Spur AI Chat Support

A full-stack AI-powered customer support chat application built as part of the Spur Software Engineer assignment.

Live Demo

https://spur-ai-chat-frontend.onrender.com/

---

Features

• AI-powered customer support chat
• Session-based conversation memory
• Persistent message storage
• Clean and minimal chat UI
• Start new conversation / reset chat
• Robust backend error handling

---

Tech Stack

Frontend

• SvelteKit

Backend

• Node.js
• Express.js

Database

• PostgreSQL (via Prisma ORM)

AI / LLM

• Groq API (LLaMA model)

---

Project Structure

/frontend /backend

Each folder contains its own README with setup instructions.

---

Notes

• The frontend and backend are deployed separately.
• Session IDs are used to maintain conversation context.
• The backend is designed with a clean service-based architecture for scalability.

---

Future Enhancements

The current version focuses on correctness, clarity, and production readiness. The following improvements can be explored to further enhance the product:

• User Authentication & Accounts Add optional user login to associate conversations with specific users across devices.

• Conversation History UI Allow users to view and switch between past conversations instead of only resetting sessions.

• Advanced Prompt Control Add configurable system prompts or domain-specific modes (e.g., sales, support, troubleshooting).

• Analytics & Monitoring Track response times, conversation volume, and common queries for better insights.

• Role-based Support Escalation Escalate conversations to a human agent when confidence is low or on explicit user request.

• Improved Error Handling & Retries Graceful fallbacks for LLM downtime, rate limits, or partial failures.

• Theming & Accessibility Support light/dark themes and improved accessibility (ARIA labels, keyboard navigation).

License

This project was built for evaluation purposes as part of a technical assignment.