Lit Protocol has just launched its new decentralised AI agent stack.

Description

Lit Protocol, which helps developers secure, manage, and automate accounts and actions across the web, has just launched its decentralized AI agent stack. It's available now, and it lets you create onchain agents that use wallet keys and transact on their own. You can also use it to encode guardrail policies.

Lit Protocol is unique in that it allows autonomous agents to operate without third parties having access to the passwords and keys they need to function. Their decentralised approach, which manages cryptographic keys across a network of trusted execution environments (TEEs), means agents can maintain control over their state, operate across public and private systems, and use sensitive information without trusted intermediaries.

As Co-Founder David Sneider sees it, "the combination of Large Language Models (LLMs) with decentralised infrastructure is leading to a huge change in software, with more and more moving towards intelligent autonomy."

The decentralised agent stack links LLMs, the agent's brain, with the internet's data, its view of the world. Lit Protocol gives developers the ability to program key management and to give agents autonomy over accounts, while setting up guardrails with signing policies.

The flexible architecture of Lit Protocol allows developers to interact with any LLM. API keys for LLM services are encrypted with the agent's decentralised keys (see example). Teams can also run private LLMs on Lit Protocol's devnet, which supports long-running compute in sealed TEEs.

 

Agents built with Lit Protocol have complete control over the accounts that represent their identities. They can use these accounts to authenticate across systems. This includes private "wallet" keys for signing blockchain transactions as well as API keys and passwords to web apps and services. No external parties, including the developer, can access these accounts. Users of the agent can set policies for spending limits, time locks, restricting unauthorised transactions, and requiring human approvals for particular actions.

"Our key insight is that while LLMs are great at semantic reasoning, they still need external tools – like blockchains and software platforms – to access keys," added Sneider. "With Lit's decentralised agent stack, developers can now build AI agents that aren't dependent on companies or human individuals managing their keys. This is a big step forward in enabling their independent agency as well as securing them against malicious actors."

Fancy building autonomous systems with Lit Protocol? Check out this video to get started with Lit as a signer. You can also fork the Lit Agent example from the video to create your first onchain agent policy and start sending autonomous transactions.