Fizo AI

Overview
Data uploaded
Upload Dashboard
PDPA consent notice
Data filtered out
Data processing

Inspiration

Financial documents contain highly sensitive information, yet many AI tools require users to upload their data to the cloud. This creates concerns about privacy, security, and data leakage. We wanted to build a solution where users could benefit from AI without sacrificing control over their financial data.

This led us to FizoAI, built around one simple idea: “Bring the AI to the data, not the data to the AI.”

What it does

FizoAI is a local-first financial intelligence platform that helps users analyze financial documents quickly and securely.

Users can upload PDF, CSV, XLSX, and image files. FizoAI uses Gemini 3.7 Flash to assist with intelligent financial analysis while also validating uploaded content.

One of our key features is smart document filtering. FizoAI can identify and reject unrelated content, such as a picture of a cat, a building, or other non-financial images, before it enters the financial analysis pipeline.

The platform also provides explainable financial insights, allowing users to trace calculations back to their source data instead of blindly trusting AI-generated answers.

How we built it

We designed FizoAI around a local-first architecture, keeping sensitive financial data within the user's environment whenever possible.

Our system combines:

Gemini 3.7 Flash for AI-powered analysis
Document processing for PDF, CSV, XLSX, and images
Financial calculations and risk analysis
Context-aware document validation
Evidence tracing for explainable results
A local audit trail for tracking analysis activities

Our goal was to combine the flexibility of AI with the reliability of deterministic financial calculations.

Challenges we ran into

One of our biggest challenges was balancing AI capabilities with privacy and reliability. Financial analysis cannot simply depend on an AI model generating an answer.

We also needed to prevent irrelevant inputs from affecting the analysis. Teaching the system to distinguish between financial documents and unrelated content such as cat or building photos was an important part of building a reliable workflow.

Another challenge was making AI-generated insights transparent and verifiable rather than treating the AI as a black box.

Accomplishments that we're proud of

We are proud to have created a working concept that combines AI, financial intelligence, privacy, and explainability in one platform.

Our key accomplishments include:

Building a local-first financial analysis workflow
Integrating Gemini 3.7 Flash
Implementing intelligent filtering of unrelated uploads
Providing traceable financial calculations and evidence
Creating a privacy-focused audit trail
Turning complex financial analysis into a simpler workflow

What we learned

We learned that building AI for financial applications is not just about making AI smarter. Trust matters just as much as intelligence.

Through this project, we learned how to combine AI with validation, deterministic calculations, and evidence-based outputs. We also gained a deeper understanding of the importance of privacy-by-design when working with sensitive data.

What's next for Fizo AI

We want to take FizoAI further by expanding its financial analysis capabilities and supporting more document formats and financial use cases.

Future improvements include:

More advanced financial risk detection
Deeper financial forecasting and insights
Improved document understanding and OCR
More comprehensive audit and compliance features
Additional local AI capabilities
Enterprise-ready deployment for financial teams

Our ultimate vision is to make FizoAI a trusted financial intelligence workspace where powerful AI analysis does not require users to give up control of their sensitive data.