Your next customer might not even be human – here's how to prepare for AI purchasing agents

One UK retailer saw its share of sales generated by AI agents grow from 0.3% to 2.5% in a year—an almost eightfold increase. Shopify reported an 18-fold increase in the volume of purchases made by AI agents in the first quarter of 2026. This isn’t a far-fetched vision: some of your customers aren’t necessarily human anymore.

If you run a webshop, this article is about what agentic commerce is, where it stands in Hungary, and what you should start doing before your competition does it before you.

What is agentic commerce and why is it exploding right now?

The essence of agentic commerce is that an AI agent does not just recommend or compare, but – with the owner’s permission – also carries out the purchase independently: it searches for the product, compares prices and conditions, and then completes the transaction. According to Mastercard’s forecast, by 2030 more than 10% of online shoppers would regularly shop with an AI agent – the process could really start in the areas of food, medicine and subscriptions.

There’s another sign that this isn’t a narrow niche: a survey of 26,000 parents and teens across 13 countries found that teens are roughly twice as likely to use AI to make purchasing decisions as their parents – about 251% of teens would use a fully autonomous shopping assistant, compared to only 161% of parents. Today’s young shoppers are used to a different paradigm than most online stores.

Mastercard-OTP precedent launched in Hungary

This is not a theoretical question here either. OTP Bank and Mastercard recently conducted Hungary's first authenticated AI agent transaction, and Gránit Bank, MBH Bank and K&H Bank have since joined the initiative. This means that from the payment infrastructure side, Hungarian banks are already working on the technical background of AI agent shopping - the question is more whether your webshop is prepared for such an agent to find and trust your products at all.

What does an AI agent do in practice today? – the Shopify example

Shopify recently expanded its support for what it calls WebMCP to the checkout process. They introduced three new tools: get_checkout, update_checkout and complete_checkout. With these, an AI agent running in a browser can not only add a product to a cart, but also view and modify shipping details, and then – with the customer's approval – complete the order.

The technical point worth noting is that the agent does not „read” screenshots or scrape the surface, but communicates with the site via a structured API. An interesting contrast is that Amazon, on the other hand, currently blocks purchases by AI agents – so the market is far from unified, and at this early stage it can be a distinct advantage for a webshop to prepare in time.

What does an AI agent need to „see” your product?

An AI agent can only make good decisions about your product if it is given good, machine-readable data about it. In practice, this means:

• Full, structured product data: title, description, image, price, stock level, variants and attributes – not only in the frontend text, but also as structured data (e.g. schema.org Product/Offer markup).
• Exact specifications in searchable format: If an important feature – say „100% merino wool” or „machine washable” – is only mentioned in the product photo or in a catchy marketing text, the AI agent is working with uncertain data and will prefer a competitor that clearly communicates this.
• Shipping and return terms as data, not just text: It is also worth making these available in a structured and clear manner.
• Up-to-date, real-time price and inventory data: AI agents verify data before recommending or purchasing – an outdated price or „in stock” indication quickly leads to a loss of trust.

Practical steps for your own webshop

You don't have to solve everything at once, but it's worth going through these few points in order:

• Structured data / schema markup review: The presence and accuracy of Product, Offer, and AggregateRating type markings on all key product pages.
• Following WooCommerce (or other platform) AI protocols: The Shopify and WooCommerce ecosystems are also constantly expanding the data formats and protocols compatible with AI agents – this is worth monitoring and implementing.
• Organizing product feeds: The same disciplined, accurate, and up-to-date data provision that you expect from a good Google Shopping or Facebook feed will now be a basic requirement for AI agents.
• This is not the end of UX and design: A beautiful and fast website optimized for human customers still matters - you simply get a second, machine "customer" who speaks a different language.

Summary: you now have two clients

Your webshop now serves two types of visitors: the human who browses, and the AI agent who reads data and makes decisions. Those who recognize this in time and organize their product data, feeds, and structured markup accordingly have a good chance of gaining an advantage by the time this topic becomes self-evident in the wider market.