> ## Documentation Index > Fetch the complete documentation index at: https://docs.manthan.systems/llms.txt > Use this file to discover all available pages before exploring further. # Verified Signals > How Parmana evaluates verified facts rather than AI outputs, user claims, or application assumptions # Verified Signals Verified signals are the foundation of policy evaluation in Parmana. A signal is a fact required by a policy. A verified signal is a fact that has been validated by an external system before it reaches Parmana. ```text theme={null} Required Signal ↓ Verification ↓ Verified Signal ↓ Parmana Evaluation ``` Parmana evaluates verified facts. Parmana does not verify facts itself. *** ## Why Verified Signals Exist Authorization decisions should be based on facts, not assumptions. Examples: ```text theme={null} Customer Is Authenticated Account Ownership Verified Balance Available Insurance Active Destination Account Verified ``` These facts must be verified before they can be trusted. Without verification: ```text theme={null} Claim ↓ Decision ``` With verification: ```text theme={null} Claim ↓ Verification ↓ Verified Fact ↓ Decision ``` *** ## Core Principle Parmana does not trust AI outputs. Parmana does not trust user claims. Parmana does not trust application assumptions. Parmana evaluates verified facts. ```text theme={null} AI Request ↓ Verified Facts ↓ Policy Evaluation ``` *** ## What Is A Signal? A signal is a piece of information required by a policy. Examples: ```text theme={null} insurance_active risk_score available_balance destination_verified customer_authenticated account_owner_verified ``` Signals are inputs to policy evaluation. *** ## Policies Define Signals Policies determine which signals are required. Example: ```text theme={null} Task ↓ Policy ↓ Schema ↓ Required Signals ``` Example schema: ```json theme={null} { "requiredSignals": { "customer_authenticated": { "type": "boolean" }, "available_balance": { "type": "number" }, "destination_verified": { "type": "boolean" }, "transfer_amount": { "type": "number" } } } ``` The policy schema determines required evidence. *** ## Signal Verification Signals must be verified before evaluation. Example: ```text theme={null} Transfer Request ↓ Verify Customer Verify Account Verify Balance Verify Destination ↓ Verified Signals ``` Verification occurs outside Parmana. *** ## Who Verifies Signals? Examples: ```text theme={null} Banking Platforms Identity Providers Fraud Systems Claims Platforms CRM Systems ERP Systems Compliance Platforms ``` These systems are responsible for verifying facts. *** ## Example Customer requests a transfer. ```text theme={null} Transfer ₹10,000 ``` Potential signals: ```json theme={null} { "customer_authenticated": true, "account_owner_verified": true, "available_balance": 50000, "destination_verified": true, "transfer_amount": 10000 } ``` Each signal must be verified before evaluation. *** ## Verified Signals Example ```json theme={null} { "customer_authenticated": true, "account_owner_verified": true, "available_balance": 50000, "destination_verified": true, "transfer_amount": 10000 } ``` These are facts. Not recommendations. Not predictions. Not opinions. *** ## What Parmana Receives Parmana receives verified signals. ```text theme={null} External Systems ↓ Verified Signals ↓ Parmana ``` Parmana does not receive: ```text theme={null} Raw User Claims AI Conclusions Unverified Data ``` *** ## AI And Verified Signals AI can help capture requests. Example: ```text theme={null} Customer ↓ AI Assistant ↓ Transfer ₹10,000 ``` The AI identifies intent. The AI does not determine authorization. The AI does not create verified facts. Verification occurs elsewhere. ```text theme={null} AI ↓ Captures Request Bank ↓ Verifies Facts Parmana ↓ Evaluates Policy ``` *** ## Banking Example Customer requests a transfer. ```text theme={null} Transfer ₹10,000 ``` The bank verifies: ```text theme={null} Customer Authenticated Account Ownership Verified Balance Available Destination Verified ``` Verified signals become: ```json theme={null} { "customer_authenticated": true, "account_owner_verified": true, "available_balance": 50000, "destination_verified": true, "transfer_amount": 10000 } ``` These signals are submitted to Parmana. ```text theme={null} Verified Signals ↓ Policy Evaluation ``` *** ## Insurance Example Customer submits a claim. The insurance platform verifies: ```text theme={null} Policy Active Customer Identity Verified Claim Amount Verified Risk Score Calculated ``` Verified signals: ```json theme={null} { "insurance_active": true, "risk_score": 12, "claim_amount": 100 } ``` Parmana evaluates policy. *** ## Why Verification Matters Without verification: ```text theme={null} AI Says Customer Is Verified ``` becomes: ```text theme={null} Trust AI ``` With verification: ```text theme={null} Identity Provider ↓ Verified Customer ``` becomes: ```text theme={null} Trust Evidence ``` *** ## Separation Of Responsibilities ```text theme={null} Customer ↓ Requests Action AI ↓ Captures Request External Systems ↓ Verify Facts Parmana ↓ Evaluates Policy External Systems ↓ Execute Action ``` Each component has a single responsibility. *** ## Auditability Verified signals improve auditability. Auditors can answer: ```text theme={null} Which Facts Were Required? Which Facts Were Verified? Which System Verified Them? Which Decision Was Produced? ``` This creates a verifiable evidence trail. *** ## Relationship To Policy Schemas Policies define schemas. Schemas define required signals. ```text theme={null} Policy ↓ Schema ↓ Required Signals ↓ Verified Signals ``` This makes evidence requirements deterministic. *** ## Relationship To Policy Evaluation Verified signals are the inputs to evaluation. ```text theme={null} Verified Signals ↓ Policy Evaluation ↓ Decision ``` Without verified signals, policy evaluation cannot occur. *** ## Complete Flow ```text theme={null} Task ↓ Policy ↓ Schema ↓ Required Signals ↓ Signal Verification ↓ Verified Signals ↓ Policy Evaluation ↓ Decision ``` Verified signals connect policy requirements to real-world facts. *** ## Core Insight AI can capture requests. External systems verify facts. Parmana evaluates policy. ```text theme={null} AI Captures Request ↓ External Systems Verify Facts ↓ Verified Signals ↓ Parmana Evaluation ``` This is how Parmana ensures that autonomous systems follow human rules before they act.