Skip to content
FrankX OS · Experiment Intelligence System

Experiments

Pre-registered, hypothesis-driven research on agentic systems, personal AI, and content engineering. Every claim is falsifiable. Every experiment is reproducible. Negative results ship as readily as positive ones.

Experiments
1
Supported
2
Not supported
1
Deferred
1

The 5-doc discipline

Borrowed from how Anthropic, DeepMind, and OpenAI publish research artifacts. Every experiment ships these five documents, authored in this exact order, before any data is interpreted:

  1. 1hypothesis.md — pre-registered claims before data
  2. 2methodology.md — locked protocol + metrics
  3. 3results.md — quantitative + qualitative findings
  4. 4recommendations.md — ranked action queue
  5. 5replicate.md — reproducibility commands so anyone can re-run

Shipped experiments