Reppo

Disambiguation finding (the main one): the tracker descriptor "AI reputation system" is misleading. Reppo is NOT an agent-identity or reputation system (ERC-8004 / World ID / KYA). It's a decentralized prediction-market protocol for AI training-data quality — closer to RLHF-as-a-protocol than to identity infrastructure. The "reputation" framing traces to its 2024 pitch about risk/reputation models for DePIN networks; the 2026 execution has pivoted to Datanets and data-quality prediction markets.

Product

Domain-specific subnetworks called Datanets (medical imaging, legal text, etc.). Inside each:

• Miners produce data
• Voters stake REPPO to validate quality
• Markets settle every 48h
• Output = scored time-stamped data feeding into an RL environment

Vetoken model (veREPPO) governs emissions. Built on Base L2, with Filecoin/IPFS storage layer (Protocol Labs heritage).

Team

• Raghav Rmadya (CEO/founder)
• Caroline Cai (co-founder)
• Founded 2024, spun out of Protocol Labs Venture Studio (March 2024)
• AI/ML credentials thin in public coverage — pitch is mechanism design over AI research

AI Substance

Honest read: no proprietary ML. The "AI" is the use case (consumers of the curated data are AI labs). Protocol itself is crypto-mechanism design treating human preferences as the oracle. Real value depends on whether labs actually buy curated Datanet outputs.

$REPPO Token

• Price: ~$0.034 | MCap: ~$12M | FDV: ~$34M (3x dilution overhang)
• Supply: 350M circulating / 1B max | Contract: 0xff8104251e7761163fac3211ef5583fb3f8583d6 (Base)
• Listed: Kraken (primary), Coinbase, MEXC, XT. Not Binance
• Utility: governance (veREPPO), subnet emission voting, voter rewards, solver fees, Datanet seed lockups
• "Top Gainers" placement = April 23, 2026 $20M strategic commitment from Bolts Capital + Kraken listing + Crypto Rover spotlight → 40-75% surges

Recent News

• April 23, 2026: $20M commitment from Bolts Capital (post-seed $2.2M from Protocol Labs, CMS, CV VC + Anoma/Ledger angels)
• Kraken listing post-funding
• Target: 100+ Datanets by end of June 2026

Traction

Light. No public named AI lab customers consuming Datanet outputs at scale. Vana partnership (May 2025) for risk/reputation models in Vana's data pools is the strongest external validation but appears bilateral. Currently Phase 1 — public subnets and early Data/Infra exchange "live" per project.

Critiques

• Sybil/coordination risk: stake-weighting only partially mitigates; whales can skew preference signals; subjective data quality has no ground-truth oracle
• Token overhang: 35% circulating, ~13-15M/month unlocks
• Pitch shift between 2024 (data infra for AI agents) and 2026 (prediction markets for data) suggests product still searching for fit
• Early community labeled it "potential scam" pre-Bolts; thin technical docs
• Narrative-stuffed: combines prediction markets + DePIN + AI data + RL envs — could be real or veneer

Competitive Positioning

• NOT in the same lane as ERC-8004 / World ID / agent identity. Wrong category for the tracker description.
• Adjacent to Vana (DataDAOs, user-owned models — partner not competitor) and Grass (raw browser data — different layer)
• Reppo's wedge: quality-curation layer on top of raw data sources

Watch Signal

The Vana integration is the real customer-substance signal; if Vana DataDAOs ship with Reppo quality scoring as default, that's traction worth tracking. Until then, Reppo is a Protocol-Labs-incubated mechanism-design experiment with a Bolts-Capital backstop.