I Built an AI Expense Tracker That Understands the Way People Actually Type
Most expense tracking apps expect users to type in a fixed format.
But real people don’t track expenses like that.
Some people type:
“Paid 100 for petrol”
Some type:
“500 रुपये का पेट्रोल भराया”
Others naturally mix languages:
“Petrol ke liye 300 pay kiya”
And sometimes people combine everything together in a single sentence.
Most expense trackers completely fail at understanding this.
That’s the problem I wanted to solve while building Vitmora.
The Problem With Traditional Expense Tracking
I’ve tried many expense tracking apps over the years.
Most of them were good at charts, dashboards, and reports.
But almost all of them had the same problem:
They expected users to adapt to the app.
You usually have to:
- Select category
- Enter amount
- Choose payment method
- Add notes
- Fill forms repeatedly
After a few days, it starts feeling like work.
And honestly, most people stop tracking because of that friction.
But there was another problem I kept noticing.
Language.
People Don’t Think About Money in One Language
In countries like India especially, people naturally switch between languages while typing.
Sometimes it’s English.
Sometimes Hindi.
Sometimes Hinglish.
Sometimes regional languages.
And most of the time, it’s mixed naturally.
That’s how people actually communicate.
For example:
- “Paid 100 for petrol”
- “500 रुपये का पेट्रोल भराया”
- “Petrol ke liye 300 pay kiya”
- “Swiggy order 450 paid by UPI kal night”
These all mean something very clear to humans.
But most apps struggle to understand them.
That’s where the idea behind Vitmora started.
The Goal Was Simple
I didn’t want users to learn how to use the app.
I wanted the app to understand users naturally.
Instead of structured forms, the experience should feel more like chatting.
Just type normally.
Vitmora should figure out:
- amount
- category
- payment method
- merchant/context
- intent
automatically.
Building Multilingual Expense Understanding
This turned out to be much harder than it sounds.
Because users don’t type consistently.
For example:
“Petrol ke liye 300 diye”
“Petrol bharaya 500”
“500 ka fuel”
“Bike me petrol 200”
All of these technically describe the same kind of expense.
Then there’s mixed language input.
A single sentence might contain:
- English words
- Hindi words
- regional language
- abbreviations
- spelling mistakes
And users still expect the app to understand them instantly.
That became one of the biggest product and engineering challenges while building Vitmora.
Understanding Intent Instead of Perfect Grammar
One thing I realized very quickly:
People don’t type for machines.
They type for convenience.
Nobody wants to stop and think:
“Am I formatting this correctly?”
The app should understand intent even if the sentence isn’t perfect.
For example:
| User Input | What Vitmora Understands |
|---|---|
| Paid 100 for petrol | Fuel expense |
| 500 रुपये का पेट्रोल भराया | Fuel expense |
| Uber ke liye 300 pay kiya | Travel expense |
| Swiggy order 450 paid by UPI | Food delivery + UPI |
| Coffee 120 with card | Food & card payment |
That flexibility completely changes the feeling of expense tracking.
It starts feeling natural instead of robotic.
The Bigger UX Lesson
While building this, I realized something important:
Most software still expects humans to behave like software.
But real communication is messy.
People switch languages.
Use shortcuts.
Mix sentences.
Misspell words.
Type casually.
And honestly, that’s normal.
Technology should adapt to people — not the other way around.
Quick Demo
I recently recorded a small demo showing how Vitmora understands different styles of expense entries in real time.
Demo Video - https://www.youtube.com/shorts/p06VeqzKGNA
Examples used in the demo:
- “Paid 100 for petrol”
- “500 रुपये का पेट्रोल भराया”
- “Petrol ke liye 300 pay kiya”
- “J’ai payé 100 pour l’essence”
The goal wasn’t just multilingual support.
The goal was making expense tracking feel effortless.
What’s Next
Vitmora is still evolving, and there’s a lot more I want to improve around:
- natural language understanding
- multilingual support
- smarter categorization
- conversational finance tracking
- AI-powered insights
But this idea of “track expenses the way people naturally speak” has become one of the core philosophies behind the product.
Because managing money shouldn’t require perfect formatting or perfect English.
It should feel natural.
Thanks for reading.
Would genuinely love feedback from developers, builders, and anyone interested in AI, multilingual UX, or personal finance tools.
You can check out Vitmora here: https://vitmora.com