LEXY — Legal Solutions

Lawyers at a firm waste time answering the same basic questions over and over, while users don't know who to ask or how to start.

Team capstone project (Duoc UC, 2024). A legal platform pairing an AI chatbot to handle general questions and route complex cases to a lawyer, with a back office where the team picks up the query and follows up.

Create a clear entry point for general legal questions and connect cases that need attention with the lawyer team.

• Built the frontend of the public site: hero, services, lawyer team and contact form, with Next.js, Tailwind and shadcn/ui.
• Worked within a team on an end-to-end project (chatbot + back office), integrating my part with the rest of the app.
• Helped tune the AI: adjusting the system prompt and context so it answered only within the legal domain.

• Next.js and TypeScript for the application, MongoDB for persistence, NextAuth for authentication and OpenAI for the chatbot.
• The public site, chatbot and back office were integrated into a codebase shared by the team.

• Public site that introduces the firm and its services, with the Lexy chatbot as the entry point.
• Chatbot powered by the OpenAI API for general questions; specific cases are routed to a lawyer.
• Back office to receive the routed query and contact the person directly.
• Authentication with NextAuth and data persisted in MongoDB.

• Scoping chatbot context in a sensitive domain and routing what it should not resolve automatically.
• Integrating several contributors without confusing personal responsibility with the product's total scope.

• First time wiring a product to an AI: constraining answers with a system prompt and context in a sensitive domain.
• Real teamwork on a shared codebase, integrating my part with everyone else's.