SugarDaddy

An AI-powered feature that surprises users with context-aware freebies, like coffee or movie tickets, based on things such as spending habits, weather, and location. Delight, reward, and grow loyalty.

Comment

Inspiration

Most rewards systems feel generic or transactional. We wanted to explore how AI could make rewards feel personal, timely, and truly delightful, not just based on spending, but also on real world context. We think AI agents could fit perfectly fit in this narrative of personalized presents, especially if there is a decent amount of context available for users based on their payment data.

What it does

Our project uses AI to analyze a user's spending habits and current context, such as weather, time, and location depending on what is relevant. It will, within a random periodic moment interval, analyse recent payments and the current context and based on that provide a small offer for the user to enjoy, such as a coffee or discount on groceries. It's a context aware micro-reward engine that drives user joy, loyalty, and engagement.

How we built it

We integrated with the bunq API to collect payment data and user's balance, then we used AI to analyze these transactions and if they could possibly be used for gifts. A backend service then implements an MCP with several context related things such as the bunq API or weather information. The agent selects the most relevant tools to use and takes this in their decision to decide whether a good gift could be provided. The frontend is a mobile web app that shows a surprise notification, it retrieves the latest offer and provides info about the freebie.

Challenges we ran into

  • Understanding and working with the bunq API efficiently in a short timeframe.
  • Designing a reward engine that feels useful and not random.
  • Implementing the MCP with tools
  • Limitations around how to present this as a demo
  • Less latency for offer creation, because of lack of time

Accomplishments that we're proud of

  • A working prototype
  • Integration with the bunq API
  • Use of NVIDIA API
  • MCP implementation
  • Interesting growth model

What we learned

  • Writing great code is tough in a short amount of time
  • Context is important and powerful
  • Keep things simple

What's next for the project

  • Expand the context engine with more inputs (calendar, social, seasonal).
  • Allow merchants to offer contextual rewards via the platform.
  • Refactor the code, it's a mess
  • A/B test reward styles (explicit vs. surprise, recurring vs. one-off).
  • Explore how this system could help with user onboarding, churn prevention, or saving goals.

Built With

Share this project:

Updates