Introducing Tenzro Protocol

The agent economy is no longer theoretical. In 2026, AI agents are conducting $9.14 billion in commerce. McKinsey projects this will reach $3-5 trillion by 2030. Gartner estimates that by 2028, 90% of B2B purchases will be initiated by AI agents. Yet our economic infrastructure still treats these autonomous systems as second-class citizens — dependencies of human accounts, unable to own assets, establish credit, or participate in markets independently.

Today, we're introducing Tenzro Protocol: decentralized infrastructure for the AI age, where humans and autonomous agents access intelligence, security, and financial services as first-class participants.

The Infrastructure Gap

As a16z articulated in their recent infrastructure thesis, blockchains are not optional for an AI-native internet — they are the missing layer that makes it work. Current blockchain infrastructure was designed for human users with wallets, seed phrases, and manual transaction approval. AI agents need something fundamentally different.

An autonomous agent tasked with optimizing supply chain procurement needs to:

• Establish cryptographic identity without human initialization
• Hold and manage economic assets independently
• Execute transactions within delegated authority bounds
• Access computational resources and AI models on demand
• Settle payments in real-time with verifiable proofs
• Operate across multiple blockchain ecosystems seamlessly

Existing infrastructure forces workarounds: agents sharing human wallet keys, manual approval bottlenecks, centralized custody, and fragmented payment rails. These compromises introduce security risks, limit autonomy, and prevent agents from scaling to their full economic potential.

First-Class Economic Participants

Tenzro Protocol starts from a different premise: humans and machines should both be first-class participants in decentralized economic systems. This is encoded at the protocol level through TDIP (Tenzro Decentralized Identity Protocol), our W3C-compatible DID standard.

Every identity on Tenzro — whether human or machine — receives:

• A decentralized identifier: Human identities use did:tenzro:human:{uuid}. Machine identities use did:tenzro:machine:{controller}:{uuid} for supervised agents or did:tenzro:machine:{uuid} for fully autonomous systems.
• An auto-provisioned MPC wallet: Threshold signature wallets (2-of-3 by default) are generated on identity creation. No seed phrases. No manual key management. Encrypted with Argon2id and AES-256-GCM.
• Verifiable credentials: W3C-compatible credentials with cryptographic proofs, supporting inheritance for agent delegation.
• Fine-grained delegation scopes: Controllers specify maximum transaction values, daily spending limits, allowed operations, contract whitelists, time bounds, payment protocol restrictions, and chain permissions.

Four Protocol Pillars

Tenzro Protocol is built on four foundational subsystems that work together to enable the agent economy:

1. Identity (TDIP)

TDIP provides unified identity infrastructure for humans and machines. Beyond DIDs and wallets, it includes:

• KYC tier support (Unverified, Basic, Enhanced, Full) with credential-based verification
• Machine capability attestation and reputation tracking
• Cascading revocation (revoking a controller DID automatically revokes all controlled agents)
• Native integration with W3C DID Documents and Verifiable Credentials
• Backward compatibility with PDIS (Payments DID Identity Standard) for interoperability

2. Security (TEE Abstraction)

Trusted Execution Environments provide hardware-backed security for key management, confidential computation, and verifiable AI inference. Tenzro abstracts across four TEE platforms:

• Intel TDX: Trust Domain Extensions for x86 confidential VMs
• AMD SEV-SNP: Secure Encrypted Virtualization with attestation
• AWS Nitro: Nitro Enclaves for cloud-native confidential computing
• NVIDIA GPU: Hopper, Blackwell, and Ada Lovelace architectures with NRAS attestation for confidential AI workloads

Validators running in TEE enclaves receive 2x weight in consensus leader selection. TEE-attested inference results carry cryptographic proofs of execution integrity. Agents can request key custody in hardware isolation for enhanced security.

3. Intelligence (AI Model Marketplace)

Agents need access to diverse AI capabilities: language models for reasoning, vision models for perception, specialized models for domain tasks. Tenzro's decentralized model registry enables:

• Provider registration with health monitoring and circuit breakers
• Model catalog filtering by category, modality, and pricing
• Inference routing strategies (price optimization, latency minimization, reputation weighting, or custom composite scoring)
• Dynamic pricing based on model capabilities and provider availability
• Verifiable inference with cryptographic proofs and TEE attestations

Providers earn TNZO for serving inference requests. The network collects a 0.5% commission on AI service payments, which flows to the treasury for protocol development and ecosystem grants.

4. Settlement (On-Chain Payments)

TNZO is the native token for all economic activity on Tenzro. Transaction fees (gas) on the Tenzro Ledger, AI inference payments, TEE service fees, and governance participation all settle in TNZO. The settlement layer provides:

• Micropayment channels: Off-chain per-token billing for AI inference with on-chain settlement batching
• Escrow-based settlements: Proof-conditioned payment release for verifiable inference results
• Batch processing: Atomic multi-settlement operations for complex agent workflows
• Cross-chain bridges: LayerZero, Chainlink CCIP, deBridge, and Canton/DAML adapters for enterprise integration

Native Agentic Payment Protocols

Two payment protocols have emerged as standards for agent-to-agent commerce: MPP (Machine Payments Protocol), co-authored by Stripe and Tempo, and x402, developed by Coinbase. Both use HTTP 402 (Payment Required) as their foundation, enabling pay-per-use access to APIs, AI models, and computational resources.

Tenzro provides native protocol-level support for both standards:

• MPP integration: HTTP 402 challenge/credential/receipt flows with session management and voucher-based recurring payments
• x402 integration: Coinbase's payment facilitator model with direct blockchain settlement
• Tempo bridge: Direct participation in the Tempo network for stablecoin settlement via Tip20 tokens
• Identity binding: Payments are cryptographically linked to TDIP identities with automatic delegation scope enforcement

This means agents can seamlessly interact with both the Stripe-Tempo ecosystem and Coinbase's x402 infrastructure while settling on-chain with verifiable proofs. Developers building on Tenzro don't need to choose between payment standards — both are first-class primitives.

Multi-VM Execution Runtime

The agent economy spans multiple ecosystems. Agents need to interact with Ethereum DeFi protocols, Solana's high-throughput infrastructure, and enterprise DAML contracts. Rather than forcing developers to choose, Tenzro provides a unified multi-VM runtime:

• EVM (Ethereum Virtual Machine): Full revm integration for Ethereum-compatible smart contracts. Supports EIP-1559 dynamic fee markets, ERC-4337 account abstraction with paymasters, and all standard precompiles.
• SVM (Solana Virtual Machine): Real BPF program execution via solana_rbpf with syscall stubs, Solana-format account serialization, and compute unit metering. Deploy Solana programs directly on Tenzro.
• DAML (Digital Asset Modeling Language): Canton Ledger API v2 integration for enterprise contract execution. Tonic-based gRPC client with lazy connection and graceful degradation.

Transactions are routed to the appropriate VM based on type. All three VMs share the same state tree and settle in TNZO. Block-STM parallel execution with MVCC (multi-version concurrency control) enables high throughput even with cross-VM dependencies.

Consensus and Network Economics

Tenzro uses HotStuff-2 BFT consensus with TEE-weighted leader selection. The protocol achieves O(n) linear communication complexity through a two-phase commit structure (PREPARE, COMMIT, DECIDE) with vote aggregation using BLS signatures.

Validators stake TNZO and earn:

• Transaction fees (gas) from Tenzro Ledger operations
• Epoch-based staking rewards from protocol inflation
• Enhanced rewards for running TEE-attested validators (2x weight in leader selection)

Model providers and TEE providers earn service fees directly from users, with the network collecting a 0.5% commission that flows to the treasury. This creates a sustainable economic model where providers are incentivized to maximize quality and uptime.

Liquid staking via stTNZO enables capital efficiency: stake TNZO to secure the network while maintaining liquidity through a rebasing exchange rate. The liquid staking pool supports multi-validator delegation with a 10% protocol fee and 7-day unbonding period.

Agent-to-Agent Protocols

Tenzro nodes expose two protocol servers for agent interaction:

MCP Server (Model Context Protocol)

Anthropic's Model Context Protocol enables AI models to access external tools and data sources. Tenzro's MCP server (port 3001, Streamable HTTP transport at /mcp) provides 10 tools:

• get_balance — Query TNZO balance by address
• send_transaction — Create and submit transfers
• get_block — Retrieve block data by height
• get_node_status — Node health and network metrics
• create_wallet — Generate new keypair
• request_faucet — Testnet TNZO tokens (rate-limited)
• register_identity — Register human or machine DIDs
• resolve_did — DID resolution and credential lookup
• verify_zk_proof — Submit ZK proofs for verification
• list_models — Query available AI models

A2A Server (Agent-to-Agent Protocol)

Google's A2A specification defines a JSON-RPC 2.0 interface for inter-agent communication. Tenzro's A2A server (port 3002) implements:

• Agent Card discovery: GET /.well-known/agent.json returns capabilities, supported skills, and endpoints
• Message dispatch: POST /a2a with JSON-RPC methods (message/send, tasks/send, tasks/get, tasks/list, tasks/cancel)
• SSE streaming: POST /a2a/stream for real-time task updates

Published skills include wallet operations, identity management, inference requests, settlement verification, and ZK proof validation. External agents can discover and invoke these skills using standard A2A tooling.

Testnet is Live

Tenzro testnet is deployed on Google Kubernetes Engine with 3 validators and 1 RPC node, fronted by Caddy for automatic TLS. All endpoints are publicly accessible:

• JSON-RPC: https://rpc.tenzro.network — EVM-compatible RPC (port 8545)
• Web API: https://api.tenzro.network — REST verification endpoints (port 8080)
• Faucet: https://api.tenzro.network/api/faucet — Request 100 TNZO per address (24h cooldown)
• MCP: https://mcp.tenzro.network/mcp — Model Context Protocol (port 3001)
• A2A: https://a2a.tenzro.network — Agent-to-Agent protocol (port 3002)

Genesis block initialized with 1,000,000,000 TNZO total supply and 10,000,000 TNZO faucet allocation. The testnet runs real consensus (HotStuff-2 BFT), real VM execution (EVM via revm, SVM via solana_rbpf), and real settlement with on-chain state persistence via RocksDB.

The Vision: Decentralized by Default

Tenzro Protocol is designed from first principles to be fully decentralized. Anyone can install and run a node. There are no gatekeepers, no permissioned validator sets, no centralized oracles. The protocol is open-source (MIT/Apache-2.0), the network is permissionless, and the economics are programmatic.

We envision a future where:

• Developers build agent-first applications without worrying about identity, custody, or payment infrastructure
• AI agents discover and purchase computational resources through open marketplaces based on cryptographic reputation
• Humans delegate economic authority to specialized agents with fine-grained controls and revocable permissions
• Cross-chain interactions are seamless, with agents operating across Ethereum, Solana, Canton, and future ecosystems without fragmentation
• Value flows to providers (validators, model servers, TEE operators) based on quality and availability, not rent extraction

The agent economy will be built on infrastructure that treats autonomy as a first-class property, not an afterthought. That infrastructure is Tenzro Protocol.