-
-
Team Photo
Inspiration
The inspiration came from observing the massive disconnect between digital velocity and physical retail. Trends like "Aura," the "Chill Guy," and the "Off-Duty CEO" create overnight demand for specific identities, yet traditional brands take months to react. We realized that any generic item could become a high-status asset if it were rebranded with the right "cultural code" at the exact moment a trend peaks.
What it does
TrendID is a trend-arbitrage engine. We identify emerging social signals and "re-skin" Shopify vendors' products to embody a specific internet subculture. By attaching viral terminology to everyday items, we allow consumers to purchase the "status" of a trend before it reaches mainstream saturation.
How we built it
In simple terms, we use a really long pipeline. We begin by identifying seven metadata tags per call via API-based web scraping. These tags are then fed into the YouTube Data API, powered by Google Cloud. This API scrapes two videos published within a short timeframe of each other with the highest viewer/hour gross rate over these seven metrics. These videos are then fed into the TwelveLabs Pegasus 1.2 API to uncover video-based semantic insights, ranging from descriptions of scenes in the video to analyses of which particular actions in these scenes place this video in the territory of these seven tags. The output is a structured JSON file focusing on the key aspects of a trend. This JSON file is paired with another JSON file that encapsulates all key data from the given Shopify store, including the types of products sold. The result is a semantic planning table that is pushed into a MongoDB database, the hive mind of marketable tweaks you can add to your Shopify store to maximize traction and sales.
Our backend is python and node-js based, used to interact with the shopify api and scraping. We are hosting the node js on render. We are using gemini. We made an embedded app for the app store. We used the Admin GraphQL to autonomously update the store page. We also made a test store to develop TrendID.
Challenges we ran into
Arguably, the coolest and hardest part to integrate was automating these changes to the Shopify store owner's admin page. The language structures required for this were new to us.
Accomplishments that we're proud of
We created an embedded app allowing the user to change their products, descriptions, themes, and pricing, among countless other data, by suggesting different changes they could make.
What we learned
Communication, especially in the early stages of finding an idea, is critical in defining the flow of the project.
What's next?
Allocated more time, this project can easily be implemented on a larger scale to exist at platforms such as Shopify
Log in or sign up for Devpost to join the conversation.