The Era of Human Search and Purchase is Quietly Coming to an End

Shopify has opened over 1 billion catalog items for AI agents. Google has established a system that allows individuals to distribute AI applications on the Play Store.

When read separately, these two pieces of news might seem like just “new features from major platforms.” However, when viewed structurally, they convey a singular meaning.

The pathway of “humans searching, comparing, and purchasing through browsers” is beginning to shift to a pathway where “AI selects and purchases on behalf of humans.”

The entrance to stores is changing. This is not just a discussion about SEO or advertising. It’s about the changing premise of “who the product information is organized for.”

For small and medium-sized e-commerce businesses, is this a threat or an opportunity? To conclude, I believe that “if they take the right actions, small and medium-sized businesses will actually find themselves in a more advantageous position.” Let me explain step by step.

—

What Shopify Opened Was Not a “Catalog” but the “Negotiation Rights for Shelves”

The core of the feature announced by Shopify is simple. When consumers ask their AI agents (such as ChatGPT or Gemini) to “buy this for me,” the AI searches for products from Shopify’s catalog, compares them, and completes the purchase.

The list of over 1 billion items becomes the AI’s “supplier.”

What we need to consider here is, “What criteria will AI agents use to select products?”

Humans buy based on the appearance of brands, the vibe of reviews, and a certain intuition. AI operates differently. It mechanically compares structured data—prices, specifications, stock status, shipping conditions, and review scores—to make selections.

In other words, to be “chosen” by AI agents, product data must be accurate, structured, and organized in a comparable format.

This may seem like a domain where large corporations excel. However, the reality is quite the opposite.

Large corporations often rely on legacy systems, taking years to unify their data. In contrast, small and medium-sized e-commerce businesses using Shopify typically have only a few dozen to a few hundred products. One person can finish organizing the data for all products in just one week.

Let’s break down the costs involved.

If done in-house, the costs are even lower. For businesses with fewer than 100 products, most of the work can be handled using Shopify’s management screen and CSV export. The actual cost is nearly zero; only time is required.

With an investment of less than 500,000 yen, a new sales channel via AI agents can be opened. It would be harder to find a reason not to do this.

—

What Happens If You Don’t Take Action

Now, let’s consider the risks of not organizing product data.

To be honest, no one knows yet what percentage of overall e-commerce purchases will come through AI agents. By 2025, it might be a few percent.

However, what is frightening when thinking structurally is the potential for a “slow but steady impact.”

Think back to the influx from search engines. Around 2010, stores that claimed “SEO doesn’t matter” found themselves buried beyond the second page of search results. AI agents are likely to follow a similar trajectory.

Let’s assume a small to medium-sized e-commerce business with annual sales of 10 million yen sees the purchasing ratio through AI reach 5-10% over the next three years. If they fail to organize their data and are excluded from AI’s options, they could lose 500,000 to 1,000,000 yen in annual sales “without even noticing.”

Even increasing advertising expenses won’t help recover this loss, as AI agents do not view ads.

—

The True Meaning of Google AI Studio—The “Cost of Creating Apps” Has Disappeared

Another piece of news: Google has made it possible for individuals to distribute apps created with AI Studio on Google Play.

Reading this as just a “developer-focused story” would be a waste.

The essence is that the development and distribution costs of AI apps have dramatically decreased.

In the past, when small businesses attempted to create their own apps, it would cost at least 3 to 5 million yen, and over 1 million yen annually for operations.

With Google AI Studio, prototypes of AI apps can be created in a near no-code format. Distribution is also completed via the Play Store. Development costs have effectively dropped to almost zero to a few tens of thousands of yen.

What does this mean for small and medium-sized e-commerce businesses?

For example, consider the following types of apps:

• “A shopping assistant exclusive to our store”—An AI app that suggests products based on past purchase history when customers ask for something similar to what they bought before.
• “Region-specific product recommendations”—An app that suggests local ingredients and specialties based on the season and purpose.
• “Automation of repeat orders”—For stores dealing with consumables, an app that notifies customers when they are running low and encourages reordering.

The era of commissioning app development companies for 3 million yen is over. Now, you can create it yourself, test it yourself, and if it doesn’t work, you can redo it. This cycle can be completed in just a few days.

This is a strong suit of small businesses. Large corporations take six months for approvals and verifications. Small businesses can release something by next week.

—

The Essence of Structural Change—The “Discovery Mechanism” Becomes Two-Tiered

Let’s summarize what we’ve discussed so far.

The opening of Shopify’s catalog is the foundation for being “discovered” by AI agents. The simplification of Google AI Studio’s distribution is a weapon for “connecting directly” from one’s own business.

In other words, the “points of contact with customers” for small and medium-sized e-commerce businesses will become a two-layer structure.

Layer 1: Being chosen by AI agents (passive)
→ Addressed through structuring and optimizing product data.

Layer 2: Direct connection through proprietary AI apps (active)
→ Control the relationship with customers on your own.

It is not correct to do just one or the other; doing both is the right approach. Moreover, both have a low cost barrier for small businesses.

—

So, What Should Be Done?

Let’s narrow down the specific actions to three.

1. What to Do This Week: Inventory of Product Data

Open the Shopify management screen and check whether the titles, descriptions, and attributes (size, material, usage, etc.) of all products are standardized. Export to CSV and eliminate any blanks or inconsistencies. Just this will complete about 80% of the “AI-readable product data.” The cost is zero. Time required is half a day.

2. What to Do This Month: Implement Structured Data

Add Schema.org Product type markup to all product pages. Many can be addressed through app or theme settings in Shopify. Even if outsourced, it would cost 50,000 to 100,000 yen. This will enable compatibility not only with AI agents but also with rich results in Google search. A win-win.

3. What to Do This Quarter: Experiment with Your Own AI App

Using Google AI Studio, start by creating a simple app. Something like “a chatbot that recommends our products in a Q&A format” is sufficient. Don’t aim for perfection. Release something functional and observe customer reactions. That will provide the greatest learning experience.

—

When the “Store Entrance” Changes, Only Those Who Adapt Will Survive

Once, the store entrance was defined by “location in a shopping district.” Then it changed to “search engine rankings.” Next, it will shift to whether or not a business is included in the “choices of AI agents.”

Throughout these changes, small businesses that responded quickly have survived. Stores that embraced SEO early won in search, and those that jumped on social media early built a fan base.

This time is no different. The difference is that the cost of adaptation is overwhelmingly lower than in past changes. Organizing product data costs tens of thousands of yen, and experimenting with AI apps can be done almost for free.

The issue is not technology or budget. It’s about whether one stops to think, “Is this okay as it is?”

The answer is clear. It’s not okay as it is. Therefore, I hope you start by opening the product data CSV this week.