AI-Receipt Parser in Power Automate

AI-Receipt Parser is a Power Automate action by PDF4me that reads a receipt PDF or image and returns merchant details, line items, financial totals, and payment method as structured data. Use it to automate expense capture, feed Excel or Power BI expense tracking, or post transactions straight into Dynamics 365 without manual keying.

What this action does

PDF4me AI-Receipt Parser reads a receipt (PDF, PNG, JPG, or JPEG) and returns merchant information, line items, financial totals, and payment method as structured JSON. It also recognizes specialized fields for hotel, fuel and energy, healthcare, and training receipts, and can extract additional custom fields beyond the standard schema.

How Do I Authenticate My Power Automate Flow?

To call PDF4me through Power Automate, the flow needs a valid PDF4me API key on the PDF4me Connect connector. Authentication ensures secure communication and validates your identity as an authorized user.

Get your API Key

Important Facts You Should Not Miss

The real field names are File Content and Body/Document/Name

The connector UI labels the two main fields File Content and Body/Document/Name, not "Receipt File Content" or "Receipt Name". Map the receipt bytes into File Content.

All 5 optional fields live under Advanced parameters

Body/receiptType, Body/analyzeItems, Body/extractMerchantInfo, Body/calculateTotals, and Body/customFieldKeys are all under Advanced parameters, confirmed in the screenshot as "Showing 5 of 5".

Check warnings and fallbackUsed before trusting a low-confidence read

A blurry photo or an unusual layout can still return success: true with a populated warnings array or fallbackUsed: true. Branch these to manual review instead of auto-posting totals.

Parameters

Required in the action: File Content and Body/Document/Name. All other fields are Optional Advanced parameters with working defaults.

Parameter	Required	What it does	Example
File Content	Required	Binary content of the receipt file, mapped from a previous action (Get file content, SharePoint, OneDrive, email attachment). Supports .pdf, .png, .jpg, .jpeg.	`body('Get_file_content')`
Body/Document/Name	Required	Filename of the receipt including its extension. Used for format detection and processing.	`receipt_2026_07_07.pdf`
Body/receiptType (Advanced)	Optional	Improves parsing accuracy for a known category: meal, supplies, hotel, fuel and energy, transportation, communication, subscriptions, entertainment, training, healthcare, or generic.	`hotel`
Body/analyzeItems (Advanced)	Optional	Yes extracts individual line items with name, quantity, price, category, and item code. No skips line-item detail.	`Yes`
Body/extractMerchantInfo (Advanced)	Optional	Yes extracts merchant name, address, phone, website, store number, and cashier name.	`Yes`
Body/calculateTotals (Advanced)	Optional	Yes extracts and calculates subtotal, tax amount, total amount, and payment method.	`Yes`
Body/customFieldKeys (Advanced)	Optional	Array of additional field names to extract beyond the standard schema. Returned inside additionalInfo.	`['loyaltyNumber']`

How Do I Set Up AI-Receipt Parser in Power Automate?

Add AI-Receipt Parser to your flow as a new step.
Sign in to PDF4me Connect with your API key.
Map File Content from a previous action (Get file content from SharePoint, OneDrive, Dropbox, or an email attachment).
Set Body/Document/Name with the matching extension (.pdf, .png, .jpg, or .jpeg).
Under Advanced parameters, set Body/receiptType if you know the category, and toggle Body/analyzeItems, Body/extractMerchantInfo, Body/calculateTotals as needed.
Run the flow. Check success before using the extracted fields downstream.
Route merchantName, analyzeItems, totalAmount, and related fields to Excel, SharePoint, Dynamics 365, or Power BI.

Output

The response includes core transaction fields, nested arrays for items/merchant/financials, specialized fields per receipt type, and processing metadata.

Field	Type	Description
receiptType	String	Type of receipt extracted (supplies, retail, restaurant, hotel, fuel, healthcare, training).
transactionDateStr	String	Transaction date in ISO 8601 (YYYY-MM-DD).
transactionTime	String	Transaction time (HH:MM:SS).
receiptNumber	String	Receipt number or transaction identifier.
analyzeItems	Array	Line items with name, quantity, unit price, total price, category, description, item code, brand, size, color, SKU.
merchantInfoList	Array	Merchant objects with name, address, phone, website, cashier name, store number.
financialInfoList	Array	Financial objects with subtotal, tax amount, total amount, payment method, card last four.
merchantName / merchantAddress / merchantPhone / merchantWebsite	String	Flattened merchant details for quick access without indexing into merchantInfoList.
subtotal / taxAmount / totalAmount	Number	Flattened financial totals with precise decimal handling.
paymentMethod / cardLastFour	String	Payment method used and last four digits of the card, if applicable.
roomNumber / checkInDate / checkOutDate	String	Hotel-specific fields, populated when receiptType is hotel.
fuelType / fuelQuantity / fuelUnit / vehicleNumber	String/Number	Fuel-specific fields, populated when receiptType is fuel and energy.
trainingProvider / courseName	String	Training-specific fields, populated when receiptType is training.
healthcareProvider / patientName / insuranceInfo	String	Healthcare-specific fields, populated when receiptType is healthcare.
additionalInfo	Object	Dictionary containing any Body/customFieldKeys values that were found.
warnings	Array	Warning messages for potential data-quality issues encountered during extraction.
fallbackUsed	Boolean	True when a fallback extraction method was used, a signal to review manually.
success	Boolean	True if extraction succeeded.
message	String	Descriptive result message.

Typical Setups

Expense capture to Excel

Trigger when a receipt is emailed or uploaded to SharePoint.
Run AI-Receipt Parser with extractMerchantInfo and calculateTotals enabled.
Append merchantName, totalAmount, and transactionDateStr as a new Excel row.
Archive the original receipt with the extracted data linked.

Line-item expense categorization

Trigger on receipt upload with analyzeItems enabled.
Apply to each item in analyzeItems and categorize by the returned category field.
Route items over a policy threshold to a manager approval.
Post approved items to Dynamics 365 Finance.

Specialized receipt routing (hotel, fuel, healthcare)

Trigger on receipt upload; set Body/receiptType if the category is known.
Check the returned receiptType and branch on hotel, fuel and energy, or healthcare.
Send roomNumber/checkInDate, fuelType/vehicleNumber, or patientName to the matching system.
Log any warnings for manual review.

Practical Tips

Always branch on success and warnings before posting totals

A populated warnings array or fallbackUsed: true means the extraction had lower confidence. Route these to manual review instead of auto-posting into an expense or accounting system.

Set Body/receiptType when you know the category

Leaving it unset still works via generic extraction, but a matching receiptType improves accuracy on category-specific fields like roomNumber or fuelType.

Turn off extraction flags you do not need

If you only need the total amount, set analyzeItems and extractMerchantInfo to No to keep the response smaller and processing focused.

Match Body/Document/Name's extension to the real file format

A mismatched extension (for example naming a JPG file receipt.pdf) can affect format detection and extraction accuracy.

Use customFieldKeys for anything the standard schema misses

A loyalty number, a promo code, or an internal reference printed on the receipt can be captured via Body/customFieldKeys without waiting for a schema update.

Cheat Sheet

Goal	receiptType	Key output fields
General expense capture	`generic`	merchantName, totalAmount, transactionDateStr
Line-item expense report	`supplies or retail`	analyzeItems (name, category, price)
Travel expense reconciliation	`hotel`	roomNumber, checkInDate, checkOutDate, totalAmount
Fleet fuel tracking	`fuel and energy`	fuelType, fuelQuantity, vehicleNumber
Healthcare expense/insurance	`healthcare`	healthcareProvider, patientName, insuranceInfo

Common Questions

What receipt types does Body/receiptType support?+

meal, supplies, hotel, fuel and energy, transportation, communication, subscriptions, entertainment, training, healthcare, and generic. Setting the correct type is optional but improves field-recognition accuracy for that category's specialized fields, for example roomNumber for hotel or fuelType for fuel and energy. For general recordkeeping guidance on what a compliant receipt should capture, see the IRS recordkeeping requirements for businesses.

What is the difference between analyzeItems, extractMerchantInfo, and calculateTotals?+

These three Advanced parameters independently control extraction depth. analyzeItems extracts individual line items (name, quantity, price, category). extractMerchantInfo extracts merchant name, address, phone, and cashier. calculateTotals extracts subtotal, tax amount, and total amount. Disable any you do not need to reduce processing and keep the response focused.

Can I extract custom fields the standard schema does not cover?+

Yes. Body/customFieldKeys accepts an array of field names beyond the standard schema. The AI engine attempts to locate and return each one inside the additionalInfo object of the response.

How do I handle a receipt the parser could not read confidently?+

Check the warnings array and the fallbackUsed flag in the response. A non-empty warnings array or fallbackUsed: true signals lower-confidence extraction, for example from a blurry photo. Route these cases to a manual review step instead of auto-posting the extracted totals, consistent with general CFPB guidance on tracking expenses accurately.

AI-Invoice Parser

Extract structured data from vendor invoices instead of receipts.

AI-Bank Statement Parser

Extract transaction data from bank statements for reconciliation.

AI-Pay Stub Payslip Parser

Extract payroll data from pay stubs for expense and HR workflows.