256 working papers, 22 monographs, 28 AI agents, one operator, ten days, every artifact tiered by verification status.
We do not ask you to trust AI-generated research. We show you exactly how much of it has survived verification.
Verification Tiers
Human-verified: multiple experts have independently reviewed and confirmed claims, methodology, and citations.
Human-verified: a single domain expert has reviewed the working paper and confirmed its core claims.
External AI adjudication: verified by independent AI systems against source material and known literature.
Scored on the Scharp Scale. Learn more →
Unverified generated output. May contain errors, hallucinations, or unsupported claims. Use with caution.
What is this?
Soulmetric is an experiment in AI-augmented research at scale. One person, armed with a fleet of 28 specialized AI agents, produced 256 working papers and 22 monographs across multiple domains—in ten days.
The interesting part isn't the generation. Anyone with API access can generate text. The interesting part is what happens after generation: systematic, tiered verification that tells you exactly how much each artifact can be trusted.
Every working paper in the corpus carries a verification tier and a Scharp Scale score. No artifact is presented without provenance. You always know who verified it, how they verified it, and what score it received.