Case study
Monetiq: AI and OCR document workflow
A personal finance app that reads receipts with OCR, sorts every expense into a category and uses an LLM to show where money can be saved.
Background: the Monetiq savings dashboard.
From a receipt photo to a savings suggestion
Monetiq turns everyday receipts into categorized expenses, then explains which spending is essential and where a saving is realistic. Screens below use sample data.
01
Receipts become categorized expenses
A user uploads a receipt image. Tesseract OCR extracts the text, such as the merchant and the amount, and a FinBERT model sorts the expense, zero-shot, into categories like Food, Transport, Bills and Shopping. No manual entry, no tagging by hand.

02
AI Financial Insights explain each finding
LLaMA 3.1, called through the Groq API, reviews the categorized spending. Essential costs such as an internet bill or daily commute are marked as ones to keep; non-essential spending gets a concrete suggestion and a potential saving.

03
A savings dashboard to track goals
The dashboard compares savings target with savings achieved, charts savings growth over time and tracks progress on goals such as an emergency fund, a vacation or a car.

Architecture
How Monetiq is built
- Web dashboardThe financial dashboard and AI Financial Insights screens.
- Spring Boot backendBusiness logic, expense records and calls to the AI service.
- Python AI microserviceEndpoints for classifying receipt images and analyzing expenses.
- Tesseract OCRExtracts text from receipt images.
- FinBERTZero-shot transaction categorization.
- LLaMA 3.1 via GroqExpense analysis and savings advice.
More of our work

HRive: one workspace for HR records, leave and recruitment
A role-based HR platform bringing employee records, attendance, leave and day-to-day HR work into one connected workspace.
AI Operations Assistant and three more blueprints
Illustrative architectures for problems we are equipped to solve, with the decisions that matter most.
Have a document workflow that stalls?
Tell us about the workflow. We will tell you honestly what it would take to build.
Discuss your project Usually answered within one working day.