CELAYA SOLUTIONS RESEARCHRESEARCH / TRINIDAD

// Research Instrument

Trinidad

Trinidad is an archival intelligence system over 5,435 rare Texas historical documents, with 1,065,357 passages indexed, deployed at trinidad.town.

What it studies

Trinidad studies how historical records can become an evidence-aware retrieval and reasoning surface while keeping source material visible.

Artifact or evidence produced

It produces source-linked retrieval paths and archival intelligence outputs over the Texas historical corpus.

Research context

Trinidad is documented as part of Celaya Solutions Research Lab's work on local-first Applied Intelligence systems, provenance-aware research instruments, industrial Applied Intelligence workflows, archival intelligence systems, multi-agent orchestration tools, and human-judgment-preserving Applied Intelligence architecture.

// Problem

A corpus is useless if you cannot trust where an answer came from.

5,435 rare Texas historical documents, with 1,065,357 passages indexed, hold real answers, but a plain language model over them would invent confident claims with no way to check them. For archival work, an answer without provenance is worse than no answer.

// Method

Retrieval bound to source, answers bound to records.

Trinidad routes a query to the records that match, returns an answer, and keeps each claim linked to the record that supports it. The path from question to source stays visible. The same provenance-as-architecture approach is what the lab builds for clients in legal and industrial domains.

Open trinidad.town

// Outcome

A public, source-linked window into a historical corpus.

Trinidad runs in public over the full corpus, answering questions with the records behind them. A fuller writeup of what it surfaced that plain search could not is being prepared. Until then, the system speaks for itself at trinidad.town.

This is the kind of system the lab builds for clients. Hand the lab a problem.

CELAYA SOLUTIONS RESEARCH / RESEARCHRESEARCH / TRINIDAD