AI ordering
AI ordering and order entry
AI ordering means customer orders reaching your system without anyone typing them. Orders arrive by email, PDF, spreadsheet, portal export and photo, and every one of them is keyed by hand today. A model reads them, matches lines to the customer and your catalogue, and creates the order. Anything it is not sure about goes to a person.
What we build on
We build it beside the inventory, warehouse or accounting system you already run. Unleashed, Cin7, CartonCloud, NetSuite or MYOB stays the system of record. The order entry team becomes a customer service team.
Software we usually meet
- Unleashed
- Cin7
- DEAR
- CartonCloud
- NetSuite
- MYOB Advanced
- Xero
What AI ordering does
Orders read in any format
Each customer sends orders their own way. Intake is designed per customer format and tested against your own order history before it goes live.
Lines matched from customer history
A handwritten line or a customer part number is matched to your product using what that customer bought before. That is far more reliable than the catalogue alone.
Created in your system
Orders land in the WMS, inventory or accounting system as drafts or live orders, depending on confidence. The rep or operations team confirms where needed.
One exception queue
Changed product codes, mismatched addresses and odd quantities surface in one place with the reason attached instead of in an inbox.
Software or a custom build
EDI and customer portals handle ordering well for the customers who will use them. The problem is the long tail who will not. A build handles that tail in whatever format it arrives, and can run alongside EDI for the customers who already have it.
How it works in practice
Every step depends on the one before it, and skipping the first two is how these projects stall.
Collect six months of orders per customer
The real files, in the formats they arrive in. This is the test set and it decides the design.
Design intake per format
Each customer's format is handled on its own terms, and the informal rules your team applies by hand are written down and encoded.
Test against history and set the threshold
The auto-create threshold goes where the error rate matches what your team already produces manually.
Go live with one review queue
Orders above the threshold are created. The rest surface in one place with the reason attached, in priority order.
Add customers in batches
The biggest formats first, then the tail, with EDI customers feeding the same queue so operations watch one screen.
What it has looked like
Not for you if
- Most of your customers already order through EDI or a portal and the tail is tiny.
- Under a few hundred orders a month. The hours are real but the build will not pay back quickly.
- Order lines that cannot be matched to a catalogue or history, such as fully bespoke work.
For a meat distributor, an agent on their product database reads text and WhatsApp orders, prices each line and drafts the sales order and the quote. Customers order the way they always did.
In an illustrative 3PL engagement with forty customers, 91 percent of orders reached the WMS without a person typing them. Three full time roles moved from order entry to customer service. Pick errors traced to order entry fell by about seventy percent.
For a building products distributor, also illustrative, reading orders from email, PDF and photos returned about four hours a day to each sales rep.
The supplier side of the same problem, purchase orders and invoice matching, is covered under AI in procurement. How long a first build takes is in our timeline guide. We do most of this work with logistics and wholesale businesses.
Case studies
How it works
- 01
Intro call
A short call to talk through your objectives, goals and timeframes, and agree what success looks like. No problem is too big or too small, whether that is construction estimation, materials procurement, job quoting or AI SEO.
- 02
Discovery
We sit with your team for an extended stretch to understand your business, the systems you run and how the work gets done day to day.
- 03
Roadmap
We put together a roadmap of action items, in order, that takes you from where you are today to what we agreed success looks like.
- 04
Execution
We do the work, check it against the goals we set on the first call and do not call it finished until you are satisfied.
Questions
Our customers all order differently. Can that be automated?
That is the normal case. Intake is designed per customer format and tested against your own history. The confidence threshold sits where the error rate matches what your team produces by hand.
What happens to orders the AI is not sure about?
They go to a single review queue with the reason highlighted. Typically around one order in ten needs a person, and that is the order worth a person looking at.
Can it work alongside EDI?
Yes. EDI keeps handling the customers who use it. The build handles everyone else, and both can feed the same exception queue so operations watch one place.
How long does it take to set up?
A first customer set is usually live within a few weeks of discovery, with more customer formats added as they are tested. We scope and quote each stage before it starts.
Industries where this comes up
Other things we build
Next step
We map where your hours go and price the work that is worth automating.