The On-Chain Execution Layer
for AI Agents
Let your agents create and run DeFi workflows through the Ditto SDK — bounded by scoped session keys, secured by restaking, and non-custodial by default.
How it works
Agents describe intent; Ditto executes. The SDK ships an AI-agent skill so an LLM can author, simulate, and deploy on-chain workflows directly — while scoped session keys and policy limits keep funds safe. Triggers and execution run on a decentralized network with economic guarantees, so autonomous actions stay reliable and bounded.
- Ships an AI-agent skill — LLM-ready
- Scoped session keys — agents can't exceed their limits
- Non-custodial, restaking-secured execution
Why agent builders choose Ditto
-
LLM-ready SDK skill
A ready-made skill lets agents author and deploy Ditto workflows from natural-language intent.
-
Bounded by session keys
Agents act only within scoped, revocable permissions — no open-ended access to funds.
-
Non-custodial guardrails
Users keep custody; policy limits constrain what autonomous agents can do.
-
Flexible triggers
Cron, event, and on-chain conditions let agents act at exactly the right moment.
-
Economic guarantees
Restaking-secured operators with slashing ensure agent-initiated jobs actually run.
-
Simulate before execute
Dry-run workflows to validate behavior before any funds move.
-
Multi-chain
Agents operate across Ethereum, Base, Arbitrum, Polygon, and Optimism.
Key Features
Give your agents a safe way to act on-chain
Add bounded, non-custodial DeFi execution to your AI and crypto agents with the Ditto SDK.
Talk to SalesPrefer email?
[email protected]
Building an autonomous agent? Ask about the SDK's AI-agent skill and session-key guardrails.
Blockchain Backend for AI Agents FAQ
Talk to Sales-
How do AI agents execute on-chain with Ditto?
Agents use the Ditto SDK — including a ready-made AI-agent skill — to turn intent into workflows of triggers and contract calls, which a decentralized network executes on their behalf.
-
What stops an agent from draining funds?
Agents are limited to scoped, revocable session keys with explicit permissions and policy constraints, so they can only perform the actions you authorize.
-
Is it non-custodial?
Yes. Users keep custody of their assets; agents never gain open-ended control.
-
Which LLMs or agent frameworks work?
Any LLM or agent framework that can use the SDK or its skill file — the integration is model-agnostic.
-
Which chains are supported?
Ethereum, Base, Arbitrum, Polygon, and Optimism.
-
How do I get started?
Explore the SDK and its AI-agent skill, or talk to sales to scope your agent use case.
Explore what else can be built with Ditto.
What will you automate next?