Inquiry automation: from messy message to structured record

An inquiry arrives as an email with three photos, a WhatsApp message that says “need 50 of these, how much”, or a web form with a PDF drawing attached and no material specified. Someone has to read it, work out what’s being asked, and write it down somewhere useful. That’s the part we automate.

What it reads

Inbound email, web form submissions, WhatsApp messages, attached PDFs and drawings, and plain free text. It works on the messy version — the one where the customer wrote in a hurry and half the details are missing. It reads English by default; other languages can be added per shop.

What it extracts

From each inquiry it pulls out a structured record: contact and company, part or item description, quantity, material or spec, requested deadline, delivery location, and any attached files. Where the customer wrote something ambiguous, it keeps their original wording next to the field, so you can see what they actually said rather than a cleaned-up version of it.

How missing fields get flagged

If the quantity, material, tolerance, or deadline isn’t there, the record is marked incomplete and the specific missing field is listed. Nothing is guessed and filled in silently. An incomplete record is still useful — it tells you exactly what to ask the customer, in one message, instead of a back-and-forth.

What the human still does

A person reads the record, corrects anything the automation got wrong, fills the gaps once the customer replies, and decides what happens next. The automation prepares; your team decides.

What the output is not

This is a record to confirm, not an auto-reply. It does not email your customer, promise a price, or accept an order. The first thing that goes back to the customer is still written or approved by a person.