alab-mart

AI-powered e-commerce with RAG product assistance and real-time voice shopping.

Comment

Inspiration

I built alab-mart to explore how AI can make e-commerce more intelligent and interactive. Traditional shopping platforms require users to search, read product information, and manually manage their carts. I wanted to create an experience where users could simply ask questions, get intelligent product assistance, and control their shopping cart using natural language and voice commands.

This led to the idea of combining Retrieval-Augmented Generation (RAG) and voice AI inside a single e-commerce application.

What it does

alab-mart is an AI-powered e-commerce platform with two main AI features:

  • RAG Product Assistant – Users can ask questions about products, store policies, and recommendations. The assistant retrieves relevant information from the knowledge base before generating an answer.
  • Voice Shopping Assistant – Users can use spoken or typed commands to add products, remove items, update quantities, and perform shopping actions in real time.

The platform also includes product browsing, authentication, cart management, checkout, and order persistence.

How we built it

The application was built using FastAPI, a browser-based frontend, SQLite, Chroma, Ollama, Whisper.cpp, and gTTS.

The RAG pipeline uses embeddings and a Chroma vector store to retrieve relevant product and store information. The retrieved context is then passed to a local Qwen 2.5 3B model through Ollama to generate grounded responses.

For the voice assistant, Whisper.cpp is used for local speech-to-text processing. The transcribed command is interpreted by the shopping assistant and connected to the backend's cart operations.

The backend manages products, users, carts, orders, and AI assistant functionality, while the frontend provides the shopping interface and AI interaction widgets.

Challenges we ran into

One of the biggest challenges was integrating multiple AI and software components into one reliable application.

Building the RAG pipeline required handling embeddings, vector storage, retrieval, and local LLM inference while ensuring that responses were based on the available knowledge base.

The voice assistant also required configuring Whisper.cpp, speech recognition models, local binaries, and real-time communication with the shopping cart.

Another challenge was making the AI actually interact with the e-commerce functionality instead of functioning as a standalone chatbot.

Accomplishments that we're proud of

We are proud that alab-mart combines two practical AI applications in one complete e-commerce experience.

Instead of creating only a chatbot demonstration, we integrated AI directly into the shopping workflow. Users can ask questions about products through RAG and use natural language or voice commands to interact with their cart.

We are also proud of using local AI components, including Ollama and Whisper.cpp, allowing important parts of the AI workflow to run locally.

What we learned

This project taught us that building an AI application involves much more than connecting an LLM to an interface.

We learned how to work with:

  • Retrieval-Augmented Generation (RAG)
  • Vector databases and semantic search
  • Local LLM inference with Ollama
  • FastAPI backend development
  • Speech-to-text with Whisper.cpp
  • AI-powered application workflows
  • Connecting AI systems with real application actions
  • Integrating frontend, backend, databases, and AI components

Most importantly, we learned how to turn AI capabilities into useful product features rather than isolated AI demos.

What's next for alab-mart

The next step is to make alab-mart more intelligent and production-ready.

Future improvements include:

  • Personalized product recommendations based on user behavior
  • More advanced conversational shopping
  • Multilingual voice shopping
  • Better product ranking and hybrid search
  • AI-powered order tracking and customer support
  • Personalized shopping agents for individual users
  • Improved speech recognition and voice responses
  • Cloud deployment and scalable infrastructure

The long-term goal is to evolve alab-mart into a fully AI-native shopping platform where users can interact with the entire e-commerce experience through natural language.

Built With

Share this project:

Updates