Skip to main content

​
Why Parmana

AI systems are becoming increasingly capable of generating recommendations, initiating workflows, coordinating services, and executing actions. What remains unresolved is a much simpler question:
Did the right person approve this action before it happened?
Parmana exists to answer that question.

​
The Problem

Modern systems are becoming more autonomous. AI agents can:
  • Initiate payments
  • Approve refunds
  • Trigger infrastructure changes
  • Escalate healthcare workflows
  • Coordinate multi-step business processes
  • Interact directly with external systems
These systems can be highly intelligent. But intelligence is not authority. The central challenge is not whether an AI system can decide. The challenge is whether the system can prove it was authorized to act.

​
The Missing Layer

Most organizations already have:
  • Authentication
  • Authorization
  • Identity providers
  • Access control systems
  • Workflow tools
  • Audit logs
Yet a critical gap remains. Most systems cannot independently prove:
This specific action was approved by the appropriate human authority before execution.
Parmana introduces a new control layer: 
Human Authority
        ↓
Authority Verification
        ↓
Cryptographic Attestation
        ↓
Execution
        ↓
Execution Integrity Proof
This layer sits between authorization and execution.

​
Why Not IAM?

Identity and Access Management systems answer:
Who are you?
and
What permissions do you have?
Examples include:
  • Okta
  • Microsoft Entra ID
  • Auth0
  • AWS IAM
IAM systems are critical infrastructure. But they do not answer:
Did the right person approve this specific action?
A CFO may have permission to transfer funds. That does not prove the CFO approved this particular transfer. IAM verifies identity. Parmana verifies authority.

​
Why Not RBAC?

Role-Based Access Control answers:
Which actions may this role perform?
Example: 
Role: Finance Manager

Allowed:
✓ Create transfer
✓ Approve transfer

Denied:
✗ Modify production systems
RBAC determines capability. It does not verify decision legitimacy. Parmana verifies:
  • Decision provenance
  • Policy evaluation
  • Approval authority
  • Execution uniqueness
  • Cryptographic evidence
RBAC answers:
Could they do it?
Parmana answers:
Should this action be executed?

​
Why Not Workflow Engines?

Workflow systems orchestrate processes. Examples include:
  • Temporal
  • Camunda
  • Airflow
  • Step Functions
They are excellent at coordinating activities. They answer:
What happens next?
They do not answer:
Was this action properly authorized?
Workflow engines optimize execution. Parmana verifies authority before execution. The two systems are complementary.

​
Why Not Approval Systems?

Many organizations use approval chains. Examples include:
  • Procurement approvals
  • Expense approvals
  • Change management approvals
These systems often record: 
Approved by:
Jane Smith

Time:
10:42 AM
This creates a record. It does not create independently verifiable evidence. Questions remain:
  • Was the policy correct?
  • Was the approval modified?
  • Was the execution altered?
  • Can an external verifier confirm the decision?
Parmana creates cryptographic attestations that can be independently verified.

​
Why Not Policy Engines?

Policy engines evaluate rules. Examples include:
  • Open Policy Agent (OPA)
  • Cedar
  • Custom rule systems
They answer:
Does this request satisfy policy?
Policy evaluation is necessary. But policy evaluation alone is not sufficient. After evaluation, organizations still need to know:
  • Which policy executed
  • Which version executed
  • Whether the runtime was trusted
  • Whether the result was modified
  • Whether execution matched authorization
Parmana extends policy evaluation into verifiable execution authority.

​
Why Not Agent Governance Platforms?

A growing category of tools focuses on AI governance. These systems often provide:
  • Agent monitoring
  • Policy management
  • Risk scoring
  • Human review workflows
These capabilities are valuable. However, most governance platforms focus on oversight. Parmana focuses on verification. The central question is different. Governance platforms ask:
How should agents behave?
Parmana asks:
Can this action be proven authorized before execution?

​
What Makes Parmana Different?

Parmana answers one question:
Did the right person approve this?
If yes: 
Execute.
If no: 
Do not execute.
Everything else exists to support that guarantee.

​
Authority Verification

Parmana verifies authority before execution. This creates a deterministic control point between intent and action.

​
Cryptographic Attestations

Every approved decision can be represented as verifiable evidence. Attestations contain:
  • Policy identity
  • Policy version
  • Execution identifier
  • Decision outcome
  • Runtime provenance
  • Digital signature
These records can be verified independently.

​
Replay Protection

The same authorization should not be executed repeatedly. Parmana enforces execution uniqueness. This prevents accidental or malicious re-execution.

​
Execution Integrity

Approval alone is insufficient. Organizations must also verify that execution matched authorization. Parmana enables execution integrity verification through signed evidence and execution lineage.

​
Independent Verification

Verification should not require trusting Parmana. Verification should not require trusting application operators. Verification should not require trusting infrastructure providers. Evidence should stand on its own. Parmana is designed around independently verifiable outcomes.

​
Example: AI Agent Wants To Send Money

An AI system determines a vendor invoice should be paid. Without Authority Verification: 
AI Decision
      ↓
Payment Execution
Trust depends entirely on application behavior. With Parmana: 
AI Decision
      ↓
Authority Verification
      ↓
Signed Attestation
      ↓
Execution
      ↓
Execution Integrity Proof
The payment is no longer based solely on trust. It is backed by verifiable authority.

​
The Core Thesis

AI can generate requests. AI can generate recommendations. AI can generate plans. AI can generate actions. But AI does not possess authority. Humans remain responsible for consequential outcomes. As autonomous systems become more capable, organizations require infrastructure that ensures authority remains verifiable before execution. That infrastructure is Parmana.

​
In One Sentence

Parmana ensures that only human-authorized actions are executed in AI systems. Or more simply:
AI has intelligence. Humans have authority. Parmana verifies that authority before execution.