An e-commerce AI chatbot that turns a browsing question into a checkout

The two questions that decide the sale

An online store lives and dies on two moments: the pre-sale question that decides whether someone checks out, and the "where is my order" that decides whether they buy again. Both arrive around the clock, both are overwhelmingly repetitive, and both are lost the moment they go unanswered — a browsing question at midnight that gets no reply is a cart that never fills, and an order-status ticket that waits two days is a customer who has already messaged you three times.

Almost none of it needs a person. "Does this run small", "when will it ship to me", "can I return it", "where is my parcel" — all answerable from your product, shipping and returns content, or from the order itself. An always-on agent handles that layer on your storefront and on messaging, and routes the genuine exceptions to your team.

What it handles, and what reaches your team

Turning pre-sale questions into checkouts

The most valuable question a store gets is the one asked with a card already in hand. Sizing, materials, compatibility, "will it arrive before the weekend" — a clear answer in the moment is often the whole difference between a checkout and a closed tab. The agent answers from your product pages and cites the source, so the shopper sees where the answer came from and trusts it, and it captures a restock lead when the size they want is gone rather than losing them to an out-of-stock page.

Cutting the "where is my order" pile

Order-status tickets are the single biggest, most mechanical drain on an e-commerce inbox. The agent looks up status from an order number or email and answers the customer directly, so the tickets that used to swamp your team simply resolve themselves. What is left — a genuinely late parcel, a damaged item, a missing package — is routed to a person with the order and the conversation attached, which turns a callback from a discovery exercise into a short, informed one.

One agent across the store and the DMs

Customers ask where they are. Some use the website widget, many message on WhatsApp and Instagram after seeing a product, and the agent handles all of them with the same knowledge and the same order lookups. Everything lands in the unified inbox, so your team works one queue, not four apps.

Where the growth actually comes from

Sales you were losing at night. Traffic does not keep business hours; the pre-sale questions arriving after you close are demand your marketing already paid for. Answering them is the cheapest revenue available.

Support hours back for real work. The same team, freed from order-status tickets, spends its time on the exceptions that actually need judgement. The analytics dashboard reports resolution rate, leads captured and hours saved, so you can see the load the agent absorbed.

The line it does not cross

It does not invent stock or dates. It answers shipping windows and stock from your published content and order data, and where something is uncertain it says so rather than promising a date it cannot keep.

It does not process refunds on its own. Refunds, disputes and damaged-goods claims are handed to a person with the order and the transcript, because money leaving the business is a human decision.

It does not guess about a named order. It reads status from your order data and nothing else, and never speculates about where a parcel "probably" is.

An after-hours order line, answered by phone

Not every customer wants to type. Someone who has been waiting on a parcel calls instead of messaging, and if that call lands at eleven at night it either rings out or goes to a voicemail nobody checks until the office reopens — by which point they have already emailed twice and left a frustrated review. Phone support outside business hours is exactly the kind of coverage a small team cannot staff, and exactly the kind an always-on agent can.

The voice agent answers that call directly, and treats an order lookup as something it can actually do, not just a question to note down: give it an order number or the email the order was placed under, and it reads the status back — dispatched, in transit, delivered — the same way it would answer a chat message, just spoken instead of typed. Return windows, exchange rules and what counts as a valid return all come from your own published policy, so a caller asking "can I send this back" gets the real answer for your store rather than a generic one.

What it does not do is move money. A refund, a dispute over a damaged item, or anything where the business has to decide to pay someone back goes straight to a person, with the order and the call attached, so nothing gets approved on the phone that should have had a human sign-off. The result is a phone line that behaves like your storefront chat already does — answered instantly, honest about what it can and cannot settle — running on the same AI answering service available across the platform. See how the voice channel works.

What to watch in the first month

Watch pre-sale questions answered outside business hours, the share of order-status tickets resolved without a person, and the questions the agent missed — knowledge gaps point straight at the sizing, shipping and returns details your product and policy pages never spelled out.

Order-status questions on WhatsApp

"Where is my order?" is the question that fills a store's inbox. On WhatsApp the agent looks the order up, gives the tracking status, and handles returns and exchange rules from your own policy — the whole WISMO queue answered before a human ever sees it, and escalated cleanly when a case is genuinely stuck. See how the WhatsApp channel works.

Product questions in Instagram DMs

Shoppers ask about sizing, materials and stock in Instagram DMs while they are still deciding. The agent answers from your product content in the moment that intent is highest, links the exact product, and captures the lead if the item is out of stock. See how the Instagram channel works.