AI and automation for e-commerce
Most shops spend their support budget answering "where is my order" and lose their conversions to unanswered pre-purchase questions. Both are solved by connecting the assistant to order data rather than by writing better canned replies. The integration is the project; the chat window is the easy part.
Reviewed:
The short answer
An assistant that cannot see the order is a FAQ page with a cursor. The value appears the moment it can look up the shipment, and that is an integration job, not a prompt.
| Bottleneck | What it costs today | What we build |
|---|---|---|
| "Where is my order?" | The largest single share of support volume | Order lookup by email or number, answered from the carrier status |
| Pre-purchase questions | An abandoned basket you already paid to fill | Assistant answering from the product data, with sizing and stock |
| Returns | Manual approval and a label sent by hand | Policy check, label generation and a status the customer can see |
| Stock across channels | Overselling, then an apology | Sync between shop, marketplace and warehouse on every change |
| Order to invoice | Someone re-typing paid orders each morning | Invoice created automatically when payment clears |
| Abandoned baskets | One generic email | Sequence keyed to what was in the basket and whether it is still in stock |
Sources: Baselinker: API documentation · WooCommerce: REST API documentation · reviewed: Jul 17, 2026
Where the money leaks
The order status question
It is the highest-volume ticket in almost every shop and it needs no judgement: match the customer to the order, read the carrier status, answer. Taking that volume out frees the team for the tickets that actually need a person.
The question asked at 22:00
Will this fit, is it in stock, when will it arrive. Unanswered, the basket is abandoned and the ad spend that filled it is gone. Answered from live product data, it converts.
The return that took four emails
Checking the policy, approving, issuing a label and telling the customer where it is are four mechanical steps. Automating them cuts the handling cost and, more usefully, removes the follow-up tickets each step used to generate.
Marketplace stock
Selling the same stock on your own shop and a marketplace without a sync means overselling. This is a plain integration with no AI in it, and it is often the first thing we build because it stops active bleeding.
What we would build first
The shops that get value here are the ones that let the assistant read the order database. The ones that do not end up with a nicer FAQ and no change in ticket volume, because the questions people actually ask are about their order, not about your policies in the abstract. So the build is mostly plumbing: a clean read path to orders, shipments and stock, then a conversational layer on top that is heavily restricted about what it may assert.
That restriction matters more in retail than anywhere else we work. An assistant that guesses at a delivery date creates a promise the warehouse never made, and the customer holds you to it. Ours are built to answer from a retrieved record or say they will check, never to infer. In Poland the integration list is specific and worth naming: Baselinker in the middle, Allegro and WooCommerce on the selling side, and one of the invoicing systems behind. Getting those talking to each other usually removes more manual work than the chatbot does. The messaging side is the same build described in the WhatsApp bot tutorial, and the back-office half is ordinary business automation.