-
-
Screenshot of Auto Shopper in action
Inspiration
We wanted to build something that saves people time. Grocery shopping is repetitive, slow, and often requires scanning through dozens of items just to find what you normally buy. Our goal was to create an automatic grocery shopper that can quickly search Instacart for your preferred items, apply filters like dietary needs, and prepare your cart with a single action. Making everyday tasks faster and smarter inspired this project.
What it does
Auto Shopper takes in a list of items you want to buy and automatically searches for them on Instacart. It finds each product, compares options, and adds the best match to your cart. You can apply filters such as • organic • gluten free • cheapest available • brand-specific preferences
The system handles the entire browsing process for you, from searching to selecting to adding items.
How we built it
We built Auto Shopper using Python, custom AI agents, and automated browser control tools. The agent interacts with Instacart’s interface, identifies product listings, evaluates attributes like price and labels, and chooses the optimal match based on user preferences. We used prompts, rules, and structured planning to teach the agent what to look for and how to navigate the site intelligently.
Challenges we ran into
Training the agent to reliably navigate Instacart’s dynamic layout was a major challenge. The site uses elements that load at different times, which caused early versions to break. We also had to learn how to properly use Browser Use API keys, authenticate correctly, and handle errors when pages refreshed or selectors changed. Debugging agent behavior and timing was one of the hardest parts of the build.
Accomplishments that we're proud of
We’re proud that we built a fully working version of Auto Shopper by the end of the hackathon. The agent successfully finds items, applies filters, and adds products to the cart, which was our main goal. We also gained a deeper understanding of agent workflows and browser automation.
What we learned
We learned • how to design agent instructions that are reliable and repeatable • how to work with browser-automation tools and selectors • how to handle dynamic page content and inconsistent loading states • how to integrate API keys, environment variables, and secure authentication • how to collaborate and iterate quickly under hackathon constraints
What's next for Auto Shopper
Next, we want to expand support for more stores beyond Instacart, improve the agent’s speed, and add features like saved shopping lists, price comparison across stores, and user profiles. Long-term, Auto Shopper could become a full AI-powered shopping assistant that handles weekly groceries automatically.
Log in or sign up for Devpost to join the conversation.