Building on Base

Trust Layer for

autonomous AI

An open-source protocol to build, ship, and govern agentic systems with cryptography. Wrap your LLMs in absolute on-chain boundaries.

+
+
Protocol Overview

Govern your agents

Enforce cryptographic boundary constraints at the protocol level. Define hard limits on execution scopes, on-chain permissions, and spend caps through immutable policy contracts.

NetworkBase L2
LicenseMIT
StatusIn Development
ArchitectureEVM + ZK
Core Features

Immutable Guardrails

Smart contract policies that cannot be altered after deployment — absolute boundaries.

Agent Registry

On-chain identity layer for registering, tracking, and revoking autonomous agents.

Sentinel Network

Decentralized verification nodes that validate every agent action against policies.

Open Source

Every contract, SDK, and node is MIT licensed, auditable, and forkable.

+
+
+
+

Performance matters

Cheapest agent verification

2,500×cheaper than Ethereum L1
50×cheaper than Arbitrum
$0for read-only policy checks
| $0.001 / Bouclier$2.50 / Ethereum L1 |

Cost per agent registration transaction (avg.)

Maximum throughput

133×more capacity than L1
faster block confirmation
2saverage block time on Base
| 15 TPS / Ethereum L12,000 TPS / Bouclier |

Network transactions per second capacity

+
+
How It Works
01

Register Agent

Deploy or connect your AI agent and register its unique bytes32 identity on-chain.

02

Attach Policies

Compose guardrail contracts that enforce specific constraints — transfer limits, scope restrictions, rate limits.

03

Verify Actions

Sentinel nodes intercept and validate every agent transaction against attached policies in real-time.

04

Audit & Govern

Full on-chain audit trail. Revoke compromised agents instantly with a single transaction.