Back to home

Bizzy - AI business expense tool for freelancers

, Designer and builder

Side project: how I built Bizzy, an AI expense app for Israeli freelancers. From sheets pain to a zero-cost, scale-ready MVP with CPA reports and Google Drive sync. Project ended.

Bizzy AI expense app interface for freelancers

TL;DR

I built Bizzy for myself: an AI-powered business expense app for freelancers under Israel tax law. Sheets were too manual and gave me no insight into what I owed. I worked with AI tools, researched competitors, shipped a one-user POC, used LinkedIn to find design partners, then expanded into a mobile-ready MVP with batch upload, income support, and CPA reports. Zero-cost to run. The project later ended.

The problem

As a freelancer in Israel, I need to manage expenses by tax law, save every receipt, record them properly, and pay taxes every two months. I am not the most organized person, and the tools I tried were not smart enough. They did not give me extra value.

Why sheets did not help

  1. Too much manual work. It took too much time.
  2. Paying the right amount of tax was a black box.
  3. No smart insights from the data I already had.

Goals

  1. Stop collecting and logging receipts by hand.
  2. Build a tool that costs me $0 to run.
  3. Use AI for smart insights, not just storage.

Process

  1. Build with AI tools I use: Base44, Lovable, AI Studio, Firebase Studio, ChatGPT, VS Code, Claude, Gemini
  2. Research competitors
  3. Research Israel tax law and edge cases
  4. POC product planning and initial scope
  5. Release for feedback
  6. Optimization
  7. MVP planning and scope
  8. Gradual release and feedback
  9. Deep optimization
  10. Go-to-market
AI tools used to build Bizzy: Base44, Lovable, AI Studio, Firebase Studio, ChatGPT, VS Code, Claude, and Gemini

Key findings

Competitor analysis showed a clear gap:

  1. Almost none of the tools used AI in a meaningful way.
  2. Pricing was high for little extra value.
  3. Users still did too much manual work.

Initial scope (POC)

Built for one user: me.

  1. Use AI to scan and analyze expenses
  2. Create full reports for my CPA
  3. Secure database
  4. Fast and reliable
  5. Not scalable yet
Bizzy initial POC product screens

Design partners

I used LinkedIn to get traction and gather design partners for feedback.

LinkedIn post used to find Bizzy design partners

After posting, feature requests poured in:

  1. Can I upload images too?
  2. Can you add batch upload?
  3. What about income? Can I add that as well?
  4. Does it work on mobile?

And more.

MVP scope (v2)

  1. Batch upload
  2. Support for images and PDFs
  3. Google Drive support
  4. Advanced reports for CPA
  5. Scale-ready for 50k users
  6. Mobile ready
  7. Income support
  8. Machine learning for smarter insights
  9. Onboarding for new users
  10. User tiers: Free and Pro

And more.

Final MVP

Scale-ready for 50k users. Full UX/UI and front end on React, with Firebase for auth and data, Google Analytics for events, and a design system I built for the product.

Final Bizzy MVP product screens

Walkthrough: Bizzy on YouTube.

What I learned

  1. How React apps are structured and what they can do
  2. How to store and use APIs securely
  3. How to structure, index, and store data in Firestore
  4. How to handle user scale with Firebase
  5. Event handling with Google Analytics
  6. How to build my own design system

Takeaway

A real personal pain is the best product brief. Ship a narrow POC, listen in public, then widen scope only when requests prove the need. The project ended after the MVP, but the build taught me end-to-end product craft: research, AI tooling, data, scale, and design as code.

Traction teaser

The short I made to get traction and design partners: