Description
Summary ## Project Overview I'm looking for an experienced AI developer to build a simple MVP that automates part of the commercial lease review process using Claude. The goal is to create an internal productivity tool for lawyers that reduces repetitive review work while ensuring a human remains in control of the final legal decision. This is **not** a complex AI agent project. I'm looking for a clean, reliable implementation that can be expanded over time. ## Workflow The application should have a very simple workflow: 1. User uploads a lease (PDF). 2. The system extracts the lease text. 3. Claude reviews the lease against a predefined review checklist. 4. The AI evaluates against review criteria and flags clauses requiring attention (e.g. break clauses, repair obligations, outside of typical parameters). 5. The AI performs a second validation pass to minimise hallucinations or unsupported conclusions. 6. The system generates: * A structured lease review report. * A draft email summarising the findings and requesting approval from CEO. The user experience should be as simple as possible - upload a PDF and receive the outputs. ## Initial Scope The MVP should include: * Simple interface with drag-and-drop PDF upload. * Claude Enterprise API integration. * Prompt engineering for structured lease review. * Multi-step AI workflow (review → validation → report → email). * Structured output (JSON or equivalent internally). * Downloadable review report. * Downloadable email draft. * Clean, maintainable code. ## Nice to Have (Not Required) * FastAPI or Node.js backend. * Experience building AI document review tools. * Experience with legal AI or document processing. * OCR support for scanned PDFs. * Docker deployment. ## Future Phases If the MVP is successful, there will likely be additional work including: * Firm-specific review playbooks. * Knowledge base / precedent retrieval. * SharePoint or Microsoft 365 integration. * Outlook integration. * Approval workflows. * Suggested amendments. * Additional review categories. ## What I'm Looking For Please include: * Examples of AI applications you've built. * Experience working with Claude or other LLM APIs. * Your recommended technology stack. * Estimated timeline for delivering an MVP. * Any suggestions for simplifying the architecture while maintaining reliability. ## Budget Please provide a fixed-price quote for delivering the MVP, along with an estimated number of development hours. I'm looking for someone who values simplicity, good architecture and clean code over unnecessary complexity. This project is intended to become the foundation for more legal AI work if the MVP is successful.