Business Central AP Automation: Dynamics 365 Invoice Capture and Vendor Invoice Processing

Business Central AP automation, explained

Business Central AP automation means capturing, coding, approving and posting vendor invoices inside Dynamics 365 Business Central without manual data entry. Microsoft ships part of this as the Payables Agent, which monitors a mailbox, extracts invoice data with Azure Document Intelligence and drafts a purchase invoice for a supervisor to finalize. Microsoft documents firm limits: PDF attachments only, 10 attachments per email, 10 pages and 5 MB per PDF, 100 emails and 500 invoices per day, no approval flows and no anomaly detection. An AI extraction layer in front of Business Central handles the invoices those limits exclude. Last updated July 2026.

What the Business Central Payables Agent does and does not cover

The Payables Agent runs as a background dispatcher against a designated Microsoft 365 mailbox. It picks up unread mail, creates an Inbound E-Document entry for every PDF attachment, sends the PDF to Azure Document Intelligence for OCR, matches the vendor, then drafts line details using AI, prior invoice history, text-to-account mapping and item references. A supervisor reviews the draft and finalizes it into a purchase invoice. It is genuinely useful, and it is scoped narrowly on purpose.

The constraints matter once real volume arrives. Microsoft states the agent does not process PDFs over 10 pages or over 5 MB, skips any email with more than 10 attachments, ignores emails that carry no PDF at all, and stops at 100 emails and 500 invoices per day. When the agent creates a new vendor it sets the Blocked field to All, so nothing posts until a person unblocks it, and Microsoft notes the agent itself provides no vendor approval capability. The feature is validated and supported in English only, and it bills through Copilot Credits on consumption-based billing rather than a flat seat price.

Where InvoiceExtractor fits, and what it does not do

InvoiceExtractor is the capture layer, not an AP suite. It does not run Business Central approval workflows, execute payments, or replace the Payables Agent for teams whose invoices already fit inside its limits. What it does is read any vendor invoice regardless of page count, file size, format or delivery route, split multi-invoice files, capture the complete line-item table and export a CSV ready to load. Teams that also need approval routing and payment execution on top pair it with accounts payable automation covering approval and payment. Because the agent only watches a single Microsoft 365 mailbox, finance teams whose invoices arrive across several inboxes often collect them first with a service that brings every mailbox into one place.

The OCR engine underneath the Payables Agent is Azure Document Intelligence, so if you are costing out a build-versus-buy decision our breakdown of Azure Document Intelligence pricing shows the real per-invoice math, and the Azure Document Intelligence alternative comparison covers accuracy and setup effort. For the wider category, read our accounts payable automation software page, and invoice data capture software explains how templateless AI reading works.

Loading extracted invoices into Business Central

Once invoices are extracted you hold a CSV with one row per line item and identical headers across every vendor. Business Central gives you three sane routes to load it. A configuration package built on the Purchase Header and Purchase Line tables handles a bulk historical load. The Edit in Excel add-in suits smaller recurring batches, since you paste rows straight into the purchase invoice list and publish. Developers post through the purchase invoice API or an AL codeunit for a scheduled, hands-off pipeline. Vendors and G/L accounts must already exist in Business Central for any of the three to match, which is the single most common reason a first import fails. Our step-by-step walkthrough on how to import invoices into Business Central covers all three routes, and importing invoices to an ERP compares the pattern across systems. For teams matching invoices to orders, purchase order data extraction captures the PO side.