LexDraft: Prelegal AI Document Curation

Problem

Why This Project Was Built

Founders and operators often need structured legal drafts quickly, but gathering required clauses and fields manually is slow, inconsistent, and error-prone.

Solution

How It Works

A two-stage AI workflow first identifies the right document type, then collects required inputs through guided chat and compiles a polished draft for export.

Core Capabilities

• Supports six document types including NDAs, offer letters, DPAs, cloud agreements, and AI addenda.
• Maintains multi-draft sessions with create, switch, rename, reset, and delete controls.
• Tracks completion across required fields and renders final sections when drafts are complete.
• Includes authentication, session-aware workflows, and browser-side PDF export.

Technology Stack

• Next.js App Router frontend (TypeScript)
• FastAPI backend (Python)
• SQLModel with SQLite persistence
• OpenAI-powered selection and field-collection turns
• Dockerized local runtime scripts

Project Outcomes

• Reduced drafting friction through a guided conversational flow.
• Created a reusable architecture for additional legal templates.
• Delivered a production-style full-stack AI product that can be demoed locally.