Inspiration
One of our team members runs a small jewelry business on Etsy with her mother. Due to the high tariffs in Canada, it is extremely expensive for sellers to track products being shipped out. Additionally, many orders also get lost during delivery due to Canada’s snowstorms. This results in increased customer complaints and growing difficulties in responding to all inquiries promptly.
What it does
ShipSmart is an AI-powered customer service assistant built for small business owners who can't afford to spend hours answering the same shipping questions every night. An owner sets up ShipSmart in minutes, entering their business name and flagging common issues like storm delays or tariff surcharges, and ShipSmart handles the rest, saving an average of 1.6 hours per day. It delivers instant, warm, accurate replies 24/7, reducing negative reviews and customer frustration by clearly explaining that delays are caused by tariffs and extreme weather.
How we built it
We built ShipSmart using a Flask Python backend connected to the Groq API, which runs LLaMA 3.3 70B. The frontend is built in plain HTML and CSS. We integrated Cloudinary's Upload Widget for customer photo uploads and used Cloudinary's URL-based image transformations for the bead color preview and packaging overlay. We also used Claude throughout development to help write, debug, and iterate on our code.
Challenges we ran into
Almost everything was a challenge!! Especially since this was our first Hackathon, Cloudinary's transformation syntax was much trickier than expected; the order of transformation steps matters a lot, and we spent a long time debugging why background removal wasn't applying correctly. Getting the packaging overlay to look natural took multiple iterations, and we ended up using PhotoRoom to pre-clean the product photo. Overall, working with so many different APIs and tools simultaneously for the first time was not easy. Still, it was a stressful but very educational experience!
Accomplishments that we're proud of
Initially, we never planned on entering a sponsorship track as we were not confident we could make it within the time constraints. However, we are proud to say we eventually found a way to integrate Cloudinary into ShipSmart to elevate customer experience. If a customer wants to order a phone charm, instead of scrambling to get the vision across, they can illustrate what they want the phone charm to look like using Cloudinary’s generative colorize feature. There's also the Packaging Preview. Using Cloudinary's image transformation and background removal, we overlay the customer's charm directly onto our actual packaging photo, so they can see exactly how their order will arrive.
What we learned
We learned how to connect a Flask backend to a generative AI API and dynamically adjust the AI's behavior based on user settings. We also deepened our understanding of Cloudinary's transformation pipeline and how powerful URL-based image manipulation can be. We learned how to use Git and GitHub for the first time under pressure.
What's next for ShipSmart
Anytime a customer requests a refund in their complaint, ShipSmart will notify the business owner and give them the authority to approve or disapprove of the transaction, rather than giving an automated response. This leaves the ultimate decision of all transactions to business owners themselves. Next, we will integrate ShipSmart in more e-commerce platforms such as Shopify, eBay, Amazon, and not just Etsy. Finally, ShipSmart hopes to connect with shipping companies like Canada Post, UPS, and FedEx for more accurate responses.
Log in or sign up for Devpost to join the conversation.