Follow the defined sequence
Patient Information ReceivedInsurance Validation
ALLOWED
Insurance validation is the next permitted step in the demonstrated policy.
Pranyx / Compliance Studio
Compliance Studio
Deterministic enforcement
for AI-driven workflows.
AI can propose an action. Pranyx independently checks whether the formal workflow permits it. See this principle in action in our Patient Intake demonstration.
Protected demo. Existing Pranyx access code required.
The problem
Putting an SOP in a prompt is not enforcement.
Instructions guide a probabilistic model. They do not guarantee that every proposed action follows a required procedure.
The AI does not judge
its own compliance.
Pranyx's approach places an independent decision boundary between a proposed action and permission to proceed. Our current simulation demonstrates that boundary for a defined Patient Intake policy.
The enforcement concept
From a written procedure
to an explainable decision.
- 01
SOP / Policy
The procedure defines the intended workflow.
- 02
Formalization
The procedure is represented as a formal model.
- 03
Proposed action
An AI or application proposes the next step.
- 04
Pranyx evaluation
Defined transitions and conditions are checked independently.
- 05
ALLOW / BLOCK
The decision follows the supported policy rules.
- 06
Formal evidence
The relevant condition and context explain the decision.
Demonstrated today with Patient Intake and a policy based on its generated formal model. Automatic runtime enforcement for arbitrary SOPs and interception of live AI actions are future capabilities.
Patient Intake / demonstration example
Same starting point.
A different decision.
Patient information has been received. What should the workflow be allowed to do next?
Attempt to skip validation
Patient Information ReceivedClinical Risk Assessment
BLOCKED
This action would bypass mandatory validation stages. Insurance validation must happen next.
Why this matters
A decision you can question.
Evidence you can inspect.
A separate decision boundary
The proposed action is checked by Pranyx, not approved by the AI that suggested it.
Rules, not another opinion
For the same supported Patient Intake state and context, the evaluator returns the same decision. No LLM decides ALLOW or BLOCK.
A reason and a compliant route
See which condition was required, which facts were used and which supported actions are permitted instead.
Formal evidence behind the outcome
Move from the business explanation to the exact transition and condition used in the evaluation.
For your technical team
The evidence is there
when you need it.
The protected Technical Evidence view exposes the formal model, state machine, rule definitions, evaluation response and generation trace.
Start with the guided Live Demo, then inspect the same system through a technical lens.
A Patient Intake proof of concept, not regulatory certification or a production enforcement gateway.