Project Info

TwinSphere AI

This project did not submit a demo video on Devpost.

Inspiration

We were inspired by the gap between cold analytics and real human behavior. We envisioned a platform where businesses could interact with their target audience not as data points, but as dynamic, thinking "digital twins." Our goal was to breathe life into market research, making it more intuitive, predictive, and empathetic.

What it does

Twinsphere AI creates a virtual focus group of AI-powered customer personas. Companies can upload their ad copy, and our digital twins will react in real-time—commenting, liking, or ignoring it based on their unique personalities. This provides instant feedback on a campaign's potential impact before it goes live.

How we built it

Our backend is a powerful FastAPI server running Python, orchestrating AI agents built with advanced language models. The frontend is a sleek, responsive interface crafted with React and TypeScript. The entire system is containerized using Docker, ensuring smooth scalability for simulating complex market scenarios with thousands of digital twins.

Challenges we ran into

Our greatest challenge was avoiding generic AI responses. We fine-tuned our models to ensure each persona had a distinct, consistent personality and didn't just "hallucinate" answers. Optimizing the simulation engine to run complex scenarios with numerous agents in parallel without compromising speed was another significant but rewarding hurdle.

Accomplishments we're proud of

We're incredibly proud of creating AI agents that feel authentically human. When a "Skeptical Steve" persona gives a cynical but in-character critique of an ad, we know we've succeeded. Moving beyond simple metrics to generate qualitative, actionable insights is our biggest accomplishment, offering a deeper understanding of audience reception.

What we learned

We learned that the most powerful insights come from combining quantitative data with qualitative reasoning. Seeing why a persona disliked an ad is far more valuable than just knowing the engagement rate. This project reinforced the idea that the future of AI in marketing lies in its ability to simulate human nuance.

What's next

for Twinsphere AI The future is about deeper simulation. We're developing a "social graph" where digital twins can influence each other's opinions, creating more realistic and dynamic campaign forecasts. We also plan to expand our agent creation process, allowing businesses to build personas from a wider range of data sources for even greater accuracy.

Analysis

Compare with all teams

View

Metric

vijay
4

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

AnthropicIn code
CSSIn code
FastAPIIn code
HTMLIn code
LlamaIndexIn code
Mistral AIIn code
OpenAIIn code
PythonIn code
ReactIn code
TypeScriptIn code
LangChainClaimed
Node.jsClaimed
Tailwind CSSClaimed

10 of 13 appear in the indexed code. 3 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

48 KB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.