CloneMe

Inspiration

The CloneMe project was inspired by the challenge of making online dating more authentic and efficient in a world where superficial swipes often overshadow meaningful connections. Conventional dating apps offer many problems: they are time-consuming, frustrating, awkward, and there is an uninviting social stigma surrounding them. Learning about the explosion of large language models and their potential impact, our team was passionate about AI’s potential to understand human behavior. Inspired by a blend of Sims, Marriage Pact, and AI, this led to our vision of an app where AI clones represent users’ personalities, making dating more exciting and fun while giving users an engaging topic to break the ice. The HackRice 2025 Social Track motivated us to leverage Google Gemini's AI capabilities to create a tool, CloneMe, that achieves these goals, revolutionizing how we can meet people!

What we learned

Building CloneMe taught us how to integrate AI with intuitive user experiences while navigating complex technical constraints. This was the first time any of us had used an API to integrate an LLM into any sort of project. We also expanded our ability to integrate front-end and back-end software. However, one of the hardest parts was designing and creating the graphical interface and designing and implementing the custom sprites we designed in our user interface.

What it does

CloneMe trains an AI based on the user’s text messages, multiple choice, and free response answers to various questions in order to generate a profile for the bot that properly encompasses the users texting style, personality, and interests. After an initial matching process, the bots get paired with a few others for whom they are to converse with. We trained another bot to be able to generate interactions between all the different bots, and so under the premise of a text conversation between two people testing compatibility, we obtain a medium-length conversation that imagines how the conversation might go in real life. Afterwards, we have a bot which judges the conversation and gives a compatibility score. The user can then scroll through their connections and follow up on the clones conversation’s, connecting with the other real person.

How we built it

We built CloneMe using a Python Flask backend to manage user interactions and serve dynamic web pages. User profiles, including names and preferences, are stored in an SQLite database for fast retrieval. Our AI module, written in Python, analyzes user answers and chat data to compute compatibility scores and generate realistic conversations. The front-end, crafted with Jinja2 templates, HTML, and CSS, mimics an iPhone interface with features like a swipeable carousel and animated GIFs for low matches. JavaScript adds interactivity, such as touch swipes and keyboard navigation, ensuring accessibility. The modular design allows easy scaling for new features like real-time chat.

Challenges we ran into

The main issue we ran into was the token limits for the Gemini API which made it difficult to work with. We also had to learn

Accomplishments that we're proud of

We’re proud that we completed this challenge despite this being all of our first real hackathons, making a product that we actually think could take off with some more work at colleges around the nation.

What's next for CloneMe

We plan on working on CloneMe beyond this hackathon as we think it could be an instant hit at campuses across the country. With some more work, we hope to release first at Yale and Rice University pairing up all bots with strong matches and then allowing those people to connect in real life, taking the pressure off of the dating experience and in doing so creating more natural, real connections. We would get a server to host our website on, make it unique to each college we release it at, expand some of the communication features, and find ways to improve the likeness between the bot’s and the person.