Offer 04Order desk

Order desk automation

Each emailed purchase order is read, checked against your items and prices, and drafted as an ERP order for approval.

A pallet of boxed fruit standing in a blue stretch-wrapping machine at the end of a packing line

Who it’s for

This offer is for the owner of a $5M to $50M wholesale distributor or manufacturer in Ontario or Quebec whose customers send purchase orders by email. It is also for the person who runs the order desk.

Orders arrive as PDF attachments, spreadsheets, scans and lines typed into an email, each on the customer’s own form. Someone reads each one, looks up the customer’s part numbers, checks the price and keys the order into the ERP. A wrong price or an old item code shows up later, at invoicing or on the customer’s dock.

The offer fits when:

  • One or two people spend much of their day retyping purchase orders.
  • Customers use their own part numbers, and only one person knows how they map to yours.
  • Price mistakes are found at invoicing, after the order has shipped.

What you get

01One inbox for ordersA mailbox or folder that collects customer purchase orders in any format: PDF, spreadsheet, scan or email text.
02Each order read into fieldsPO number, customer, ship-to address and terms, and each line’s part number, quantity, price and due date.
03Each line matched and checkedMatched to your item list, including the customer’s own part numbers, and checked against that customer’s price.
04Flags for a personA price that differs from the price file, an unknown item, an unusual quantity or a PO number already received.
05A draft order for approvalIn the layout your ERP takes, shown beside the original PO. A person approves, changes or rejects it.
06Entry only after approvalApproved orders go into the ERP through its import, or a person keys them from the draft. Nothing is entered or sent to the customer before that approval.
07A logEvery PO, draft, approval and edit, with the person who approved each order.

How it runs

Four weeks from the first call to daily use.

Week 1A test set of real orders with known answersWe collect recent purchase orders and the orders your team entered from them, plus the item list, customer part numbers and price files. We map the fields your ERP needs and set up the order inbox.
Week 2Drafts checked against orders already in your ERPWe build the reading and matching, then run them on the test set until the drafts match what your team entered.
Week 3A record of how the drafts compare on live ordersThe system drafts each new order while your team keeps entering orders as usual. We compare every draft with the order that was entered and fix what differs.
Week 4Drafted orders in daily use, with a person approving each oneYour order desk works from approved drafts, entered through the ERP import or keyed from the draft. We hand over and decide with you whether stage two is worth doing.

The first stage takes four weeks, at a fixed price agreed before work starts. It runs on your own workspace, and if you stop there, everything built in it is yours to keep.

What we need from you

AccessA mailbox or forwarding address for customer orders, and read access to the item list, customer part numbers and price files.
DataRecent purchase orders from your main customers, with the orders entered from them, so drafts can be tested against known answers.
Your ERP’s importThe import format it accepts, or a test company where drafts can be tried, and an hour with whoever runs the ERP.
TimeThe person who leads the order desk, about two hours in week one, then a short daily review of drafts in weeks three and four.

What changes

Your order desk checks drafts instead of retyping orders. Price differences and unknown items are caught before the order is entered, rather than at invoicing.

The customer part numbers that lived in one person’s head are written into a cross-reference anyone can use. An order waiting in the inbox is visible, so a busy week or a vacation does not stall the desk. The people who keyed orders have more time for customers.

After week 4

The order desk keeps drafting orders on your workspace after week 4.

Your workspace
One database with your data and your definitionsDesigned to run on its own server in Canada. A chat answers from it and shows the query.
Your systems
ERPAccountingEmailSpreadsheetsFiles and drawings

Each offer adds one module. The modules share one database and one set of definitions, so each new one starts from data that is already clean. If you stop, what was built is yours.

What can come next

Your orders, items and customer part numbers now sit in your workspace, ready for reporting and for margin by customer. The options below build on the same data.

  • Approved orders written straight into the ERP through its API, where it has one, as AI for ERP describes.
  • For an older ERP with no API or import, a computer use agent that keys approved orders through its screens, as legacy ERP automation explains.
  • Order acknowledgements drafted for a person to send.
  • More documents, such as supplier order confirmations and supplier invoices.

Questions people ask

What does it cost?
The first stage is a fixed price agreed before work starts, based on the job. ThriveAI does not publish rates.
Does anything go into the ERP without a person approving it?
No. Every draft waits for a person to approve, change or reject it, and nothing is sent to the customer on its own. The log records who approved each order. Human in the loop explains how that approval step works.
Our ERP is old and has no API. Does that matter?
No. In the first stage, approved orders go in through the ERP’s import or are keyed from the draft. Where an older ERP has no import at all, a computer use agent can key approved orders in stage two.
What formats can it read?
PDF attachments, spreadsheets, scans and orders typed into the body of an email. Each customer’s format is tested in the first stage on orders your team already entered.
What about customers who use their own part numbers?
Each customer’s part numbers are matched to yours through a cross-reference. A part number with no match goes to a person, and the answer is added to the cross-reference.
Where does our data go?
The platform is designed to keep each client’s data on its own server in Canada. You choose a model on that server or a hosted model under a written zero data retention agreement. A hosted model may process requests outside Canada. For the options, see private AI for business.

Contact

Take the retyping off your order desk

Tell Derik how customer orders reach you, how many arrive in a week and which ERP they go into. He will tell you what a four-week order desk build would cover.

Prefer to talk? Book a meeting.

Your message goes to Derik Lawlis, the founder.