Roadmap
From registry to verification network
Status labels are part of the methodology here too: live means shipped and verifiable today; planned means committed engineering; research means under active design with open questions; draft means published for critique. Nothing on this page is a promise of token value, returns, or timing.
Launch registry
5/6 liveEstablish the public record: claims labeled against evidence, scores computed in the open, hashes that cannot quietly change.
- First audit — Base AI Claims Audit: 9 projects, 41 labeled claimslive
- Project dossiers with evidence records and Claim Reliability Indexlive
- Free API + MCP server (agents consume the same evidence as humans)live
- x402 premium route — 402 challenges live; USDC settlement on production facilitator configlive
- Correction process with public changelog and 2-business-day acknowledgmentlive
- Report anchoring in the append-only BasisReportRegistry (Base Sepolia first)planned
Coverage expansion
0/5 liveWiden the lens: more projects, more onchain ground truth, a re-check cadence that keeps every label current.
- Additional Base AI projects beyond the seed nineplanned
- x402 seller-address coverage — sybil-adjusted seller metrics from the indexer pipelineplanned
- x402 settlement classification (EIP-3009 selector analysis) in productionplanned
- Published quarterly re-check cadence with lastChecked sweepsplanned
- Source-monitoring pipeline: claim-relevant events trigger re-checksresearch
Marketplace integrations
0/5 livePut verification where decisions happen: inside agent loops, wallets, and marketplaces — before the payment, not after the loss.
- Agent-marketplace trust layer: claim checks surfaced at listing/purchase timeresearch
- Wallet and agent pre-checks — spend-time label lookups via MCP/APIresearch
- MCP client integrations beyond Claude (Codex, Gemini, custom stacks)planned
- API licensing for exchanges, funds, and diligence platformsplanned
- Structured diligence workflows on the registry dataplanned
Tokenized verification market
0/10 liveResearch phase. A possible $BASIS token that lets anyone request, prioritize, fund, or challenge verification work — under hard constraints that keep outcomes unbuyable. Nothing in this phase is launched, promised, or an offer.
- Virtuals Stage 1 — Basis agent profile (read-only verification lookups; operator runbook + profile copy ready)planned
- Virtuals Stage 2 — ACP service listings (USDC, x402-compatible, corrections always free)planned
- Venue conflict regime — Basis audits Virtuals; disclosure + instrumentation, never recusalplanned
- Public audit-request mechanism (burn-to-request, no outcome guarantee)research
- Public work queue with capped, logarithmic priority weighting + guaranteed free editorial trackresearch
- Challenge bonds routing into the existing corrections process (bad-faith-only slashing)research
- Source/coverage curation staking (signal only)research
- Reviewer reputation — non-transferable, separate from any tokenresearch
- Anti-capture safeguards: parameter-only governance, never findingsresearch
- Draft economic model published for public critiquedraft
Protocolization
0/5 liveLong-horizon: the registry outgrows its operator. Standards, attestations, and a reviewer network make the evidence layer durable.
- Decentralized reviewer network with staked, adjudicated workresearch
- Onchain attestations for claims and scores (EAS on Base)research
- Reproducible evidence bundles — third parties re-derive every labelresearch
- Governance over parameters, never over findingsresearch
- Safe/multisig publisher model for the report registryplanned
Phase 3 is planning, not a launch: $BASIS is a planned work-request token — no contract exists, no date is set, and no mechanism described here can buy, alter, or suppress a label. The position change from the original no-token stance is public and versioned (RFC-001) — see the draft economic model and the whitepaper.