What is RINNAI?
RINNAI is a Solana-native AI automation layer that enables deterministic execution without custody risk. It transforms natural language user intents into secure, verifiable on-chain execution via autonomous agents.
Key Insight: RINNAI separates AI planning (probabilistic) from execution enforcement (deterministic). This enables intelligent automation while maintaining the security guarantees of blockchain.
Core Components
Intent Engine
Natural language processing system that converts user goals into structured execution parameters. The intent engine understands context, extracts constraints, and decomposes complex multi-step workflows.
Policy Engine
Validation layer that enforces user-defined constraints before and during execution. Policies define spending limits, allowed programs, slippage tolerance, and custom rules. All policies are validated on-chain for immutability.
Agent Orchestration
Coordinates multiple specialized AI agents to handle complex workflows. Manages parallel execution, dependency resolution, retry logic, and fallback strategies automatically.
Execution Program
On-chain Solana program that enforces policies using Program Derived Addresses (PDAs). Provides cryptographic proof of policy compliance and ensures deterministic outcomes regardless of off-chain component behavior.
How It Works
- 1User defines intent:"Swap 10 SOL to USDC when price reaches $150"
- 2AI generates plan:Select protocol, route, timing strategy
- 3Policy validates:Check limits, programs, slippage
- 4User signs:Approve execution with wallet
- 5On-chain enforces:Execute or revert safely
Why RINNAI?
Non-Custodial by Design
You always control your keys and funds. Agents never have custody.
Deterministic Outcomes
Despite AI's probabilistic nature, execution is guaranteed deterministic through on-chain enforcement.
Composable & Extensible
Build on templates, create custom policies, integrate with any protocol.
Complete Transparency
Every action is recorded on-chain with full audit trails and verifiability.