To align AI with human values at Internet scale it will take a new kind of collaboration between humans, machines, and infrastructure. Today, we’re announcing a partnership with Reppo.ai to scale Verifiable AI. Reppo is building the on-chain system for generating high-quality AI alignment data through a dynamic interplay of creators and voters. That system becomes verifiable, scalable, and globally distributed with the Nexus blockchain and zkVM. Together, we’re enabling a future where anyone can help shape better AI, and get rewarded for doing it.
How Nexus Powers Reppo’s Alignment Engine
With Nexus, every step in Reppo’s workflow will become verifiable. When a voter ranks a model completion or a creator uploads a new dataset, those actions generate zero-knowledge proofs on the Nexus prover network. This brings provable trust to the entire feedback pipeline. Every data point, ranking, or evaluation is logged and auditable. Dispute resolution becomes trustless. Reward flows become transparent. Ownership over improvement contributions becomes enforceable. We move beyond black-box fine-tuning into a realm of verifiable participation.
Nexus will also power the execution of Reppo’s collaborative workflows. Whether it’s routing prompts, comparing multiple outputs, or resolving disagreements over votes, these processes can run verifiably both on-chain and off-chain using the Nexus zkVM. This hybrid execution model enables global scalability while maintaining the cryptographic integrity of every action. Voters and solvers engaging with RLHF tasks can run workloads on Nexus nodes and receive verifiable rewards. In this future, your feedback doesn’t just help models improve—it becomes a cryptographically staked asset.
Integrating with Reppo.exchange
Beyond training, we’re collaborating on Reppo.exchange, the execution layer for agentic systems built on the Model Context Protocol (MCP). Reppo.ai focuses on generating alignment data; Reppo.exchange activates that intelligence by letting agents post Requests for Delivery (RFDs) and solvers fulfill them. Nexus will add a verifiable compute backbone to these interactions. Agent workflows gain audit logs, reward transparency, and zero-knowledge verification.
What This Means for the Nexus Community
This partnership opens new doors for the 2.6 million Nexus users around the world. They can earn rewards for contributing to AI development, participating in evaluation and governance, and helping shape the next generation of intelligent systems. Developers will be able to deploy agents and solvers through Reppo.exchange, supported by Nexus’ scalable infrastructure. And through joint incentive programs, Nexus users will find more opportunities to earn rewards by contributing compute and feedback to Reppo-driven tasks.
We're excited to partner with Reppo in building the Verifiable Internet, where human creativity meets cryptographic trust.
