How RINNAI Works

From natural language intent to deterministic on-chain execution. A step-by-step walkthrough of the RINNAI execution flow.

Policy Bounds & PDA Enforcement

What are Policy Bounds?

Policy bounds are immutable constraints enforced on-chain. They define the acceptable range of execution parameters, ensuring that no transaction can exceed user-defined limits even if off-chain components are compromised.

PDA-Based Enforcement

Program Derived Addresses (PDAs) are used to create verifiable on-chain state. Each policy is associated with a PDA that can only be modified by the user. The execution program validates all transactions against the policy PDA before any execution occurs.

Why This Matters

Traditional bots require complete trust. RINNAI's PDA-based enforcement means security is guaranteed by the blockchain itself, not by trusting a third party. Even if the AI components are compromised, they cannot execute transactions outside your defined policy bounds.

Glossary

Intent

A user's desired outcome expressed in natural language, including goals and constraints.

Policy

A set of rules and constraints that govern execution, such as spending limits and allowed programs.

PDA (Program Derived Address)

A Solana account address derived from a program ID and seeds, used for secure on-chain state management.

Policy Hash

A cryptographic hash of policy rules, ensuring immutability and verifiability on-chain.

Deterministic Execution

Execution that produces the same result given the same inputs, with no ambiguity or partial states.

How RINNAI Works

1. User Defines Intent

2. AI Generates Execution Plan

3. Policy Engine Validates

4. Agent Submits Signed Request

5. On-Chain Program Enforces

6. Result Recorded & Verified

Policy Bounds & PDA Enforcement

What are Policy Bounds?

PDA-Based Enforcement

Why This Matters

Glossary

Intent

Policy

PDA (Program Derived Address)

Policy Hash

Deterministic Execution

How RINNAI Works

1. User Defines IntentUsers express their automation goals in natural language. The intent includes the desired outcome, timing, and any constraints or preferences.

1. User Defines Intent

2. AI Generates Execution PlanThe intent engine analyzes the goal and generates an optimal execution plan. This includes selecting protocols, routing, timing, and transaction construction.

2. AI Generates Execution Plan

3. Policy Engine ValidatesBefore any execution, the policy engine checks all constraints. This includes spending limits, allowed programs, slippage tolerance, and custom rules.

3. Policy Engine Validates

4. Agent Submits Signed RequestOnce validated, the agent constructs the complete transaction and the user signs it. The signed request is submitted to the on-chain execution program.

4. Agent Submits Signed Request

5. On-Chain Program EnforcesThe Solana program verifies all signatures, validates PDAs, and enforces policy constraints immutably. No execution happens without passing all checks.

5. On-Chain Program Enforces

6. Result Recorded & VerifiedThe execution result is recorded on-chain with complete audit trail. Success or failure is deterministic and verifiable by anyone.

6. Result Recorded & Verified

Policy Bounds & PDA Enforcement

What are Policy Bounds?

PDA-Based Enforcement

Why This Matters

Glossary

Intent

Policy

PDA (Program Derived Address)

Policy Hash

Deterministic Execution