Tashi

  • Name: Tashi
  • URL: https://docs.tashi.network/whitepaper/whitepaper.md
  • Category: real-time coordination infrastructure / leaderless edge-consensus stack / DePIN coordination-and-settlement middleware
  • Summary: Tashi is worth cataloging not as just another DePIN, AI-agent network, or generic high-throughput consensus project, but as a coordination stack that explicitly splits local agreement, global infrastructure validation, and public-chain settlement into separate layers. The current whitepaper and docs make that decomposition unusually legible: Vertex provides leaderless DAG-based consensus inside application-specific meshnets, Lattice runs the global node market and reward-validation control plane through Orchestrators and Resource Nodes, and Arc bridges validated coordination outcomes to public chains only when payment, token issuance, or public finality is needed. That makes Tashi a useful comparison point for Streamr, Waku, replicated-service middleware, shared-sequencing systems, and crypto×AI coordination layers because the real control surfaces are not just latency claims. They are meshnet membership, Orchestrator power over validation and job routing, Reward Point issuance, staking-and-reputation policy, proprietary-versus-open consensus control, and when private coordination is translated into public-chain settlement.
  • What it does:
    • Provides application-specific peer-to-peer meshnets where robots, AI agents, games, or IoT systems reach fast local consensus through Vertex without publishing every event to a public chain
    • Uses a leaderless DAG plus virtual-voting design in Vertex to produce sub-100ms Byzantine-fault-tolerant ordering and a signed Proof of Coordination for each completed coordination session
    • Runs a global Lattice layer where Orchestrators handle discovery, routing, validation, reputation, failover injection, and reward co-signing while Resource Nodes provide proxying, relay capacity, and application-specific execution services
    • Treats Proofs of Coordination as the main portable output of the system: Orchestrators validate them for rewards and Arc can then bridge them to external chains for settlement, signaling, or token distribution
    • Pays for infrastructure through a USD- or USDC-denominated interface, issues Reward Points backed by Treasury reserves, and requires Resource Nodes to stake $TASHI for participation and slashing exposure
    • Positions most coordination as private and offchain by default, with Arc used only when outcomes need public-chain finality or tokenized settlement
  • Key claims:
    • The most reusable analytical split is Tashi’s three-layer architecture. Vertex handles local consensus, Lattice handles infrastructure coordination and economic policy, and Arc handles public-chain settlement. That is much more informative than filing it as a single AI coordination network.
    • Vertex’s docs stress leaderless DAG ordering, virtual voting, fairness, and Proofs of Coordination rather than ordinary block production. That makes Tashi closer to offchain replicated-service and coordination middleware than to a conventional smart-contract chain.
    • Lattice is where practical control concentrates. Orchestrators accept jobs, choose Resource Nodes, audit sessions, validate proofs, maintain reputation, inject failover, and co-sign reward proposals. Resource Nodes do the work, but Orchestrators own the higher-order routing and reward-validation layer.
    • The proof model is deliberately compact and session-scoped: supermajority signatures over agreed data hashes, optionally with more structure attached. That makes Proof of Coordination a reusable output object rather than a full public ledger.
    • The token and payment design is also revealing. Apps do not need 1 and convert those to $TASHI. That pushes token demand toward operator extraction and staking rather than ordinary app usage.
    • Governance is materially centralized in the current materials. The Foundation controls protocol development and key parameters during the early phase, and Vertex is still proprietary though promised for later open-sourcing. That is an important maturity and governance caveat for any decentralized coordination reading.
    • Tashi clears the corpus bar because it makes a distinct control stack legible beneath the usual DePIN for AI/robots pitch: leaderless local consensus, global orchestration and reputation routing, proof-based reward validation, and optional public settlement only at the boundary.
  • Whitepaper: Tashi has a current official executive whitepaper and supporting docs rather than a separate PDF-heavy source packet. The main primary-source notes for this pass are in ../whitepapers/tashi-primary-sources-2026-05-15.md.
  • Sources:
  • Last reviewed: 2026-05-15 UTC