Inspiration We were inspired by the daily challenges faced by small and medium store owners across Latin America. Many of them only buy a narrow range of products, ignore new launches, or drop out of loyalty programs due to lack of personalization. We wanted to create a smart, fun, and data-driven way to re-engage them — with the help of Tua, our friendly guide.
What it does Tua is a personalized loyalty system for store owners that uses machine learning to generate weekly challenges based on their real purchase history. The platform motivates stores to diversify their product categories, increase purchase volume, and reactivate if they've stopped ordering. All while earning progressive cashback up to 2% based on their challenge performance.
How we built it We used:
Python to train ML models.
MongoDB to store product and store challenge data.
Gemini API that delivers personalized challenges.
Next.js to display challenge progress, rewards, and league rankings.
Challenges we ran into Designing challenges that are both personalized and scalable across thousands of stores.
Training accurate models with incomplete or imbalanced data.
Creating a simple user interface that store owners of any digital level can understand and enjoy.
Keeping the logic engaging but still tied to business KPIs (volume, category mix, reactivation).
Accomplishments that we're proud of Building an end-to-end loyalty engine that connects real sales data to ML-generated incentives.
Designing a gamified interface with clear progression, rewards, and a mascot character that humanizes the experience.
Creating a cashback system that aligns user motivation with business growth goals.
What we learned Personalization and behavior-based incentives outperform generic loyalty systems.
Gamification isn't just visual — it needs thoughtful progression, feedback loops, and small wins to keep users engaged.
Simplicity wins: every screen, message, and challenge must be clear and relevant to each store.
What's next for Tua
Expand challenge types (combo deals, limited-time streaks, team missions).
Scale the cashback model into a regional partner loyalty ecosystem.
Log in or sign up for Devpost to join the conversation.