IAET AI – OCR Expense Assistant
A workflow-driven AI system that converts payment slip images into structured financial records.
Product Surfaces
Screens from the IAET AI workflow covering dashboard review, chat-based slip parsing, API structure, and the OCR parsing contract.
Financial Dashboard
Summary view with expense, income, and balance cards, plus filters by date, transaction type, and category.
Chat-based OCR Interaction
Users submit a payment slip and the assistant responds with parsed transaction data that can be confirmed into the expense workflow.
Expense Timeline Panel
The application stores categorized transaction records such as food, transport, shopping, bill, transfer, and other for dashboard review.
Slip Parsing Result
OCR text is passed through an AI parsing route that returns only normalized JSON for the transaction fields.
Overview
IAET AI – OCR Expense Assistant combines OCR, AI parsing, and a structured expense tracking workflow. Instead of leaving users with raw OCR text, the system converts payment-slip content into normalized financial records.
The available source screenshots show a full-stack Next.js application with API-driven processing, OpenAI-backed parsing, dashboard summaries, transaction routes, and an OCR route that returns structured JSON.
Problem
OCR alone produces noisy and inconsistent text. Payment slips may contain mixed language, bank metadata, reference numbers, timestamps, and merchant names that are not directly usable for expense tracking.
Manual expense entry is repetitive and error-prone because financial data from slips does not naturally flow from image to validated record to dashboard.
Solution
OCR text is passed into an AI parsing layer using the OpenAI SDK. The OCR route prompt instructs the model to extract financial transaction data, correct common OCR errors, and return only valid JSON.
The parser normalizes output into a strict transaction schema and integrates the parsed data into the dashboard and transaction workflow.
Delivery Scope
OCR Endpoint
API route for receiving OCR text from payment slip processing.
AI Parsing
OpenAI-backed parsing layer that normalizes OCR text into transaction JSON.
Transaction Schema
Structured fields for item, amount, type, category, merchant, bank, datetime, and reference.
Dashboard UI
Financial summary and transaction review surfaces for processed records.
Transaction Schema
Fields visible in the OCR route schema from the provided source screenshots
Description of payment
Normalized financial amount
Transaction direction
Expense classification
Merchant or counterparty name
Bank name when available
Datetime returned by the parser
Reference number when available
API Structure
API areas visible from the source tree and route screenshots
OCR parsing endpoint using OpenAI and strict JSON output.
General AI route visible in the source tree screenshots.
Transaction API area for stored expense records.
Summary API area for dashboard totals and financial overview.
Operational Modules
- OCR endpoint integration
- AI parsing logic using the OpenAI SDK
- Transaction schema validation through a strict JSON contract
- Chat interface for user interaction and confirmation
- Dashboard and financial summary UI
- API routes for transactions and summaries
Results
- Converted unstructured OCR data into usable financial records
- Reduced manual effort in expense tracking
- Created a consistent pipeline from image to structured data to dashboard
- Improved reliability of financial categorization
Tech Stack
Based on the portfolio repo and the provided source screenshots; no separate IAET source directory was present in this workspace.
Need an OCR expense workflow?
Let's discuss how I can help turn OCR output into structured, reviewable records for finance and operational workflows.