Jev-as-a-judge🤝 RL Environment Verifiers 🤝 way cheaper + faster RL a lot of things in Agent World look like classification problems…including grading RL - many RL Tasks need a judge to verify pieces of outputs or trajectories - at scale this can be very expensive and slow with tons of rollouts to score - Jev is incredibly fast and cheap meaning, can potentially remove the verification bottleneck for many tasks we previously worked with @harvey on their LAB benchmark which used LLM-as-Judge for tens of criteria per task we found that Harness/Prompt Engineering + Open Models meant we could judge tasks orders of magnitude cheaper like any verifier, it needs to be calibrated to make you’re giving back a good signal but Jev has the opportunity to massively bring down verification costs (and tuning friction) for many tasks which means more teams will do RL, which is great!

$RELY tokenomics are public. no issuance path. no emissions for sitting. no pay-for-score. rewards follow the work that fills the datanet. anyone can fork a contract. nobody can fork the receipts. full doc available on relyxyz.com/

$REPPO dropped a ridiculous amount of new information today and I don’t think CT has processed what this is becoming. 1. Eval API is ready. Short eval: $0.05–$0.50 Normal agent/document eval: $1–$5 Deep research: $12–$25 Instead of hiring Mercor/Scale experts by the hour and going through enterprise sales, Reppo wants: send output → get independent judgment → pay per call 2. The Eval API has a router. Think OpenRouter, but for judgment. A request comes in and Reppo automatically routes it to the relevant specialized evaluation market. So Reppo isn’t building one giant generic judge. It’s building a network of specialized signal factories behind one API. 3. Real Web2 customers are appearing. Luca Theory — which provides sports/betting data to large funds, is working toward using Reppo as its QA/QC layer. VORTX AI — satellite imagery is also in the works. 4. Orquestra Lite is basically one-click participation. No Docker. No RPC. No API keys. Reppo handles the infrastructure. They want to go from ~20–25 highly active nodes to 100+, then 1,000+. More Eval API demand → more evaluation markets needed → more agents/nodes needed. 5. A completely new $REPPO value-accrual mechanism is coming. New Datanet reward tokens can be paired directly with $REPPO. Proposed fee split: 60% → builder 40% → automatic $REPPO buyback + burn RG said the buyback component will likely be mandatory. And they estimate this alone could contribute 10–20%+ of Q3/Q4 buyback targets. On top of the existing enterprise-revenue buybacks. 6. They explicitly rejected endless inflation as the growth model. Most crypto networks subsidize activity by printing tokens until nobody wants them anymore. Reppo wants activity funded by: customers + usage + fees + external demand. 7. Institutional interest is starting. RG specifically mentioned conversations around Singapore with institutions including OKX and Temple Digital. And the feedback they’re getting is important: Institutions don’t just want one app. They want to see an ecosystem being built on top of the protocol. Which is exactly why Reppo is expanding into: Eval API Orquestra nodes Datanet builders reward-token ecosystems robotics data QA agent evaluation inference-time RL A ~$6M protocol is attempting to turn human + machine judgment into an API, already talking to Web2 customers, while building automatic revenue → $REPPO buyback rails underneath the entire ecosystem. Wild.