• 

  Save2Win Proposed Architecture Diagram

• 

  UI

Inspiration

Our journey for the GKE 10th Anniversary Hackathon began with an unconventional idea: a "Demotivator Tool." The concept was to build an AI that would actively discourage impulse purchases and help users delay consumption to build long-term wealth. However, we realized a more positive, rewarding approach would be more engaging. We pivoted this core idea into "Save2Win," transforming the concept from demotivation into a gamified quest where users are celebrated for their good financial habits, creating a project specifically for the hackathon.

What it does

Save2Win is a gamified AI microservice, built on top of the Bank of Anthos application, that acts as a personal financial coach. It makes saving fun by turning it into a game. Users receive personalized quests generated by an AI (e.g., "Coffee Crusader: Make coffee at home 4 times this week for 500 XP!"). By completing quests and making smart financial decisions, they earn XP, unlock achievement badges, and build savings streaks. Crucially, Save2Win builds trust by only analyzing transaction data to provide insights; it never touches or manages the user's money directly.

How we built it

We built Save2Win for the GKE 10 Hackathon using a microservices architecture orchestrated with Google Kubernetes Engine (GKE).

The core of our unique approach is a custom AI Context Engine. This Python/Flask microservice calls the internal Bank of Anthos transactionhistory API to fetch raw transaction data. It then sends this data to the Google Gemini model to invent plausible merchant names and assign financial categories, enriching the basic ledger. The development process for the AI logic itself was accelerated significantly using AI Studio for rapid prototyping of prompts.

The main Save2Win Engine consumes this AI-enriched data to generate the user-facing quests and insights.

The frontend is a React dashboard (served by Nginx). All components are containerized with Docker and deployed on GKE using a Skaffold CI/CD workflow.

Challenges we ran into

Our greatest challenge was purely infrastructural, which severely limited the time we could spend on developing the actual application functionality. Our main bottleneck was getting the base Bank of Anthos application deployed and stabilized on GKE. We found its services to be brittle; logging into the generated user personas was fickle, and the application often failed to handle basic transaction data as expected.

As a result, a substantial amount of our hackathon time was consumed by numerous rounds of deployment debugging within the GKE environment. While AI Studio accelerated our creative process, this speed was consistently negated by the slow, frustrating cycle of tweaking configurations, debugging GKE permission errors, and redeploying the entire stack just to test a small change.

Accomplishments that we're proud of

Given that the majority of our time was spent wrestling with the infrastructure, we are immensely proud that we managed to build and deploy a stable, working end-to-end demo at all. Our biggest accomplishment was persevering through the deployment and debugging challenges to create a functional application. Taming the complexities of the Bank of Anthos deployment and successfully integrating our own AI services on top of it felt like a major victory and a testament to the team's persistence.

What we learned

This hackathon was a masterclass in the realities of cloud-native development. We learned firsthand that in a complex microservices project, a huge portion of the work is in deployment, configuration, and infrastructure management, not just writing feature code. We gained significant experience in debugging applications within a Kubernetes environment, navigating GKE permissions, and working around the limitations of a pre-existing, complex system. It also highlighted the stark contrast between the fast-paced, creative work enabled by tools like AI Studio and the meticulous, time-consuming nature of enterprise-grade infrastructure.

What's next for Save2Win: Your Personal Finance Quest

The future for Save2Win is focused on making personal finance a more social and goal-oriented experience. Our roadmap includes:

• Community Leaderboards: Introducing an optional feature to allow friendly competition among friends or community members.
  
• Goal-Oriented Quests: Allowing users to set real-world goals (e.g., "Vacation to Hawaii," "New Laptop") and having the AI generate a tailored quest-based savings plan to achieve them.
  
• Expanded Data Integration: Leveraging our scalable architecture to integrate with other data sources for more holistic financial coaching.
  

Built With

• gke
• mcp
• nextjs