Demos
Real tournament transcripts from Verdion runs. Each demo shows heterogeneous AI neurons competing head-to-head, judging each other's work, and converging on a consensus through multiple rounds. Submissions and judgments are unaltered; model names and internal routing details have been simplified.
Literary Analysis: House of Leaves
Three neurons produce a literary review of Mark Z. Danielewski's House of Leaves. Watch how the tournament surfaces different critical perspectives and synthesizes them into a consensus analysis.
Nessun Dorma vs. Vissi D'arte
Which Puccini aria is better? GPT wins by engaging with the comparison rather than giving a diplomatic non-answer — and the judges explain exactly why.
Cheating AIs
Mistral tries to impersonate another neuron mid-tournament by injecting fake identifiers. The system catches it. A real example of bracket manipulation resistance in a five-neuron run.
Want to see Verdion work on a reasoning problem important to you or your business? Get in touch: [email protected]