This post is my submission for DEV Education Track: Build Apps with Google AI Studio.
What I Built
This is AURA, a new (and ethical) app for small companies and freelancers who want discover new scenarios for their excels and PDF. You will discover new decisions based in the data that you have in your Excel or PDF files. Developed, thought, designed and every detail that I have made was did with these tools:
AI Models Evolution:
Started with Claude Sonnet 3.5 - For initial development and architecture decisions
Implemented Google Gemini Pro - As the core financial analysis engine in the backend proxy
Used Claude Sonnet 4 - For advanced debugging, UX improvements, and code optimization
Multi-model approach - Leveraged different AI strengths: Claude for development, Gemini for financial domain expertise, ChatGPT, Google AI Studio and of course, with Cursor PRO.
What it does:
Upload Excel, CSV or PDF → get instant analysis.
Ask questions in plain English: “Where am I overspending?”, “What if I raise prices 10%?”
Generates automatic insights, projections & what-if scenarios.
Why? Financial analysis often costs thousands €, out of reach for small businesses. Aura wants to democratize it.
And for me, this is the most important part and every developer should be consider take part of it: Ethics first:
No storage, local processing.
Not available for defense companies or countries violating human rights.
Demo
My Experience
What I Learned:
User feedback beats assumptions - I initially made UX recommendations without real data, highlighting the importance of user testing over theoretical "best practices".
Incremental improvements matter - Small fixes like scroll behavior and animation polish significantly impact user experience.
Cache management is critical - Proper file naming/hashing prevents deployment headaches and ensures users get updates.
Simplicity often wins - The overly complex "glassmorphism" thinking animation was confusing; clean, simple gradients worked better
What Was Surprising:
How much polish details matter - Users notice subtle things like scroll appearing during file uploads or thinking animations not showing properly
My own inconsistency - I contradicted my previous advice about UI patterns without solid justification, showing the importance of data-driven decisions.
Deployment complexity - Managing Google Cloud Storage, cache busting, and manual uploads requires more attention than expected
The iterative nature - We went through multiple animation versions before finding the right balance.
Technical Insights:
React useEffect dependencies matter for performance (scroll behavior)
CSS animation complexity doesn't equal better UX
Build pipeline optimization helps with cache invalidation
Error handling and loading states are crucial for production readiness
Process Learnings:
Beta-first approach is smart - get real user feedback before building advanced features
Version control and backups prevent lost work during rapid iterations
Clear communication about deployment processes saves time and confusion
Bottom line: Building a polished product requires attention to detail, user-centered thinking, and willingness to iterate based on real feedback rather than assumptions.
AURA BETA V2:
https://storage.googleapis.com/aura-financial-app/aura-complete.html
Feedback, feature ideas or questions are very welcome!
✉️ info@vertex-flow.com
Thanks for reading! 🚀
Top comments (0)