Skip to content
OxARCA

Archos

A source-grounded LLM system that turns the epistemic requirements of historical scholarship into architecture.

Archos is an epistemically grounded LLM system for archival exploration. It is built on one commitment: an AI system is only usable in the humanities if the requirements of scholarly practice are encoded into the architecture itself, rather than requested of the model in a prompt and hoped for in the output.

What makes it different

With Archos, historians remain in control of interpretation throughout. The system assists with evidence discovery, organisation and documentation; it does not adjudicate the history.

Our workflow

The Archos pipeline. Steps, left to right: a historian writes the research statement; the system retrieves archival evidence, applies an epistemic constitution and produces source-grounded synthesis; a historian verifies the result. The first and last steps are marked as domain expert involvement, the three middle steps as scholarly AI.

Tap the diagram to open it full size. Figure 1. The division of labour across a run. The interpretive decisions at either end stay with the historian; the scholarly AI carries the evidentiary work in the middle, and every claim it produces remains traceable back to a retrieved document.

Where the historian sits in the loop

Archos is not an end-to-end answering machine. A domain expert writes the research statement that opens a run, and a domain expert verifies the report that closes it. Between those two points the system works under constraint: it retrieves archival evidence, holds it to an epistemic constitution, and synthesises only from what retrieval actually returned.

Reading an Evidence Report

An Evidence Report, side by side. Left: the transcribed archival source, with one passage highlighted and marked 1. Right: the synthesis for subsection 3.1, Chaloner's Printed Reformer Claims, where each sentence carries its own numbered footnote, marked 2, and a provenance card is open over the text showing the source descriptor, its id MINT01535 and the verbatim quotation, marked 3.

Tap the screenshot to open it full size. Figure 2. The Evidence Report for one subsection of the Chaloner Prosecution run (Royal Mint papers, 1692–1700). Left: transcribed source with a highlighted citable passage. Right: source-grounded synthesis in PEEL structure, where every claim is traceable to a specific retrieved document.

The system's output is an Evidence Report: the transcribed source on one side, the source-grounded synthesis on the other.

  1. A citable passage is highlighted in the transcribed archival source.
  2. Sentence-level citation — every evidence sentence carries its own footnote or footnotes.
  3. A provenance card appears on hover, giving the source id, date, verbatim quote and epistemic status.

Try it on your own material

We are looking for collections to test against. See get involved for what we need and what you get back.

Last updated .