part was never the repository. It is the description work that has to happen before anything can be ingested.

Start free Talk to us

30 free credits a month · No card needed

What you get

Not a repository. Never asking to be one.

Archivers.ai does not ask you to migrate off AtoM, ArchivesSpace, Preservica or Archivematica.

A completed Archivers.ai accession listing five reviewed items above an export panel offering a PDF report, BagIt, and catalogue exports for EAD3 XML, AtoM CSV, ArchivesSpace CSV and Archivematica CSV
Ingest-ready output

The metadata your workflow expects.

EAD3 finding aids, BagIt packages with checksum manifests, and PREMIS events recording processing and review.

  • BagIt to RFC 8493 with SHA-256 manifests
  • PREMIS events for processing and review
  • Transfer CSVs for Archivematica ingest
All export formats
An Archivers.ai record awaiting review, showing a personal-data flag, an expandable evidence panel for how the date and location were inferred, and evidence marks beside each drafted descriptive field
Provenance

The ingest record tells the whole story.

Each AI-assisted field carries the model family, the timestamp and the review state — and nothing exports until a reviewer has signed off. Machine and reviewer are both recorded.

  • Model family, timestamp and review state
  • Original suggestion kept beside the override
  • Sensitivity and personal data flagged early
Trust and procurement
The Archivers.ai reference code panel with Manual, CSV upload and Auto-sequence options, a base reference applied across a batch of files, and each file showing its own editable reference code
Export profiles

Agreed in writing, tested first.

Round-trip tests run against a staging instance of your target system during onboarding, before the full run.

  • Identifiers preserved exactly as supplied
  • Round-trip validation on a staging instance
  • Profile documented for audit and handover
Systems we work with
What stands behind it

Four things we state plainly.

Not a repository and not a preservation system

Open formats only, no proprietary blob

Provenance on AI-assisted fields

Exit in the formats you would have used anyway

Common questions

What preservation leads ask first.

What exactly lands in our ingest workflow?

EAD3 finding aids, BagIt packages with SHA-256 manifests and bag-info provenance, Dublin Core, and CSV transfer profiles for AtoM, ArchivesSpace and Archivematica.

How is AI processing recorded?

As PREMIS events. Each AI-assisted field also carries model family, timestamp and review state, so the machine step and the human step are both auditable.

Do you validate against our instance?

Yes. During onboarding we test imports into a staging instance of your target system and validate the round trip before the full run.

What happens if we leave?

You export in the open formats you would have used anyway. There is no proprietary container and no format that only we can read.

Where does processing happen?

Application compute and the database both run in London (AWS eu-west-2), with AI inference in EU data centres. The Trust page sets out sub-processors and regions in full.

Plan the cataloguing phase before ingest.

Start free on a sample bundle, or tell us the shape of the work and we will scope it together.

Start free Talk to us