Noloop

Overview

Noloop is a full-stack healthcare insurance claims platform built around one idea: make medical claim adjudication faster, more explainable, and more transparent for everyone involved. It pairs a role-based web experience for patients, hospitals, and insurers with an AI-assisted backend that handles OCR, document understanding, policy-grounded review, fraud-risk analysis, and decision support.

It is the production evolution of an earlier prototype (ClaimHeart, v1) — rebuilt as a real, end-to-end system.

The Problem

Medical claims get stuck. Manual document review, incomplete submissions, unclear policy interpretation, and fraud risk all create delay. Patients can't understand why a claim was approved or rejected, hospitals get caught in endless back-and-forth, and insurers need real auditability before they sign off.

Noloop closes that gap by turning unstructured claim documents into structured review workflows with visible reasoning — every automated decision is traceable, not a black box.

What I Built

• Role-based dashboards for three distinct workflows — patient, hospital, and insurer — each seeing the same claim through a lens built for them
• OCR upload pipeline for images and PDFs, converting raw documents into structured claim data
• Gemini-assisted document understanding — classification and structured extraction from messy real-world paperwork
• Claim queue, claim detail, fraud alerts, decision letters, reports, and policy views
• Glass-box adjudication — policy-grounded, evidence-backed recommendations with supporting rationale surfaced to the operator
• A multi-agent architecture spanning intake, policy review, medical/fraud review, and cross-validation

How It Works

1. A hospital (or stakeholder) uploads claim documents.
2. OCR + extraction convert raw files into structured claim data.
3. Policy and medical/fraud analysis review the claim against the evidence and the governing policy.
4. A final recommendation is surfaced with its reasoning, not just a verdict.
5. Patients, hospitals, and insurers each view the claim through their own role-specific dashboard.

Tech Stack

Frontend: Next.js, React 19, TypeScript, Tailwind CSS, Framer Motion, Zustand

Backend: FastAPI, Uvicorn, Python, SQLAlchemy, PostgreSQL

AI / OCR: Google Gemini, Groq, EasyOCR, Tesseract, OpenCV, PyMuPDF, pdf2image

Async & Infra: Celery, Redis, Docker, Prometheus + Grafana for observability

Use Case

Insurers and hospitals processing high volumes of medical claims who need to cut manual review time while keeping every decision auditable and explainable — surfacing fraud and anomaly signals early, and giving patients clear, understandable outcomes.

Status

The platform ships a polished, end-to-end multi-role frontend with demo workflows, working OCR + extraction services, and a FastAPI backend; the deeper multi-agent orchestration and async infrastructure are being expanded toward full production.