Runtime Governance and Execution Control Plane for AI Systems
Explore RIO through interactive demos, then install it locally. Each demo builds on the previous one, walking you through the complete Three-Loop Architecture.
Follow these demos in order to experience the full RIO protocol. Each one demonstrates a different aspect of governed execution.
Submit a high-risk action (e.g., transfer funds) and see how RIO requires human approval before execution. Approve or deny the request and watch the full pipeline execute with v2 cryptographic receipts.
Compare what happens when you approve vs. deny a request. See how RIO generates different receipts for each outcome and how denial receipts are cryptographically signed just like approvals.
Run multiple actions and explore the full audit trail. Each action produces a receipt and ledger entry. View the complete runtime log showing every pipeline stage.
Watch the complete governed execution pipeline run step by step. See every stage from intent creation through verification, receipt generation, and ledger append.
See how execution outcomes feed back into policy refinement. RIO analyzes patterns across multiple executions and proposes policy updates to improve future governance.
Paste any receipt JSON and independently verify its cryptographic integrity. Check hash validity, signature verification, and look up receipts by ID.
Browse the full hash-chain ledger. See how each entry links to the previous one via previous_ledger_hash, creating a tamper-evident chain of custody.
See what happens when someone tries to tamper with a receipt. Choose from 5 attack types and watch RIO detect the modification through cryptographic verification failure.
Run a sample governed request through the full pipeline with v2 cryptographic receipts.
Expected Output
Runs all 57 tests including 10 v2 receipt/ledger tests.
Then open your browser to http://localhost:8050
Then open http://localhost:8050
RIO is a governed execution system that sits between AI, humans, and real-world actions. It translates goals into structured intent, evaluates risk and policy, requires approval when necessary, controls execution, verifies outcomes, and generates cryptographically signed receipts recorded in a tamper-evident ledger. The system enforces the rules, not the AI.