AIHealthcare

AI-Powered Healthcare Intelligence Platform

Built by

Bill Blackmon

Senior AI Platform Engineer

I built AIHealthcare from the ground up to demonstrate what a single engineer can deliver with modern Spring Boot, Spring AI, and disciplined hexagonal architecture. Every line of code, every architectural decision, and every test in this 899-class codebase is mine.

Platform at a Glance

2,891
Automated Tests
104+
Data Sources
6
LLM Providers
88
Dashboard Pages
899
Java Classes
107
Domain Models
23
Prompt Templates
16
Pipeline Steps

What This Platform Does

AIHealthcare is a fully automated intelligence platform that monitors the AI-in-healthcare industry 24/7. Every day, it harvests articles from PubMed, openFDA, the Federal Register, HuggingFace, Healthcare Dive, Beckers, Google News, Perplexity, Y Combinator, and 47 other sources. Then it runs an 11-step pipeline that:

1

Scrapes and deduplicates articles from RSS, web pages, and APIs with SHA-256 change detection

2

Classifies sentiment on every article with LLM confidence scores and rationale

3

Scores companies across 6 competitive dimensions with radar chart visualization

4

Detects deal signals (M&A, funding, partnerships) with extracted amounts and counterparties

5

Harvests regulatory events from FDA and CMS, matching against subscriber watchlists

6

Compiles a knowledge wiki with source provenance, contradiction detection, and revision history

The output: a daily AI-generated newsletter, searchable dashboards for every dimension of the industry, and a multi-model AI search engine that queries Claude for Healthcare, OpenAI for Healthcare, Google for Health, and Perplexity Health simultaneously with numbered citation references.

Architecture: Why It Matters

This isn't a demo app bolted together with tutorials. It's a production system built on hexagonal architecture (ports and adapters) with a framework-free domain layer. The architecture enforces a rule: business logic never depends on infrastructure.

web (104 controllers) ──>  application (use cases)  ──>  domain (107 models + 119 ports)
                                                                  ^
                    infrastructure/* (adapters) ──────────────────+
                    - ai/          9 Spring AI adapters (Claude, OpenAI, Google, Perplexity, Amazon)
                    - ingestion/   RSS, web scraping, HuggingFace, PubMed, FDA, Perplexity
                    - persistence/ 70 JPA entities, repositories, storage adapters
                    - delivery/    Email (SES), NotebookLM export, transactional emails
                    - enterprise/  Data export jobs, push delivery, remote connectors, signed downloads
                    - scheduler/   Pipeline orchestrator, 16 externalized cron jobs

Swap Anything

Switch from Claude to OpenAI? Add a new data source? Change the database? Each change is a single adapter — zero domain modifications. I've done this live with 5 LLM providers.

Pipeline Isolation

The 16-step pipeline orchestrator wraps each step in try-catch. If sentiment analysis fails, trend detection still runs. If the FDA API is down, the newsletter still generates. No cascade failures.

Config-Driven

Add a company to competitive analysis? One YAML entry. Add a data source? One YAML block. Add a news topic? One config line. The system is designed to grow without code changes.

Built to Be Repurposed

The engine behind AIHealthcare is vertical-agnostic. The domain layer contains zero healthcare-specific logic. All industry knowledge lives in configuration: YAML feed definitions, prompt templates, and company lists. To create AI Finance, AI Legal, or AI Energy, you swap the config:

YAML

Feed Sources

104+ RSS feeds, APIs, and scraping targets — all defined in application.yml. Swap them for SEC filings, legal databases, or energy APIs.

TXT

Prompt Templates

23 externalized prompts control all AI behavior. Change the domain expertise without touching Java code.

CFG

Company Lists

Competitive analysis companies, watchlist defaults, and topic groups — all config-driven. No code changes to track a new industry.

Tech Stack

Java 17 Spring Boot 3.4.5 Spring AI 1.0.0 Spring Security 6 Spring Data JPA PostgreSQL 16 PGVector Claude for Healthcare OpenAI for Healthcare Google for Health Microsoft for Healthcare Perplexity Health Amazon Health Stripe Billing Amazon SES AWS EC2 Thymeleaf Tailwind CSS Chart.js TinyMCE Swagger UI Rome RSS Jsoup PDFBox CommonMark JUnit 5 Mockito AssertJ MockMvc

Plans & Pricing

Individual researchers to enterprise strategy teams.

Free
$0 /month

Daily AI healthcare newsletter — no platform access required.

Sign Up Free
Most Popular
Subscriber
$19 /month
  • Full platform — 88 dashboard pages
  • Semantic / vector search
  • Multi-model AI research queries
  • Vendor comparison reports
  • Regulatory alert watchlists
Start Free Trial
Try Enterprise
7-Day Demo
$0 / 7 days

Full enterprise access — 5 heavy AI calls per day.

  • All Enterprise features unlocked
  • Multi-LLM AI Audit (GPT-4o, Gemini)
  • Market Digest & Deal Signals
  • Framework competitive analysis
  • 5 heavy API calls/day (AI research)
Start 7-Day Demo
Enterprise
$399 /month

Less than an analyst-hour per day — and it runs 24/7.

  • Multi-LLM AI Audit (GPT-4o, Gemini, Nova)
  • Source provenance — study-level attribution
  • AI Accountability Tracker (19 health systems)
  • Confidence trend heatmap
  • What Changed? weekly diff
  • Enterprise Data Console — async export
  • Push Schedules — cron-based delivery
  • Remote Connections — HTTPS connectors
  • REST API access (10 keys)
  • 2,000 queries/month
Sign up for Enterprise

What I Bring

I'm a senior Java/Spring engineer who builds complete, production-grade systems — not prototypes. I write comprehensive tests, think in architecture, and ship features that work end-to-end from database to UI.

Full-Stack Delivery

This entire platform — domain models, AI integrations, persistence layer, REST APIs, 88-page UI, email delivery, Stripe billing, AWS deployment — is the work of one engineer.

AI-Native Engineering

Real experience integrating 5 LLM providers at production scale. Prompt engineering with evaluation frameworks, not just API calls. Cost-conscious design with caching, batching, and model selection.

Quality Mindset

2,891 tests aren't decoration. They're how I ship with confidence. Every feature has tests. Every adapter is mocked. Every controller has MockMvc coverage. The 7-day QA audit found zero critical bugs.