AI-Powered Healthcare SaaS Platform Development

We are building an AI-powered healthcare SaaS platform focused initially on cardiothoracic surgery. The goal of the software is to help surgeons maximize documentation quality and reimbursement opportunities before claims are submitted by analyzing operative reports immediately after dictation. This is NOT a generic chatbot project and NOT a simple OpenAI wrapper. The platform will process operative reports through a structured multi-stage AI pipeline developed in collaboration with a practicing cardiothoracic surgeon. The MVP workflow will be: 1. Operative Report Extraction Engine * Extract diagnoses, procedures, findings, complications, operative details, personnel, medical necessity statements, and complexity factors. 2. Medical Necessity Analysis * Determine whether the report adequately documents why the surgery was necessary. * Identify missing elements that may create reimbursement or denial risks. 3. Complexity Analysis * Identify documentation supporting increased procedural complexity. * Analyze cardiac surgery-specific factors such as: * difficulty weaning from bypass * ECMO conversion * multiple heparinizations * graft issues * unusual anatomy * additional procedures * bleeding complications * additional surgeon involvement * and other cardiothoracic-specific complexity indicators. 4. Revenue Opportunity Analysis * Identify documentation gaps that may reduce reimbursement. * Identify missed opportunities before claim submission. 5. Follow-Up Question Engine * Generate high-value questions that could uncover missing documentation, complexity factors, modifier opportunities, or reimbursement opportunities. 6. CPT Suggestion Engine * Suggest CPT codes based on documented evidence. * Provide confidence levels and supporting documentation. 7. Coder Summary Engine * Generate a concise coder-facing report summarizing findings, opportunities, supporting evidence, and unresolved questions. Important Architecture Requirements The application must NOT use one giant prompt. The system must operate as a staged pipeline where outputs from one stage feed subsequent stages. However, every stage must also retain access to the original operative report as the source of truth. The AI must provide: * supporting evidence * source references * confidence levels * traceable reasoning The platform should minimize hallucinations and unsupported conclusions. Technology Requirements Required: * Next.js * TypeScript * React * OpenAI API * Vercel deployment * API route development * Prompt orchestration * Modern UI development Strongly Preferred: * LangChain * AI workflow orchestration experience * Healthcare software experience * Medical documentation experience * HIPAA-aware development practices * Experience building production AI applications MVP Features Phase 1: * User pastes operative report * AI executes pipeline * Results displayed in separate sections * Evidence and confidence levels displayed * Exportable coder summary Future Versions: * Authentication * Secure storage * Appeals engine * Modifier engine * Global period analysis * Revenue recovery workflows * Multi-specialty expansion What I Need From You Please answer: 1. Have you built AI SaaS products before? 2. Have you built multi-step LLM workflows? 3. Have you worked with OpenAI APIs in production? 4. Have you deployed applications on Vercel? 5. Have you worked on healthcare software? 6. What architecture would you recommend? 7. Estimated timeline? The ideal candidate is someone who has built real AI products and understands how to design reliable multi-stage LLM systems, not just someone who can make API calls to ChatGPT. I am looking for a long-term technical partner if the MVP is successful.