Calling VerseDocs from Power Automate

The one step every flow shares

1. 1Add an action and search for the Microsoft Dataverse connector.
2. 2Choose Perform an unbound action.
3. 3Set Action Name to the action you want — vdocs_TemplateDocx, vdocs_PdfMerge, and so on. They appear in the dropdown once the solution is imported.
4. 4Fill in the parameter fields the designer generates for that action.
5. 5The outputs — OutFileContent, Result, Meta — become dynamic content for later steps.

Parameter conventions

Every action reuses the same handful of parameter names, so once you know the pattern you can use any action without looking it up.

Worked example: invoice PDF, emailed

The flagship pattern. Render a Word template straight to PDF and send it.

1. 1Get the template file. From SharePoint, OneDrive, or a vdocs_template row.
2. 2Compose the data as JSON.
3. 3Perform an unbound action → vdocs_TemplateDocx, with FileContent = the template, Data = your Compose, and Options asking for PDF.
4. 4Send an email, attaching base64ToBinary(...OutFileContent) with the name from OutFileName.

{
  "customer": { "name": "Contoso Ltd", "city": "Seattle" },
  "invoiceNumber": "INV-1042",
  "total": 1250.00,
  "lines": [
    { "item": "Consulting", "qty": 10, "price": 100.00 },
    { "item": "Support",    "qty": 1,  "price": 250.00 }
  ]
}

{
  "outputFormat": "pdf",
  "culture": "en-GB",
  "onMissing": "error"
}

{
  "outputFormat": "pdf",
  "pdfConformance": "PDF/A-2b"
}

Worked example: straight from a Dataverse row

When the data already lives in Dataverse, do not assemble JSON by hand. Give vdocs_TemplateFromRow a template and a row id, and it resolves the template’s binding against that record — related records included.

• Data names the template and the target row.
• Everything in Template Syntax §7 applies: formatted values by default, _raw when you want the underlying one, related tables as arrays.
• Use vdocs_GenerateAndAttach instead if you want the result attached to the record’s timeline in the same call — it is what the Generate document button uses.

Worked example: many rows at once

vdocs_TemplateBatch generates for many records in one call. Name the rows explicitly with RowIds, or let it select them with FetchXML in Options. When both are supplied, RowIds wins.

Worked example: a barcode in a document

1. 1vdocs_BarcodeGenerate with Data carrying the payload — an order number, a serial, a URL — and Options naming the symbology.
2. 2Pass the returned image into vdocs_TemplateDocx as a named image in Options.images.
3. 3Place {{image:barcode width=40mm}} in the template where it should appear.

{ "type": "qr", "format": "png", "margin": 10 }

Digitally signing a PDF

vdocs_PdfSign applies a cryptographic signature to a PDF using a certificate you supply. The signature proves two things to anyone who opens the document: that it has not changed since it was signed, and who signed it.

{
  "pfx": "<base64 of a .pfx/.p12 file, including its private key>",
  "password": "..."
}

{
  "reason": "Approved by finance",
  "location": "London, UK",
  "contactInfo": "finance@contoso.example",
  "fieldName": "FinanceApproval"
}

• Sign more than once by calling it again with a different fieldName. Reusing a name replaces that signature instead of adding one.
• Signatures are invisible — they carry full cryptographic weight and every reader shows them in its signature panel, but nothing is drawn on the page. Placing a visible signature graphic would require a drawing library that is not available in the Dataverse plug-in sandbox.
• Sign last. Anything that modifies the PDF afterwards — a watermark, a merge, adding pages — invalidates the signature, which is exactly what a signature is for.

Handling errors

Every action fails as a standard Dataverse action error: the step turns red and the error body is JSON.

{ "code": "TEMPLATE_FIELD_MISSING", "message": "Template field 'customer.vat' was not found in Data." }

Add Configure run after → has failed on a following step to branch, and parse the body to read code. Branch on the code, never on the message text — codes are stable, wording is not.

The full list is on Limits & Errors.

What VerseDocs does not do

Worth knowing before you design a flow around an assumption. These are deliberate boundaries, not gaps waiting to be filled:

• No OCR. Text is extracted from PDFs that contain real, selectable text. A scanned image is not made searchable. Recognise the text upstream with AI Builder, then hand VerseDocs the result.
• No AI extraction or classification. VerseDocs reads structure it already understands — PDF text layout, form fields, template tokens. Pulling unstructured fields out of a free-form document is AI Builder’s job, upstream.
• No translation. Translate the data before it enters a template, not the rendered document afterwards.
• No CAD or InfoPath conversion. The conversion ladder handles Word, Excel and PowerPoint.