Inspiration
Google's GenAI Xchange hackathon inspired us to create this project. The hackathon was a unique one in which 10 different companies had given 10 different problem statement to work on, each to be awarded one winner.
Me and my colleague are always on the lookout of creating our own startup, and this hackathon just happened to be the right place as most of the problem statements already had some sort of market validation. Also, we were pretty excited to experiment with Google's Gemini LLM model to see what we can build.
We chose the marketing track offered by Netcore to create our project as we believe that the future will be driven by social media and creator based economy, in which hyper personalized marketing will play a key role for businesses to thrive.
Alas, we won the track with our prototype and vision.
What it does
The project that we built, Marketing Genius, allows businesses to segment their customers and generate personalized ads for their customers specific to a particular platform. All of this can be done right inside our platform without having to depend on a 3rd party.
We provide the users to select from Instagram DM, Meta Messenger DM, Instagram Post , Meta Post, YoutTube shorts (yes, we generate videos as well) and email campaigns. All the multimedia content is generated at the click of a button.
How we built it
We used Google's VertexAI suite of GenAI APIs to create a web app that can generate multimedia content (images + text) according to the customer segment. Customer segmentation was done using K-Means algorithm, while the rest of the backend was created in Python Flask and hosted on GCP App Engine.
The frontend was built using NextJS and deployed on Firebase. We used Firebase's NoSql DB to store customer segment information.
Challenges we ran into
Integration of Facebook's API for programmatic content creation on Instagram and Messenger. Apparently we thought we would get a sandbox environment/ free credits to experiment like in GCP, so that we can play around with the API to learn more about its functionality. But this was not the case.
Accomplishments that we're proud of
1) Won Google's hackathon and got a lot of recognition. 2) Learnt how to develop backend in Python and utilize GCP to deploy the backend in a scalable and resilient manner.
What we learned
1) Thorough native development on GCP. 2) System Design for scalable and resilient software applications.
What's next for Marketing Genius
1) We are in talks to tie up with a very early stage startup called Solarplexus based out of Sweden. We have decided to join as co founders with the CEO of Solarplexus.
Log in or sign up for Devpost to join the conversation.