Critical judgment at speed and scale.
Generation of research-level content is growing exponentially. Expert review capacity is not. The gap between what can be produced and what can be evaluated is the defining epistemic challenge of the AI era.
Most of what is published — in journals, in preprints, in AI-generated research, in corporate filings, in litigation support — has never been checked by anyone qualified to check it. Not because people are lazy, but because checking is harder than producing, and always has been.
Solonic exists to close that gap.
Consulting. Verification engagements for M&A due diligence, pharma claims, litigation support, patent review, and academic research. Custom scope, human expertise backed by AI instruments.
Self-serve reviews. Submit a paper or manuscript for multi-model AI review. Up to 12 model families (effective independence ~4 at ρ ≈ 0.2) review your work. From $49. See a sample →
Free instruments. The methods, the Atlas, the packages — all free, no account, no paywall. The instruments are public because the moat is the track record, not the tools.
Kevin Scharp is a Professor of Philosophy at the University of Illinois Urbana-Champaign — a university that is also home to one of the world’s top-ranked computer science programs. kevinscharp.com
His academic work focuses on defective concepts — ideas that are internally inconsistent and need to be replaced rather than refined. His book Replacing Truth (Oxford University Press, 2013) argues that truth itself is an inconsistent concept and constructs two replacement concepts (ascending truth and descending truth) that do the work truth was supposed to do without the paradoxes.
This might sound abstract, but it turns out to be directly practical: the same methods that diagnose broken philosophical concepts diagnose broken claims in any domain. When a concept is defective, every argument built on it inherits the defect. Identifying the defect is verification. Solonic industrializes that process.
In 2024, Kevin realized that AI is disrupting every folk concept simultaneously — not one at a time, as technology usually does, but all at once. Truth, knowledge, authorship, expertise, evidence, originality — all of these concepts are breaking under pressure from AI systems that produce outputs indistinguishable from human work.
The tools of conceptual engineering — identifying inconsistencies, finding maximal consistent subsets, constructing replacements — are exactly the tools needed for verification at scale. Solonic is the application of 20 years of philosophical methodology to the most pressing practical problem of the AI era: telling what's right from what sounds right.
There is no mark of truth. There is only evidence, and it points both ways. The Neurath gate — our verification framework — never certifies. It tags, tracks, and corrects.
Every claim carries its status. PROVED, VERIFIED, OPEN, CORRECTED, or REFUTED. Untagged output is a bug, not a result.
Our moat is not secrecy. It is a public track record of being right — and of saying so when we were not. See the ledger and the errata.
Methods are free. Judgment is the product. The methods are published because verification that requires secrecy is not verification.