Methodology

UAP Intel applies a consistent, auditable process to every item. The goal is clarity: to show what the evidence is, where it came from, and how far it can be trusted.

Five-tier evidence classification

Each item is assigned an evidence tier based on the authority and verifiability of its source — from Tier 1 (official) records at the top, down through progressively lower tiers for secondary reporting, attributed claims, and uncorroborated material.

Retrieval

We combine dense (semantic) and keyword retrieval so relevant evidence surfaces whether it matches by meaning or by exact terms. Every retrieved item carries its provenance forward.

Human review before publication

Nothing is published automatically. Submissions pass through review and explicit admin approval before they appear in the public archive.

Versioned archive

The published archive is versioned. Answers and citations are stamped with the archive version they were drawn from, so results are reproducible.

Grounded AI

The assistant answers only from cited sources in the archive. When the available evidence is insufficient to answer responsibly, it refuses rather than speculating.