An ecommerce chatbot is software that answers shopper questions on a store and, if it is an AI agent, completes the action behind the question, such as looking up an order or issuing a refund. The practical test of an ecommerce AI chatbot is whether it reduces tickets that would otherwise reach a person, not whether it holds a pleasant conversation. For most stores, the volume sits in a small number of repeated, rule-bound requests.
This guide describes where a chatbot helps an online store, how it connects to store systems, and the limits a merchant should understand before relying on it. It is written for support leaders and store operators, and it is candid about the integration path.
In brief: order status, refunds and changes, and policy questions account for most ecommerce contacts and are well suited to automation. The decisive questions are how the agent reaches your order data and which actions you permit it to take without approval.
What can an ecommerce chatbot resolve?
The most common category of ecommerce contact is the order status question, often abbreviated as WISMO ("where is my order"). Volume rises when a sale begins or a carrier is delayed, and each request requires the same lookup. We cover the measurement and reduction of this category in our explainer on what WISMO is.
Bund AI's ecommerce configuration performs six kinds of work. It tracks an order by number or email and returns status, carrier and estimated arrival. It issues a refund within the merchant's policy, pausing for one-tap human approval where a rule requires it. It changes or cancels an order, whether by editing items, updating the shipping address or cancelling before fulfillment, within the merchant's change window. It answers product and policy questions from real returns, shipping and warranty content. It logs a damaged, lost or wrong-item problem and opens an escalation with full context. And it captures pre-sale interest as a lead or turns a hard case into a ticket with the transcript attached.
The difference between a chatbot that links to a returns policy and an agent that processes the refund is the difference between moving work and removing it. Our article on chatbots versus AI agents develops the distinction, and the self-service actions page describes the actions in more detail.
How does an ecommerce AI chatbot connect to a store?
This is the point on which merchants should be most careful, because vendors often blur it. Bund AI does not have a native Shopify app. Out of the box it answers from the knowledge a merchant provides, such as crawled help center pages, uploaded policy documents or pasted text. To act on live order data, a merchant connects its own systems through custom HTTP actions, which let the agent call order, refund and fulfillment endpoints over HTTP.
In practical terms, a store that exposes an API, whether through its platform's admin API, a middleware layer or its own back end, can define actions for order lookup, refunds or address changes and decide which require approval. Authentication headers are stored encrypted, and requests are guarded against reaching internal network addresses. This path is flexible, but it needs someone with the technical ability to configure the endpoints, and it will take longer than installing an app from a marketplace.
A merchant that wants a one-click installation with prebuilt order actions on a specific platform should confirm that expectation against what is described here. The wider comparison between store-integrated support tools is covered in our article on Gorgias alternatives, which discusses the same question from the help desk side.
Does AI for ecommerce reduce WISMO tickets?
It can, and the mechanism is simple. When a shopper asks where an order is, an agent with access to order data returns the status and carrier information immediately, at any hour, and without a ticket. During a peak, it answers every conversation at once, so a delay does not become an overnight backlog. Only cases that genuinely need a person reach the team, and they arrive with the conversation attached.
The wider context is covered in our guide to AI for ecommerce, and in our piece on how to reduce support tickets more generally. A merchant should resist treating automation as a substitute for proactive communication. Shipping notifications and delay messages sent before the shopper asks remove the question entirely, and the agent then handles what remains.
Early rollouts of Bund AI have shown 86.7% of tickets resolved with no human touch and a 1.8-second median first response. These results come from early rollouts across industries, and a merchant's own result will depend on its catalog, policies and how complete the connected order data is.
Which actions should require approval?
Refunds and cancellations have financial consequences, and the right policy is a management decision. A common approach is to allow low-value refunds inside a clear window to run automatically and to require approval for larger amounts, repeat requests or items outside the policy. Bund AI lets sensitive actions pause for one-tap approval on the web or a reply of YES by email, so that a team member confirms in seconds and not after a ticket has aged.
The same logic applies to order changes. Editing an address before fulfillment is low-risk, whereas cancelling after a pick has started may not be. The merchant's change window defines the boundary, and the agent respects it on every request. Where a request falls outside policy, the better behavior is escalation, which hands off with the full transcript. The escalation and handoff page explains how that works.
What does an ecommerce chatbot cost?
Bund AI is priced by flat monthly plans with no per-resolution fees. Basic is $0.99 per month for 50 AI replies, Starter is $49 per month for 1,000 replies and three seats, Growth is $99 per month for 5,000 replies and 10 seats, and Scale is $199 per month for 10,000 replies and 25 seats. Enterprise is custom, with unlimited replies and seats. An AI reply is each message the agent sends, and a conversation commonly takes two or three. Stores can estimate volume from last month's ticket count, and our AI support cost calculator helps compare options. The pricing page has current detail, and the ecommerce industry page describes the vertical in full.
When Bund AI is not the right fit
A store that wants a prebuilt, platform-native integration with order data, subscriptions and returns portals, installed and configured in minutes, may find a help desk built specifically around its platform a closer fit. Bund AI's connection path depends on custom HTTP actions, which are powerful but require technical setup.
It is also a chat and email agent. It does not handle telephone calls, SMS or WhatsApp, so a merchant whose customers contact it mainly through those channels should plan for them separately. Bund AI accepts image and voice input from shoppers inside chat, which helps with damage photos, but it is not a phone line. Finally, the agent can only act on data it can reach, so a store with fragmented order systems will see more handoffs until its endpoints are connected.
Frequently asked questions
What is an ecommerce chatbot?
An ecommerce chatbot is software on an online store that answers shopper questions about orders, shipping, returns and products. An AI agent version also completes actions, such as looking up an order or issuing a refund within rules the merchant sets.
Is there a native Shopify app for Bund AI?
No. Bund AI does not have a native Shopify app. A store connects its order, refund and fulfillment systems through custom HTTP actions, which require someone able to configure the endpoints, and it otherwise answers from the knowledge the merchant provides.
What is the difference between an ecommerce AI chatbot and a rule-based bot?
A rule-based bot follows scripted flows, while an ecommerce AI chatbot interprets free-form questions and can use connected actions. The practical difference is that an agent can resolve a request, not only direct the shopper to a page.
Can a chatbot reduce WISMO tickets?
Yes, where it can read order and carrier data. It returns status immediately and at any hour, which removes the need for a ticket. Proactive shipping updates reduce the volume further.
Can the chatbot issue refunds automatically?
Yes, within the policy the merchant configures. Refunds can run automatically or pause for one-tap human approval, depending on the rules the merchant sets for amount, window and reason.