IRIS AI Assistant

Main page of our web application
Support multi-language conversation, including English, Hindi, and Japanese

Inspiration

Millions of visually impaired individuals face challenges navigating e-commerce websites due to poor accessibility, complex layouts, and lack of auditory feedback. We wanted to create Iris an AI-powered shopping assistant that allows users to shop online independently and confidently through natural conversations. Our goal was to blend accessibility with intelligence, making online shopping inclusive for everyone.

What it does

Iris is an AI agent that helps visually impaired users search and shop for products on e-commerce websites like Amazon — using voice commands and conversational interaction instead of manual browsing. The user simply says what they want (e.g., “Find budget-friendly wireless earphones”). Apify scrapes real-time product data (price, reviews, rating, description) from e-commerce sites. Iris filters and summarizes the top product options, reading them aloud with clear descriptions. The conversation and preferences are stored in Snowflake, enabling Iris to learn and improve future recommendations. Stack AI automates the workflow capturing the conversation, invoking the scraper, and returning the results to the user seamlessly.

How we built it

Data Scraping: Used Apify to scrape structured product data (title, price, rating, link, description) from Amazon. Data Storage: Connected Snowflake to store user queries, product results, and conversation logs securely. Workflow Automation: Designed a Stack AI workflow to automate communication between the app, Apify scraper, and Snowflake database. Conversation Layer: Integrated natural language understanding to process user requests and return meaningful product summaries. Frontend: Created an accessible interface that supports voice interaction and can be extended for screen readers.

Challenges we ran into

Handling dynamic website structures during web scraping on Amazon. Ensuring consistent accessibility support for all interaction types (voice, screen readers, keyboard navigation). Managing real-time data synchronization between Apify and Snowflake without latency. Automating end-to-end workflows in Stack AI with reliable triggers and error handling.

Accomplishments that we're proud of

Successfully built a voice-based AI shopping assistant that retrieves and summarizes real e-commerce data. Integrated Apify, Snowflake, and Stack AI into a cohesive, automated workflow. Created a prototype that enhances accessibility for visually impaired users making online shopping more inclusive. Achieved smooth data pipeline automation and natural conversational flow between components.

What we learned

How to integrate multiple AI tools (Apify, Stack AI, Snowflake) for end-to-end intelligent automation. The importance of accessibility design and user empathy when building AI solutions for impaired users. Efficient data handling and transformation for dynamic, real-time applications. How small optimizations in workflow automation significantly improve user experience.

What's next for Iris

Add multilingual voice support to serve users globally. Integrate secure payment and checkout automation for hands-free shopping. Implement sentiment-based product ranking (e.g., prioritize positively reviewed items). Partner with major e-commerce platforms to make Iris a built-in accessibility feature for all shoppers. Expand to mobile and smart speaker platforms for a truly hands-free experience.