A Website Chatbot with Access to a YML/XML Feed — Why an Online Store Needs It

In our article on AI agents for business, we looked at an example: a customer asks about product availability and shipping costs, and a standard chatbot, at best, sends them to search for themselves. For an online store, this isn't an abstract example—it's the most common chat scenario.

Why a chatbot without a feed is just a polite placeholder

Most chatbots on store websites are trained on static catalog descriptions: categories, general specifications, delivery terms. As soon as a customer asks about a specific SKU—"is this TV available in blue?"—the bot either defaults to generic phrases or honestly admits it doesn't know and suggests writing to a manager. For a customer who is ready to buy, this is an unnecessary step where a portion of the audience simply leaves for a competitor with a faster response.

How YML/XML feed integration works

YML and XML are standard product feed formats that online stores already use for Yandex.Market, Google Merchant, and ad campaigns. An AI agent connects to the same feed and gets direct access to up-to-date data:

• Product availability — by a specific SKU, not a broad category.
• Current price, including discounts and promotions, without the risk of quoting an outdated figure.
• Specifications and modifications — color, size, configuration — from the same source as the website.

The feed updates automatically on a schedule, so the agent cannot "lag behind" the catalog—as soon as the stock or price changes in the accounting system, the agent answers with the new data.

What this gives businesses in practice

In our implementation experience, this setup immediately resolves several bottlenecks:

1. Fewer manager requests for standard questions. Questions like "is it in stock" and "how much does it cost" are the most common in chats, and the agent closes them without human involvement.
2. The customer gets an instant answer, not 20 minutes later, which is especially critical in the evenings and on weekends when managers aren't around.
3. Fewer communication errors. A manager might accidentally quote the wrong price or forget about a promotion—the agent always replies based on the current feed.

When this is especially justified

Feed integration provides the most impact for catalogs of several hundred items or more, where manually keeping track of stock and prices is physically impossible, as well as for highly fluctuating prices—seasonal promotions, new shipments, or exchange rate fluctuations for imported goods.

If you have a catalog of 20-30 items that rarely change, a feed makes less sense—it's easier to keep the current information directly in the agent's knowledge base and manually update it once a month.

The Bottom Line

A chatbot that answers using templates is an interface. An AI agent with access to a YML/XML feed is a salesperson who sees exactly what your accounting system sees in real-time. The difference is felt in the very first minute of dialogue with a customer who is asking about a specific product, not general terms.

If you already have a YML or XML feed for Yandex.Market or Google Merchant, you can connect it to an AI agent on your website without rebuilding your catalog.