Ritual has launched its public mainnet, moving its AI-focused Layer 1 from testing into production infrastructure for autonomous software. According to the official mainnet announcement published on October 26, the network is designed to bring identity, AI inference and settlement into a shared blockchain environment, allowing software agents to operate with persistent permissions, capital and execution capabilities rather than relying entirely on human-triggered workflows.
The launch completes the production milestone anticipated when Ritual previously detailed its AI-native mainnet architecture and token structure. Ritual is now a live network rather than a proposed or testnet architecture, although the transition alone does not establish sustained developer demand or production adoption. The Foundation said its testnet hosted more than 2,000 agents and processed over 90 million transactions before mainnet, figures that describe pre-launch activity rather than current mainnet usage.
EVM++ Brings Specialized Compute Into the Protocol
Ritual differs from a conventional EVM Layer 1 through EVM++, its backwards-compatible extension of the Ethereum Virtual Machine. The architecture exposes specialized computation through native precompiles, sidecars and scheduling mechanisms, giving Solidity applications access to AI inference, trusted execution environments, zero-knowledge proving and other workloads that would normally require separate infrastructure. Ritual’s earlier architecture documentation also describes scheduled transactions that can trigger execution repeatedly under predefined conditions.
That architecture requires an important distinction between native access and literal onchain computation. Resource-intensive AI inference can execute in specialized sidecars rather than being replicated across every validator, while Ritual coordinates access, verification and settlement through the chain. The model is therefore different from simply placing an LLM inside blockchain consensus and is designed to avoid forcing all network nodes to reproduce computationally expensive AI workloads.
The approach also pushes farther toward autonomous execution than systems that retain explicit human authorization. For example, Aave’s MCP infrastructure for AI-assisted DeFi lets software prepare transactions while leaving signatures with users, while Algorand’s AC2 approval framework similarly separates agent requests from cryptographic authorization. Ritual instead targets agents that can operate within protocol-defined permissions and budgets without requiring a person to approve every individual action.
Mainnet Makes Adoption Measurable
Launching the chain makes Ritual’s architectural thesis testable under production conditions. Mainnet activity can now show whether developers actually use native inference, scheduling, external-service access and autonomous execution often enough to justify a dedicated AI-oriented Layer 1. The network’s native RITUAL asset pays transaction fees and specialized AI computation while also supporting staking, with an initial supply of 10 billion tokens and 18% unlocked at launch.
Ritual is entering a broader market where several blockchain projects are experimenting with machine participants through different architectures. Gensyn has connected autonomous agents to live market and settlement infrastructure, while other networks are building payment, authorization and execution layers around agent activity. Ritual’s differentiating claim is that these functions should exist closer to the base execution environment rather than being assembled from separate middleware services. Whether that distinction produces meaningful network effects will depend on observable production activity rather than architectural breadth alone.
The confirmed milestone is that Ritual’s mainnet is live and its autonomous-agent infrastructure has moved into a public production environment. Developer deployments, recurring agent execution, compute demand, validator participation and economically meaningful transactions can now provide a clearer measure of whether the network’s AI-native design translates into durable usage beyond its testnet phase.