From email to draft sales order: an agent in Microsoft Business Central

Client:
Manufacturing company
Service:
Integration
Status:
in progress August 2026
Overgrown olive trunk on a travertine plinth, smothered in water sprouts, dry twigs, bast and loose cables whose plugs dangle into empty space

Starting point

Starting point

Orders and inquiries arrive by email, often as PDFs. The inside sales team enters the line items into Business Central by hand. The challenges:

  • Formats: every customer orders in their own format.
  • Item numbers: the customer's numbers don't match the company's own.
  • Transcription errors: only surface at order confirmation.
  • Quotes: are built from copies of old quotes.
Open frosted-glass envelope set in a travertine block, out of which grow moss, lichen and a young olive branch; a braided cable with moss at its clamp leads to a Calacatta block with a gear plaque, on which stands a frosted-glass panel with three order lines, and in front of it lies a turned champagne-metal flange in a ring of moss

Our solution

Solution

  • Read: An agent reads incoming orders and inquiries in the inbox and identifies the customer, line items, quantities and dates.
  • Match: It matches everything against the master data in Business Central.
  • Draft: It creates a draft sales order or quote and flags unclear line items.
  • Approve: The inside sales team reviews, corrects and approves. Nothing is posted until it has been approved, and every step is logged.

Flow: email agent reads matching against master data draft in Business Central inside sales approves posting.

The same trunk in full leaf: a dense olive crown with green olives and one silicon leaf; the cable bundle and plugs are overgrown with moss and lichen

Status

Outcome

As of August 2026: The process runs end to end for the first time, from email to draft document, on three sample cases and still against test systems rather than the live ERP.

What we measure: line items recognized automatically, time per order, corrections after approval.

More case studies