Customer AI Assistant
NovaCare is an AI-powered customer support assistant built with Next.js and Azure OpenAI, combining conversational memory, company-specific knowledge and safeguards against fabricated customer information.

Overview
NovaCare is an AI-powered customer support assistant designed to demonstrate how generative AI can be integrated into a modern customer experience.
The application allows customers to have natural conversations about products, orders, shipping, returns, refunds, subscriptions and warranties.
Rather than building a generic chatbot, the project focuses on a practical customer-support use case with company-specific knowledge, conversation context and safeguards around information the assistant cannot access.
NovaCare is a fictional company, and all products, policies, customers and orders used by the application are demonstration data.
The Challenge
Modern customer-support systems need to provide quick and useful answers while maintaining trust.
A generative AI assistant introduces additional challenges:
- Maintaining context across multiple messages
- Providing accurate company-specific information
- Avoiding fabricated order or customer information
- Protecting API credentials
- Clearly separating AI capabilities from unavailable backend systems
- Providing a responsive and intuitive customer experience
The goal was therefore not simply to connect a chat interface to a language model.
The goal was to build a small but complete AI application that demonstrates how conversational AI can be integrated responsibly into a customer-support experience.
The Solution
NovaCare combines a responsive Next.js frontend with a server-side Azure OpenAI integration.
The customer interacts with Nova through a conversational interface.
Messages are sent to a Next.js API route together with the current conversation history. The server combines the conversation with NovaCare-specific instructions and fictional company knowledge before sending the request to GPT-5-mini through Azure OpenAI.
The generated response is then returned to the frontend and displayed as the next message in the conversation.
This keeps the architecture intentionally simple while still demonstrating the core components of a real AI application.
Key Features
AI-Powered Customer Support
Nova provides natural-language assistance for topics including:
- Products
- Product pricing
- Orders
- Shipping
- Returns
- Refunds
- Damaged products
- Subscriptions
- Warranties
- Payment methods
Responses are generated dynamically using Azure OpenAI rather than predefined chatbot flows.
Conversation Memory
NovaCare sends the current conversation history with each AI request.
This allows Nova to understand natural follow-up questions.
For example:
Customer:
How much does the NovaBand cost?
Nova:
The NovaBand costs €89.
Customer:
How long is its warranty?
Nova can understand that "its" refers to the NovaBand discussed in the previous message.
This creates a more natural multi-turn customer-support experience without requiring persistent conversation storage for the demonstration.
Company Knowledge Grounding
Nova receives a dedicated NovaCare knowledge layer containing fictional
More Projects
Explore more AI projects, architectures and practical implementations.
AI Agent Framework
An architecture for building intelligent agents that can reason over tasks, use tools, collaborate and execute multi-step enterprise workflows.
Enterprise RAG Knowledge Assistant
A Python-based enterprise knowledge assistant that uses semantic retrieval, FAISS and Azure OpenAI to generate grounded answers with visible document sources.