Concierge

• 

  Agent-page

• 

  Client-page

• 

  Client-profile

Concierge was inspired by a simple problem: real estate clients often feel overwhelmed by contracts, inspection reports, and disclosures, while agents spend too much time repeating explanations and trying to spot which clients are anxious or at risk of dropping off. We wanted to build a tool that makes closings feel clearer, more human, and less stressful for everyone involved. Building this project taught us a lot about designing trustworthy AI experiences, especially how important transparency, fallback behavior, and real-time updates are when people are making major financial decisions. We built Concierge as a full-stack TypeScript web app using React, Express, Prisma, SQLite, Groq for AI responses, ElevenLabs for voice features, and Server-Sent Events for live updates between the client and agent dashboards. One of our biggest challenges was making all of these moving parts work together smoothly, especially syncing agent and client activity in real time, handling AI failures gracefully, and making sure the product felt useful and polished instead of just technically impressive.

Built With

• css
• elevenlabs
• express.js
• github
• groq
• node.js
• prisma
• react
• sqlite
• tailwind
• vite