Home/Work/NoteDoctor.AI
AI · API · MCP

NoteDoctor.AI
Prior Auth Engine.

Client
NoteDoctor.AI
Year
2025
Timeline
8 weeks
Services
RAG ArchitectureNext.js DevelopmentLangChainOpenAI IntegrationPublic REST APIMCP ServerOAuth 2.1AuthenticationStripe PaymentsUI/UX DesignDeployment
8wk
concept to production deployment
RAG
retrieval-augmented generation with LangChain + OpenAI
PDF
branded export of every generated authorization
API
public REST API and MCP server on the same engine
Overview

A full-stack RAG application that generates prior authorization summaries from clinical inputs — with a split-panel UI, swappable layout, PDF export, authentication, and Stripe billing.

AI that reads the rules so physicians don't have to.

Prior authorization is one of healthcare's most expensive administrative burdens — 80M+ requests processed annually, with 40% of denials ultimately overturned on appeal. Physicians waste hours each week navigating payer-specific medical necessity guidelines. NoteDoctor.AI asked us to build the engine that changes that.

We designed and built a production RAG application from scratch. Clinicians enter a diagnosis, CPT codes, and patient history; the system retrieves the exact payer guidelines for that case, runs them through an OpenAI-powered generation layer via LangChain, and surfaces a structured prior authorization summary with cited medical necessity criteria — all in seconds.

Every request produces a structured readiness report: a request overview, clinical context, authorization and medical-necessity criteria, relevant codes, and a required-documentation checklist that marks what the note already covers, with a clear determination such as "More information needed". Reports can be saved, reopened, and exported as branded PDFs. The application also ships with full user authentication, Stripe subscription billing, light and dark themes, a fully responsive layout, and a swappable split-panel interface (Request/Report tabs with a Swap Layout toggle).


Developer platform

The engine, opened up to other software and to AI assistants.

Screening API · public beta

The screening engine, as an API.

The same engine clinicians use in the app, exposed as a versioned REST API, so a health system can send a case from its EHR or internal tools and get an authorization-readiness determination back.

  • Two scoped APIs: Agents runs a full prior-auth screening (45–65 seconds); Chat answers questions grounded in the note, the payer policy and the run.
  • An in-app API Playground: pick an endpoint, edit the JSON, send it with a short-lived test key that never reaches the browser, and copy the call as cURL, JavaScript or Python.
  • Idempotency keys, so a retried request replays the same response instead of running (and billing) twice.
  • Per-key rate limits and /me and /usage endpoints, with usage metered straight into Stripe billing.
RESTAPI PlaygroundIdempotencyStripe metering
API keys · self-serve

Keys you can govern.

Organizations issue their own keys from the dashboard. Each key is a server-side secret scoped to the organization, and it is shown once, at creation.

  • Live and Test environments, labelled on every usage row; test keys run against a sandbox with simulated cases.
  • Scopes per key (agents, chat), granted at the minimum needed and never widened later: a new need means a new key.
  • A rate limit and an optional expiry (30 days, 90 days, 1 year or none) on every key.
  • Rotate by creating a replacement and revoking the old key; revoked and expired keys stop working immediately.
Scoped keysLive / TestRotationExpiry
MCP server · available now

Prior auth inside Claude, Cursor and ChatGPT.

The same engine, published over the Model Context Protocol. A provider adds one URL to Claude, ChatGPT or Cursor, signs in with their NoteDoctor.AI account through OAuth, and can run a full screening from the assistant they already use. No API key to copy.

  • A remote server: one HTTPS endpoint, nothing to install, host or keep up to date.
  • Five tools, from run_prior_auth_screening for the whole determination to Medicare and commercial guideline search, a policy extractor and key usage.
  • The same scopes, plan checks and rate limits as the REST API, and a tool the connection can't use is never listed.
  • OAuth 2.1 sign-in makes it a claude.ai or ChatGPT connector, including ChatGPT deep research; clients set up by config file can still connect with an API key.
MCPOAuth 2.1ClaudeCursorChatGPT

Screens
The challenge

Building a RAG system accurate enough for clinical decision support — where retrieval errors have real patient and billing consequences — while delivering a polished, full-featured SaaS product.

Clinical accuracy and product quality at the same time.

Most RAG demos fall apart under real-world use. Healthcare is less forgiving than most — payer guidelines span hundreds of pages, update frequently, and vary by plan and state. A hallucinated medical necessity criterion or a missed CPT requirement isn't just a bad answer; it's a delayed or denied treatment. We built a multi-layer retrieval strategy (dense + sparse search, reranking before generation) and enforced strict source citation on every output. The system declines to answer rather than guess.

On the product side, the challenge was delivering a full SaaS application — auth, billing, responsive design, adaptive UI, and PDF generation — within the same eight-week timeline as the AI work. We used Next.js throughout, integrated Stripe for subscription management, and built the swappable split-panel layout to give clinicians flexibility in how they use the tool across different screen sizes and workflows.


How we worked

A 8 weeks process built for speed and quality.

Week 01–02
Architecture & design
RAG pipeline design, vector store selection, UI/UX for the split-panel interface, auth flow, and Stripe integration planning.
Week 02–05
Core build
LangChain + OpenAI RAG pipeline, Next.js application, authentication, Stripe billing, responsive layout, and Swap Layout feature.
Week 05–07
PDF export & QA
PDF generation with branded templates, accuracy benchmarking against real prior auth scenarios, cross-device QA, and HIPAA compliance review.
Week 07–08
Deploy
Production deployment, monitoring, documentation, and handoff to the NoteDoctor team.
NoteDoctor.AI Team
Founding Team
NoteDoctor.AI
★★★★★
"CenterPoint built the entire product — the RAG pipeline, the UI, auth, billing, and PDF export — in eight weeks. They understood the clinical accuracy requirements from day one and the application they shipped is exactly what we envisioned."
Next project
1 Stop Property Maintenance
Web Design & Development
1 Stop Property Maintenance
Full website design, development, content creation, SEO, and deployment for a property maintenance company.
View project

Want results
like these?

Book a free 30-minute call and let's talk about what you're building.

Book a free call →