What was the problem?
Orders came in as texts and phone calls, written the way people talk: “Send the usual cups and those brown bags.”
Someone on staff had to work out which products that meant across the whole catalog, look up that customer’s pricing, check stock, and key the order in. It took minutes per order, mistakes crept in on busy days, and anything that arrived after five sat until the next morning.
What did Airu Labs build?
An ordering assistant inside the chat customers already use. Every message goes through the same five steps:
- The customer writes the order the way they would say it out loud.
- The assistant matches every item against what that customer has bought before.
- It checks live stock and the customer’s agreed pricing.
- It builds the cart and asks the customer to confirm.
- Once they confirm, it places the order.
When it can’t tell what someone means, it asks. When it is still stuck, it emails a person on the team with the exact question.
How does it know what “the usual cups” means?
It reads the order against that customer’s own history instead of the whole catalog. “The usual” means what this customer usually orders, and “those brown bags” means the brown bags they bought last time, at the price they agreed to. A new hire would have to learn that customer by customer. The assistant looks it up on every order.
What changed for the business?
| Before | After |
|---|---|
| Orders arrive as texts and calls, and someone retypes each one | The assistant reads the message and builds the order |
| Staff look up pricing and stock for every order | Pricing and stock are checked on every order before the cart is built |
| Orders that arrive after five wait until morning | Customers order whenever they think of it |
| 20 hours a week of order entry | A person only hears about the orders the assistant could not work out |
An order that used to take a phone call now places itself, with the right products at the right price. That is 20 hours a week the staff got back, about half of a full-time job, and it went into work that needs a person.
Could an ordering assistant work for my business?
It fits if three things are true: your customers already send orders by text, email or chat; someone on your team retypes those orders into another system; and each customer’s pricing and history live somewhere the assistant can read.
The first call covers where orders arrive today, where they have to end up (your ERP, your accounting software, your order system), and which mistakes cost you the most. Airu Labs builds around that one process first.
The person who built it explains the system in a short video: One AI system now saves this business 20 hours a week.
See the purchasing system built for a paper and packaging distributor, or the ten questions to ask before you hire an AI developer.