Χ-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?

CQF Clinical Reasoning on FHIR for Java

HIPAA compliant SaaS platform for Claims Processing, CMS 057-F CMS Regulatory FHIR APIs, EDI integration platform for healthcare payers. Process X12 transactions (270/275/277/278/837), FHIR R4 APIs, and CMS-0057-F compliance. Deploy on Azure, AWS (EKS), GCP (GKE), Digital Ocean (DOKS), Kubernetes.

The open-source AI layer for everything after the patient visit — prior auths, discharge summaries, insurance appeals

Atomic medical clinic AI workflow skills for EHR-integrated healthcare administration automation, SMART on FHIR planning, patient intake, prior auth, referrals, scheduling, chart prep, and human-in-the-loop clinic operations.

Production-grade Healthcare Prior Authorization framework featuring AI-driven OCR pipelines, HIPAA-compliant architecture, and secure JWT authentication.

An interoperable healthcare AI server (MCP) providing real-time insurance intelligence, prior authorization checks, and automated packet generation for clinical workflows.

Real-time agentic RCM prevention pipeline: Kafka streaming + LLM tool-use to intercept and score claims pre-submission against real NCCI/LCD rules · $0.003/claim · $150 denial value prevented · 50,000x ROI · 251 tests

Multi-agent AI platform that automates healthcare prior-authorization decisions. RAG-grounded, harness-engineered, observability-first.

AI governance audit system for healthcare decision models. Evaluates fairness (DPD/EOD), calibration, and robustness with NIST AI RMF-aligned reporting.

Go SDK and CLI for Smart Health Network participants: eligibility and Da Vinci prior-authorization exchange over FHIR.

FHIR-native prior authorization agent - turns a 5-day PA workflow into a 30-second automated clinical letter.

Provider-side oncology prior-authorization copilot · 7-agent LangGraph DAG · AWS Bedrock + Claude Sonnet 4.6 · CMS-0057-F native

LLMs auto-fill regulated prior-authorization forms from patient notes — with a validation pass and an LLM-as-a-Judge eval that scores correctness on clinical knowledge, not string matching. FastAPI + Streamlit, GPT structured outputs, SSE field streaming, F1 pipeline that exposed a flawed golden dataset.

Pre-launch payer coverage scorecard with RAG status and PA denial rate early warning system

TypeScript/JavaScript SDK for the Verity API - Medicare coverage policies and prior authorization

Simulated prior auth readiness handoff agent built with LangGraph, mock cases, deterministic rules, workflow traces, and HITL review.

Digital Health operator surface for prior-authorization evidence routing, payer rules, and approval-safe escalation.

Agent-native prior authorization protocol reference implementation and TypeScript SDK.

ML Model for Predicting Auto-Approvals of Prior Authorization Requests