Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{205257,
        author = {Aditya Bagal and Saurabh Kokare and Yogesh Deokar and Prof.Richa Agarwal},
        title = {Multimodal AI System for Real-Time Expense Analysis and Forecasting},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {1},
        pages = {6712-6717},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=205257},
        abstract = {Keeping track of your spending isn't as easy as before. Now, your money info pops up everywhere—bank messages, UPI payments, scanned receipts, wallet apps, QR codes, and even voice commands. Most expense trackers still expect you to type every detail or toss your spending into bland categories. It’s annoying, and let’s be honest—your records end up messy and you lose interest pretty fast.
So we took a different approach. We built a Multimodal AI-Based Real-Time Expense Tracking and Forecasting System. It’s not just another app where you manually enter numbers. Ours lifts info straight from receipts using OCR, pulls out transactions from SMS by running NLP, records your spoken expenses, tags your location automatically, and ties everything together. You get it on Android, and your data stays safe thanks to Firebase Firestore and Authentication—it’s instant, secure, and tailored for you.
The real magic is under the hood. We use a hybrid Transformer-LSTM model with an Attention mechanism to predict your spending. Traditional models fall short since expenses aren’t just random numbers—they come in order, change over time, and always have context. Transformers grab the context, LSTMs remember the sequence, and the attention layer pinpoints what really matters. When we put it to the test with real and dummy transaction data, this combo outperformed plain LSTM, GRU, and old-school regression. Predictions got sharper, and the errors dropped.
But it’s not just about forecasting. The app automatically sorts your expenses, flags odd activity, shows everything on a live dashboard, builds graphs comparing your spending to predictions, and sends you alerts before you blow your budget. By connecting all these sources and layering on automation and deep learning, the app goes way beyond just tracking money—it helps you actually understand your spending and get better at managing it.},
        keywords = {Natural Language Processing (NLP), Optical Character Recognition (OCR), SMS Auto-Synchronization, Voice-Based Input, Transformer-LSTM, Attention Mechanism, Firebase Cloud Integration, Multimodal AI, Tracking Expenses, and Forecasting.},
        month = {June},
        }