Inspiration
A huge amount of online shopping is driven by people who are interested but not quite ready to buy. They’ll look at the same product multiple times, compare different sizes or colors, maybe even visit a store to try it on, and then leave without checking out. Most of the time, it’s not that they don’t want the item but that they’re unsure about the price or the timing. They might be waiting for a sale, a discount code, low-stock pressure, or simply a sign that it’s the right moment to purchase. The issue is that this hesitation isn’t captured in any meaningful way. Once someone leaves a site, the momentum disappears. Shoppers forget about the product or get distracted, and merchants have no way of knowing whether that visitor was casually browsing or genuinely close to buying. There is real purchase intent in those moments, but nothing connects that intent to the right signal at the right time. As a result, a large number of potential transactions never actually happen.
What it does
OmniCart uses AI to identify when someone is genuinely interested in buying something, both online and in physical stores. On your laptop, our browser extension looks at signals like how long you stay on a product page, whether you come back to it multiple times, and whether you’re clicking through different sizes or variations. When those patterns suggest real intent rather than casual browsing, the item is automatically added to a tracked list. In stores, you can open the Omnicart app and scan a barcode or take a photo of something you’re considering. The app will find the exact product or comparable listings online across different retailers and marketplaces. You can either buy it right away or add it to the same unified watchlist you use on desktop. Once an item is being tracked, Omnicart continues monitoring it in the background. It keeps an eye on price changes, availability, and new discounts. If something significant happens, like a price drop, low inventory, or a special offer, you’re notified right away so you can act while the opportunity is still there.
How we built it
We built Omnicart as two separate systems, web and mobile, connected through a shared backend so everything stays synchronized in real time. On the web side, Bauer built both the frontend and backend infrastructure. This included developing the Chrome extension interface, implementing the shopping intent detection logic, designing the unified watchlist system, building the backend API, creating the price and availability monitoring engine, and setting up the real-time notification and email workflows. On the mobile side, Daniel built both the frontend and backend of the app. He implemented barcode scanning and camera integration, developed the product recognition pipeline, integrated OpenAI’s API for image understanding, connected Bright Data’s API to retrieve listings across retailers and marketplaces, and ensured that mobile activity synced seamlessly with the shared backend and watchlist system. This structure allowed both platforms to operate independently while feeding into the same core intelligence layer that powers Omnicart.
Challenges we ran into
One of the biggest challenges we faced was building a backend system that could reliably take a photo of a product and return accurate, usable results. Image recognition by itself isn’t sufficient, since many products look similar but differ by brand, model, size, or seller. A small mismatch can lead to completely irrelevant listings, which breaks trust in the product. To solve this, we carefully integrated OpenAI’s API to extract structured product information from the image rather than relying only on visual similarity. That information was then used with Bright Data’s API to pull precise listings from retailers and marketplaces. We spent a significant amount of time refining how these systems worked together to make sure the results were relevant, consistent, and fast enough to feel seamless for the user. Finding the right balance between accuracy, speed, and reliability ended up being one of the most technically demanding parts of building Omnicart.
Accomplishments that we're proud of
We’re proud that we were able to take this idea and turn it into a working product within a single hackathon. As first-time hackers, building a system that connects a Chrome extension, a mobile app, and a shared backend felt like a significant accomplishment. It required us to think beyond individual features and focus on how everything would function together as one cohesive system. Throughout the process, we learned a great deal, especially about integrating external APIs, structuring backend logic, and ensuring that different components could communicate reliably in real time. Getting the entire system to work end to end, from scanning a product to tracking it to receiving a notification, was a moment we’re genuinely proud of.
What we learned
Through building Omnicart, we learned how to connect web and mobile systems into one cohesive product. Working with Chrome extension architecture taught us how service workers, background scripts, and content scripts communicate and manage shared state. On the backend, we gained experience integrating APIs, building a monitoring system, and handling real-time data updates. Integrating OpenAI and Bright Data required us to structure API calls carefully, manage asynchronous responses, and ensure reliable results. Overall, we gained hands-on experience building a full-stack product from scratch and learned how to debug quickly, iterate under time pressure, and adapt when things didn’t work as planned.
What's next for Omnicart
Next, we plan to improve how we model shopping intent by incorporating more behavioral signals and historical pricing data. We also want to expand retailer coverage and make product matching from image scans more accurate and consistent. On the merchant side, we see potential in building tools that allow brands to offer targeted incentives to high-intent shoppers instead of relying on broad, site-wide discounts. Over time, we would like to test Omnicart with real users to refine the alert logic and better understand how timing and pricing signals influence purchasing decisions.
Log in or sign up for Devpost to join the conversation.