New sample: Invoice Data Extractor

prompts/ai-builder/invoice-data-extractor/README.md

@@ -0,0 +1,43 @@
+# Invoice Data Extractor
+
+## Description
+
+This prompt uses AI Builder's **image input** capability to extract structured data from invoice images or scanned PDFs. It returns a complete JSON object with all billing fields — vendor, line items, totals, tax, currency, and payment terms — ready for direct consumption by Power Automate without manual data entry.
+
+This is the first prompt in the repository to use an **image-type input** (`invoice_image`) rather than text. It demonstrates how AI Builder's multimodal capabilities can replace dedicated OCR or document processing pipelines for common finance automation scenarios.
+
+A null value is returned for any field not found on the invoice, making the output safe to use in downstream flow conditions without causing parse errors.
+
+## Prompt
+
+> You are an invoice processing assistant. Analyze the [invoice_image] provided and extract all relevant billing information. Return the result strictly as valid JSON with fields: invoice_number, invoice_date, due_date, vendor_name, vendor_address, bill_to, line_items (array), subtotal, tax_amount, tax_rate, total_due, currency, payment_terms. Set missing fields to null.
+
+### Supported Language(s)
+
+- [English - US](./en-us/prompt.md)
+
+## Authors
+
+Solution|Author(s)
+--------|---------
+Invoice Data Extractor | [OwnOptic](https://github.com/OwnOptic) ([@OwnOptic](https://twitter.com/OwnOptic))
+
+## Minimal Path to Awesome
+
+1. Navigate to [AI Builder](https://make.powerapps.com/) and select **AI Builder** > **Explore**
+2. Choose **Prompt** and click **Build your own prompt**
+3. Copy the content from [prompt.md](./en-us/prompt.md) and paste it into the prompt editor
+4. Add an input named `invoice_image` of type **Image**
+5. Test by uploading a sample invoice image (PNG or JPEG)
+6. Save and publish the prompt
+7. Use it in a Power Automate flow:
+   - Trigger: **When an email with attachment arrives** (Outlook) or **When a file is created** (SharePoint)
+   - Action: Call this AI Builder prompt, passing the attachment as the image input
+   - Action: **Parse JSON** with the schema above
+   - Action: Create a record in Dataverse or post to an AP system
+
+## Disclaimer
+
+**THIS CODE IS PROVIDED *AS IS* WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING ANY IMPLIED WARRANTIES OF FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR NON-INFRINGEMENT.**
+
+<img src="https://m365-visitor-stats.azurewebsites.net/powerplatform-prompts/samples/ai-builder/invoice-data-extractor" aria-hidden="true" />

prompts/ai-builder/invoice-data-extractor/assets/sample.json

@@ -0,0 +1,53 @@
+[
+  {
+    "$schema": "https://developer.microsoft.com/en-us/json-schemas/pnp/samples/v1.0/metadata-schema.json",
+    "name": "pnp-powerplatform-prompts-invoice-data-extractor",
+    "version": "1.0.0.0",
+    "source": "pnp",
+    "creationDateTime": "2026-03-02T00:00:00Z",
+    "updateDateTime": "2026-03-02T00:00:00Z",
+    "title": "Invoice Data Extractor",
+    "shortDescription": "Extracts structured billing data from invoice images using AI Builder's image input — vendor, line items, totals, tax, currency — returned as strict JSON.",
+    "longDescription": [
+      "First repository prompt to use an image-type input. Analyzes invoice images or scanned PDFs and returns a complete JSON object with all billing fields (invoice number, dates, vendor, bill-to, line items array, subtotal, tax, total, currency, payment terms). Null values for missing fields ensure safe downstream processing in Power Automate. Replaces dedicated OCR pipelines for AP automation."
+    ],
+    "url": "https://github.com/pnp/powerplatform-prompts/tree/main/prompts/ai-builder/invoice-data-extractor",
+    "products": [
+      "AI Builder",
+      "Power Platform",
+      "Power Automate"
+    ],
+    "tags": [
+      "INVOICE",
+      "OCR",
+      "IMAGE INPUT",
+      "JSON",
+      "FINANCE",
+      "ACCOUNTS PAYABLE"
+    ],
+    "categories": [
+      "AI Builder"
+    ],
+    "metadata": [
+      {
+        "key": "prompt",
+        "value": "Analyze the [invoice_image] and extract all billing fields as strict JSON. Set missing fields to null."
+      }
+    ],
+    "thumbnails": [
+      {
+        "type": "image",
+        "order": 100,
+        "url": "https://raw.githubusercontent.com/pnp/powerplatform-prompts/main/templates/previewprompts.png",
+        "alt": "Preview"
+      }
+    ],
+    "authors": [
+      {
+        "gitHubAccount": "OwnOptic",
+        "name": "OwnOptic",
+        "pictureUrl": "https://github.com/OwnOptic.png"
+      }
+    ]
+  }
+]

prompts/ai-builder/invoice-data-extractor/en-us/prompt.md

@@ -0,0 +1,41 @@
+You are an invoice processing assistant. Analyze the [invoice_image] provided and extract all relevant billing information.
+
+Extract the following fields:
+- Invoice number
+- Invoice date (format: YYYY-MM-DD)
+- Due date (format: YYYY-MM-DD, or "Not specified" if absent)
+- Vendor name
+- Vendor address
+- Bill-to company name
+- Line items: for each line, extract description, quantity, unit price, and line total
+- Subtotal
+- Tax amount and tax rate (if shown)
+- Total amount due
+- Currency (ISO 4217 code, e.g. USD, EUR, GBP)
+- Payment terms (e.g. "Net 30", or "Not specified" if absent)
+
+Return the result strictly as valid JSON using this schema:
+{
+  "invoice_number": "string",
+  "invoice_date": "string",
+  "due_date": "string",
+  "vendor_name": "string",
+  "vendor_address": "string",
+  "bill_to": "string",
+  "line_items": [
+    {
+      "description": "string",
+      "quantity": number,
+      "unit_price": number,
+      "line_total": number
+    }
+  ],
+  "subtotal": number,
+  "tax_amount": number,
+  "tax_rate": "string",
+  "total_due": number,
+  "currency": "string",
+  "payment_terms": "string"
+}
+
+If a field cannot be found on the invoice, set its value to null. Do not include any text outside the JSON object.