Power Automate AI Builder Document OCR + GPT revolutionizes your document workflow. It extracts text from files automatically and uses GPT to generate actionable insights, cutting down manual data capture and analysis. This solution not only speeds up processing but also boosts accuracy and transparency, letting you focus on what matters most.
1. Create new Power Automate flow
Create a new Power Automate flow that triggers when a file is created in a folder and then retrieves the file content.
The main part of the flow is the action Recognize text in an image or a PDF document ,which uses Power Automate’s native OCR model.
2. Create prompt for GPT in Compose Action
My Prompt:
Hi, here are the clear instructions: 1.) I will send you the OCR-extracted output from a received supplier invoice or receipt. 2.) I need you to create the following JSON: { "Document": "insert text from the OCR document here", "Document Number": "insert document number from OCR here", "Supplier": "insert supplier name from OCR here", "Company ID": "insert supplier's company ID from OCR here", "VAT ID": "insert supplier's VAT ID from OCR here", "Supplier City": "insert supplier's city from OCR here", "Supplier Address": "insert supplier's address from OCR here", "Supplier Postal Code": "insert supplier's postal code from OCR here", "Taxable Date": "insert taxable date from OCR here", "Due Date": "insert due date from OCR here", "Total Price Excl. VAT": "insert total price excluding VAT from OCR here", "Total Price Incl. VAT": "insert total price including VAT from OCR here", "Currency": "insert currency from OCR here" } Only populate the JSON with the requested information—nothing more.
3.) Here is the data type specification for each JSON field: Document – string Supplier – string Supplier City – string (maximum 30 characters) Supplier Address – string (maximum 30 characters) Supplier Postal Code – string (maximum 20 characters) Company ID – string VAT ID – string Taxable Date – date (format: yyyy-MM-dd) Due Date – date (format: yyyy-MM-dd) Total Price Excl. VAT – float Total Price Incl. VAT – float Currency – string (values: "EUR", or "USD")
4.) Here is the OCR-extracted invoice data: @{outputs('Recognize_text_in_an_image_or_a_PDF_document')?['body/responsev2/predictionOutput/fullText']}
5.) Provide only json, nothing else
3. OCR + GPT Combo
Now the fun begins with the Create text with GPT using a prompt action.
Use AI Reply in a Prompt and, for Input Text, use the Prompt Compose from step #2.
Voilá!
Automatic OCR and GPT integration in Power Automate transforms your document processing. It automatically extracts text from images and PDFs, eliminating tedious manual work.
By recognizing text with a built-in OCR model and converting it into actionable insights using GPT, you simplify workflows and boost transparency. This approach reduces errors and allows you to focus on strategic tasks.
Embrace this innovative solution to streamline data handling and enhance team collaboration, driving smarter decisions and increased efficiency.
