Ritual has launched Ritual Chain, an EVM-compatible blockchain designed to make AI inference, private computation and scheduled tasks part of its core infrastructure. Its central bet is that agents need more than a model API: they need persistent identity, permissions and funds of their own.
Watch Desk analysis
What happened
Ritual says the network lets smart contracts call language models, access the web, run private computation and schedule future work through 15 native precompiles. Its intended long-running agents can keep their on-chain identity and state, hold funds in non-custodial wallets and continue within set permissions and budgets.
The launch announcement also describes a system that delegates intensive computation to specialised nodes, then verifies the result on-chain rather than making every validator repeat the work. Ritual says the network’s public testnet has processed more than 40 million transactions since June. The release names Phala Network, Nous Research, LayerZero and Stargate among its launch partners. Ritual Chain’s launch announcement sets out the design and the company’s claims; it does not establish how the system performs under real-world use.
Our top picks
- On-chain model access
Contracts can call language models without treating inference as an add-on service. - Agents with persistent identity and state
An agent is designed to continue operating beyond the session or machine that created it. - Agent-controlled wallets
Agents can hold and spend funds within the permissions and budgets they are given. - Scheduled autonomous work
Native scheduling and a heartbeat mechanism are intended to keep permitted tasks running without an external cron service. - Delegated computation
Ritual’s execute-once, verify-many design aims to support heavier workloads without every validator repeating them.
Why it matters
This is a concrete attempt to build infrastructure around agents that act over time, rather than chatbots that answer and clock off. Persistent state, budgets and scheduled execution could make services such as recurring paid tasks possible without a human approving every transaction. Private computation is also part of the pitch, with trusted execution environments intended to shield sensitive inputs from operators.
That architecture makes the security and trust questions unusually important. An agent with a wallet and ongoing permissions has more room to do useful work, and more room to make a mess. Ritual’s announcement describes safeguards and design goals, not independent evidence that they work as intended.
Our read
Ritual is making a properly specific infrastructure bet: put model calls and autonomous tasks inside the chain, rather than bolting them on afterwards. If developers can use those tools through familiar EVM contracts, that is a lower-friction proposition than learning an entirely new development stack. The interesting test is whether persistent agents prove useful enough to justify giving them funds and authority, not whether a blockchain can fit “AI-native” on the box.
What to watch
- Whether developers can deploy applications and use the 15 precompiles as described.
- How Ritual documents permissions, budget limits and agent restarts.
- What independent testing shows about the delegated-computation and privacy mechanisms.
- Whether the launch partners produce working applications, beyond infrastructure integrations.
Discussion spark: Should long-running AI agents be able to control their own wallets and scheduled work, or should a person approve every meaningful action?
Sources and evidence
- Ritual Launches AI-Native Layer 1 for Autonomous Agents – TradingView (26 October 2026, 16:30 UTC)
- Source update (26 October 2026, 16:30 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.