About ICME Labs
ICME Labs is building zero-knowledge proof technologies for privacy preserving verifiable ML. Jolt Atlas is the state of the art zero knowledge machine learning (zkML) repo. We have made verifiable agentic guardrails into a problem solvable with maths. Don’t trust Blackbox AI security, verify.
We use 'folding scheme' to build very fast and space efficient ZKP, making it the ideal solution for constrained devices like browsers, DePin, and consumer hardware. On top of this we are building novanet.xyz a non-uniform incremental verifiable compute (NIVC) based prover network that allows for ultra fast proving via parallelization, a mathematically based DeSci layer for specialized provers (machine learning/zkFHE), privacy, and aggregation. It is peer 2 peer. It works for verifiable artificial intelligence, scaling blockchains, DePin, and privacy centric use-case.
On the consumer side we incubated kinic.io an AI memory store that works as a 'Plaid for AI memory'. Users have full control of their AI memory and can use ZKP to prove attributes about it and verify additions, and updates. The zero knowledge machine work at Kinic — is the first specialized prover in NovaNet and uses ICME's zkML stack extensively.
Interested in learning more about our tech?
Reach out on Twitter. Or visit our main site at https://icme.io/