Roadmap & priorities

Archetype is built in the open. This page sets out where the platform is headed: what is partly built and being finished, what is planned next, and the research directions being explored. Everything here describes the software, so it holds for any instance.

For what already works, see What works today.

Funding priorities

In order of how directly funding turns into results:

1. Annotation statistics and quality control — planned, with a design proposal. The reporting half is cheap; the supervisor review step is separable.
2. Sustained maintenance — dependency and security updates, the public instance, and keeping these docs true to the code. Not a feature, but what everything else rests on.
3. Interoperability — importing W3C annotations from other tools, and IIIF collections across many manuscripts.
4. A second corpus — bringing another collection onto the platform, to prove it beyond the charters it was rebuilt around.
5. AI-assisted palaeography — a phased research programme that depends entirely on grant funding.

The full picture

In progress1

Partly shipped; more underway.

Interoperability & Standards

In progress

Standards-based export and import so the scholarship is portable and citable.

• Shipped: W3C Web Annotations (JSON-LD) and IIIF Presentation 3 export; TEI and HTR (PAGE-XML/ALTO) import.
• Next: a reverse W3C-annotation → graph import path.
• Next: a IIIF Collection endpoint spanning multiple manuscripts.

Learn more →

Planned1

Designed; not yet built.

Annotation statistics & quality control

Planned

Give editorial teams a view of coverage and consistency across a corpus.

• Coverage reporting per manuscript and per hand.
• Per-annotator statistics and activity.
• An optional, off-by-default supervisor review step and quality-control queue.

Exploratory1

A research direction, not a committed feature.

AI-assisted palaeography

Exploratory

A future, grant-funded research direction — with the scholar always the author of record.

• Today the platform imports externally-produced HTR transcriptions only; it runs no recognition model.
• Not part of the released platform: the programme depends on grant funding. A phased design exists, and exploratory work is kept on a separate branch.
• Research directions: handwriting recognition, scribe attribution, and computational dating — every machine suggestion routed through the editorial review workflow.

Reviewed against the codebase · 2026-10-01