Find unknown sources in a historical archive

Goal: Use agents to surface overlooked primary sources for a research question.
Setup: agents used in source: Claude; tools/MCPs stated: GLOBALISE archive, embedding model.
Steps:
1. Start from your own specialist knowledge to define a specific research question.
2. Pick a large, free, well-edited corpus (here the GLOBALISE Dutch East India Company archive).
3. Download it and run it through an embedding model for semantic search.
4. Surface candidate passages; have frontier models read them and rank the best matches for human review.
5. Iterate on promising passages, or try new archives and terms. Give the agent an API key so it can run its own embedding searches on new sources.
Rules: ask me before anything irreversible (sending, posting, paying, deleting); don’t invent data, results or tools I haven’t given you; tell me what you couldn’t verify.
Based on: Benjamin Breen — https://resobscura.substack.com/p/using-opus-55-to-discover-a-new-eyewitness

Steps are from the source. Includes a “Based on” credit line.