No Closed Form
Projects
experiment since 2026-04

Research Intelligence Engine

Discovering evidence across sources, comparing patterns, and producing traceable research synthesis.

Most "research agents" summarize. I wanted one that argues — that forms a hypothesis, tests it against sources, and shows its work.

A multi-source system for finding evidence, comparing patterns across it, and producing a synthesis that reads as an argument rather than a list — with the sources it rests on kept traceable.

The distinction it’s chasing

The difference between listing and arguing: a listing agent retrieves and paraphrases; an arguing one states a hypothesis, marks what would confirm or refute it, and revises when the evidence pushes back. That extra structure — hypothesis, test, revision — is what separates research from summary.

Current state

Experiment. Earlier than the others here; I’m mostly probing how far the “state a hypothesis with a falsification condition” pattern carries before it needs heavier machinery.