System 02
AI order handling
Email to order solves the arrival. This system covers everything after it: is the price still right, is there stock, is this customer allowed to order this, and who hears about it when the answer is no. That is where most of the time goes.
Why orders sit waiting
An order that arrives cleanly is not yet an order that ships. Between the two sits a set of checks almost nobody has written down: does this pricing agreement still hold, has the article been discontinued, is it in stock, is the customer on credit hold, and does the delivery address match what the system holds.
As long as someone does that by hand it works fine until it gets busy. Then the checking slips to the back, orders go out that should not have, and the bill arrives later as returns and credit notes.
How the system fits together
Four steps between an order that has arrived and an order you can send to the warehouse without worrying.
- 01
Merging every channel
Webshop, email, marketplace, EDI and the customer portal land in one stream in one format, so you need one set of rules instead of five.
- 02
Checking before it proceeds
Pricing agreement, stock, credit room, minimum quantity and availability are tested on arrival, not at picking time.
- 03
Exceptions to a person
Whatever fails a check lands on one screen with the reason attached and a proposed fix beside it. No hunting through mailboxes to work out what went wrong.
- 04
Status without handling
The customer hears when something changes: confirmation, delay, partial delivery or cancellation. That removes most of the "where is my order" calls.
How it gets built
- 01
Find where it stalls today
We take a few hundred historic orders and look at which ones sat waiting and why. That list of reasons is the real specification.
- 02
Shadow without intervening
The system first assesses alongside for a few weeks without stopping anything, so you can see how often it was right before you let it decide.
- 03
Let go step by step
First only flagging, then blocking on the most certain rules, then wider. Every step is reversible.
What the system does
Orders that are correct pass through untouched, and everything needing attention surfaces with the reason already attached.
- One order stream for webshop, email, marketplace and EDI
- Checks on price, stock and credit at the moment of arrival
- One work queue of exceptions instead of scattered mailboxes
- Status messages going out to the customer on their own
What it costs
What an engagement costs depends on its scope and on how many systems are involved. I work with a fixed price per phase rather than open-ended billing: you know what a phase costs before it starts, and you decide after each one whether to continue.
I only quote a figure once I know what we are talking about. That takes a half-hour conversation, at no charge.
Whether this fits
This works well when
- orders arrive through several channels
- there are checks that currently live in someone’s head
- things go wrong precisely on the busy days
- your order system can be reached through an API
This fits less well when
- every order comes through one channel and is already checked
- every order is bespoke with negotiation per line
- your volume is too low for the checks to pay off
Questions about order handling
How is this different from email to order?
That system turns an incoming message into an order: it solves the arrival. This one starts where that stops and watches what happens to the order afterwards. They work separately, but together they cover the chain from message to ready to ship.
Will it block orders that were actually fine?
Early on it will, which is why it first runs alongside without intervening. You then see per rule how often it fired correctly. Only once a rule has proven reliable is it allowed to actually block. Rules that raise too many false alarms come out.
Where exactly is the AI?
In interpreting messy input and spotting patterns in exceptions. The checks themselves are ordinary deterministic rules. A language model should not decide whether a credit hold applies; you want that kind of decision predictable.
Let's discuss how I can help your business grow
A short call is enough to tell whether there is a good fit. No sales pitch, just an honest read on what is possible.