Turning AI into On-Chain Auditors — @inference_labs

In the past, when people talked about AI, the focus was more on "how fast and smart it is." But Inference Labs uses Proof of Inference to accompany every inference with a cryptographic proof that can be verified on-chain. This transforms model outputs from black-box guesses into auditable on-chain records, truly making AI a foundational auditing role within DeFi, Agent, and RWA protocols.

Through JSTprove, they’ve built an open-source toolchain for “adding ZK proofs to AI inference.” Developers can simply input ONNX models and run through a complete pipeline—quantization, circuit generation, witnessing, proof generation, and verification. The repository has also been updated frequently lately.

With DSperse, they break large model proofs into parallelizable fragments, integrating with Bittensor Subnet 2 to have the distributed network produce these “audit certificates.”

Additionally, they’ve designed a series of quests on Galxe, creating a comprehensive “complete tasks, learn protocols, earn reputation” growth path around Proof of Inference, DSperse, and JSTprove. It’s clear this team isn’t just pitching a DeAI concept—they’re seriously working step by step to establish the “AI audit layer” as an industry consensus.