ManualTrigger Automate

Target Audience

• Banking Professionals: Individuals working in finance who need to analyze bank statements efficiently.
• Data Analysts: Professionals who require a streamlined method to extract and analyze data from documents.
• Developers: Those looking to integrate document processing workflows into applications.
• Small Business Owners: Entrepreneurs who manage their own finances and seek to automate data extraction from bank statements.
• Students: Learners studying finance or data science who need practical examples of document processing.

Problem Solved

This workflow automates the tedious process of extracting and analyzing data from bank statements, transforming them from PDFs to markdown format. It effectively handles both downloaded and scanned PDFs, ensuring that critical data such as deposits and withdrawals are accurately captured and formatted for easy analysis. By leveraging Vision Language Models (VLMs), it provides a solution where traditional OCR methods may fail.

Workflow Steps

1. Manual Trigger: The workflow starts when the user clicks 'Test workflow'.
2. Get Bank Statement: Downloads the specified bank statement from Google Drive.
3. Split PDF into Images: Converts the PDF into separate images for each page using a web service.
4. Extract Zip File: Unzips the downloaded images to prepare for processing.
5. Images To List: Converts the images into a list format for easier handling.
6. Sort Pages: Organizes the images based on their filenames.
7. Resize Images For AI: Resizes the images to optimize them for processing by the AI model.
8. Transcribe to Markdown: Uses a Google Gemini Chat Model to transcribe the images into markdown format, ensuring all text and tables are accurately captured.
9. Combine All Pages: Merges the markdown outputs from all pages into a single document.
10. Extract All Deposit Table Rows: Analyzes the combined markdown to extract specific data related to deposits, returning structured results.