How I Built an AI Tool to Read Invoices and Predict GST — Without Losing My Mind

#webdev #programming #ai #react

Why I Did This to Myself

Invoices are like bosses: they all look different, they all demand money, and they all ruin your day.
So I thought: “Why not let AI suffer instead of me?”
Boom. Invoice Digitization & Tax Prediction Tool.
Now my AI squints at messy PDFs and scans, while I sit back and pretend I’m productive.

The Tech Stack (aka Weapons of Mass Frustration)

- OCR → For turning “blurred coffee-stained PDF” into “blurred text file.”
- Regex → Because nothing screams fun like 200-character patterns.
- ML-ish GST Predictor → Basically a model that guesses taxes better than me on exam day.
- Python + pandas → To wrangle the chaos into something that looks like data.

How It Works

You upload an invoice.
OCR panics but spits out text.
Regex plays “Where’s Waldo” with GST numbers.
AI pretends to be an accountant.
You get structured output and tax predictions.

Bugs That Almost Broke Me

OCR Hallucinations:

Me: “That’s ₹1200.”
OCR: “Did you mean ‘12008S’?”

Regex PTSD:
Finding GST numbers in text feels like hunting shiny Pokémon.

The BTS Incident:

OCR once read “18% GST” as “BTS.”
For 2 minutes, I thought I’d invented a K-Pop tax predictor.

Code Snippet: My Regex Therapy Session

import re

def extract_gst(text):
    pattern = r"\d{2}[A-Z]{5}\d{4}[A-Z]{1}[1-9A-Z]{1}Z[0-9A-Z]{1}"
    match = re.search(pattern, text)
    return match.group(0) if match else "No GST found, cry harder."

invoice_text = "Invoice #123 GSTIN: 22ABCDE1234F1Z5 Amount: ₹1200"
print(extract_gst(invoice_text))  
# Output: 22ABCDE1234F1Z5

Results (Sort of)