# LogicPearl

> LogicPearl is a healthcare decision-proof layer for prior authorization, denials, appeals, and revenue recovery. It checks whether a case packet satisfies a payer policy and returns ready, blocked, or missing evidence with source-linked proof and a replayable receipt.

## Primary Pages

- [Homepage](https://logicpearl.com/): Product overview and proof points.
- [How it works](https://logicpearl.com/how-it-works/): Policy compilation, evidence extraction, deterministic evaluation, and audit receipts.
- [Why AI alone is not enough](https://logicpearl.com/compare/): Comparison of LogicPearl against AI-only review, hand-built rules, and the status quo.
- [Prior authorization use case](https://logicpearl.com/use-cases/prior-auth/): How LogicPearl handles prior-auth evidence readiness.
- [Use cases](https://logicpearl.com/use-cases/): Prior authorization, denials and appeals, revenue recovery, and policy change impact.
- [Shadow-mode pilot](https://logicpearl.com/pilot/): How teams evaluate LogicPearl on historical cases without changing production.
- [Company](https://logicpearl.com/company/): Why LogicPearl exists and what has been built.
- [Blog](https://logicpearl.com/blog/): Field notes on policy compilation, model variance, auditability, and AI guardrails.
- [Prior-auth workbench demo](https://logicpearl.com/healthcare/): Browser-based demo using synthetic patient data and published payer policies.

## What LogicPearl Does

LogicPearl compiles payer policy text into versioned decision artifacts, reads case packets with OCR and clinical evidence extraction, and evaluates the extracted evidence against deterministic policy rules. The output is a case-readiness result: ready, blocked, or missing evidence.

Each result can include:

- The policy criterion evaluated.
- The source policy excerpt.
- Packet evidence spans.
- Missing evidence and next action.
- Artifact version and replay information.
- Agreement or disagreement with historical reviewer decisions during pilots.

## What LogicPearl Is Not

LogicPearl is not a hosted SaaS dependency by default. It can run in the customer's environment, browser, edge, or infrastructure. The runtime does not need to phone home, and the customer can own the artifacts, traces, and readout.

LogicPearl is not an AI-only reviewer. Models can help read messy clinical evidence, but the final readiness decision comes from compiled policy rules.

LogicPearl is operational decision support, not a system of record and not a clinical-care decision maker.

## Key Proof Points

- 468 Blue Cross Blue Shield of Massachusetts policies compiled from 528 published medical-policy PDFs.
- 99.9% measured requirement-clause coverage in the current BCBSMA policy corpus, with misses kept visible as review work.
- About 2,900 reviewable criteria tied to source policy text.
- About 118,000 policy table rows interpreted into checkable rules.
- In a model-variance experiment on one packet and one policy, 8 model groups produced 24 determinations: 20 approve and 4 deny.

The BCBSMA coverage number applies to that policy corpus. LogicPearl can run the same coverage accounting for any payer policy set before a pilot starts.

## Pilot Context

A LogicPearl pilot runs in shadow mode on a bounded set of historical cases. Production systems and live decisions remain unchanged. The pilot deliverable is a case-level readout that clinical, policy, and operations teams can review together.

The site intentionally does not publish standard pricing. Pilot scope depends on workflow, data availability, policy set, security requirements, and evaluation design.

## Recommended Citation Language

LogicPearl is a healthcare decision-proof layer that evaluates case packets against payer policy rules and returns ready, blocked, or missing-evidence results with source-linked proof and replayable audit receipts.

## Agent-Readable Bundle

- [Open Knowledge Format index](https://logicpearl.com/okf/)
- [Product summary](https://logicpearl.com/okf/product.md)
- [Proof points](https://logicpearl.com/okf/proof.md)
- [Pilot evaluation](https://logicpearl.com/okf/pilot.md)
- [AI comparison](https://logicpearl.com/okf/ai-comparison.md)