The network coordinates the resources.
Gensyn Lab
Machines that predict the future need to be trained against reality, which means building the infrastructure that supplies it: proof that outputs are real, resolution for events in the world, and a market that surfaces information the model hasn't seen. Gensyn is where that machinery gets built and released.
open-1b
The first model you don't have to trust.
Every training step fingerprinted, published and replayable. Auditable training is the missing layer of transparency in AI – it starts here.
Verifiable Oracles
Whoever captures the oracle captures the outcome. A credibly neutral AI report layer makes that capture prohibitively expensive – resolution for events in the world, verified rather than trusted.
Delphi
Thousands of markets, each representing a specific question, Incentive-aligned belief and outcome, continuously aggregated patterns of queryable information.
Protocol Infrastructure
Cryptographic verification
Cryptographic evidence that computations ran as specified, for training, inference, and settlement.
Persistent on-chain identity
On-chain identity for every actor (human, model, or agent), so reputation and stake accrue.
Peer-to-peer communication
Low-level networking for ML nodes to exchange weights, gradients, and signals peer-to-peer. AXL is open source and available now.