Inspiration
Living in India means navigating a hyper-fast financial landscape—from instant Zepto or Blinkit grocery deliveries and constant Swiggy/Zomato orders, to handling local maid salaries, monthly rent, and automatic Jio fiber recharges. It is incredibly easy to lose track of where your money goes. We were inspired to build a highly tailored, ultra-premium wealth management companion that feels like a private family office advisor, respecting both Indian spending habits and native regional languages.
What it does
VedaCFO reads transaction data and acts as a premium personal wealth dashboard. It visualizes local outlays using elegant, highly responsive horizontal custom bar graphs, highlights recurring subscription commitments, flags abnormal spending spikes, and includes a real-time smart entry interface. Most importantly, it features an interactive, dynamic AI CFO Chat Box that communicates perfectly across 6 major languages: English, Tamil (தமிழ்), Hindi (हिन्दी), Malayalam (മലയാളം), Kannada (ಕನ್ನಡ), and Telugu (తెలుగు).
How we built it
The application is built completely as an advanced React architecture styled with bespoke Tailwind CSS primitives. It handles high-performance data processing directly on the client side using robust React state configurations. The core intelligence engine is powered by the Gemini 2.5 Flash API, configured to return high-integrity, structured JSON arrays for precise local analytical graphs, multilingual system mapping, and conversational context-aware user responses.
Challenges we ran into
One major hurdle was optimizing complex transactional prompt contexts to consistently output flawless regional language semantics without breaking the JSON layout structure requested by our charts. We resolved this by creating explicit system guidelines, cleaning up potential raw markdown wrappers returned by the model, and building an automated mathematical calculation engine to act as a seamless fallback safety net.
Accomplishments that we're proud of
We successfully designed an incredibly immersive, luxury UI featuring 4 royal cultural theme options (such as Royal Gold, Monsoon Emerald, and Maharajah Ruby) alongside sleek interactive cursor trailing effects. We are especially proud that the entire application operates seamlessly across 6 distinct language modes, dynamically recalculating complex insights and switching conversational logic instantly.
What we learned
We learned how to design rigid system rules for Generative AI to enforce structured JSON output boundaries. We also mastered handling intricate fluid trail mathematics in pure React for desktop cursor rendering, and gained deep insights into designing accessible, localized software architectures tailored directly for diverse regional communities across India.
Built With
- gemini-2.5-flash-api
- javascript
- lucide-code-icons
- react
- tailwind-css
Log in or sign up for Devpost to join the conversation.