“The Day ‘Customers’ Are No Longer Human Has Arrived”

DoorDash launched its AI agent feature in May. When users send a text saying, “I want something spicy today,” the AI combines past order history, ratings, and location data to select a restaurant, choose a menu, and complete the order.

Airbnb also fully introduced AI search around the same time. If you input, “I want to stay near the beach with two kids next month,” the AI interprets the conditions and suggests the best accommodations. Users no longer need to endlessly scroll through lists.

What is the essence of what is happening here?

The act of ‘choosing’ has shifted from humans to AI.

Until now, restaurants and accommodations were dealing with the “human eye.” The appeal of photos, the number of stars in reviews, and the ranking on the homepage were all factors that humans considered when deciding, “Let’s go with this one.”

However, AI agents do not look at photos. They do not sense the atmosphere. They read data. They analyze structured information, score it, and return the top results.

The question is simple: Is your restaurant providing the ‘information’ that AI can ‘choose’ from?

DoorDash: In a World Where AI Places Orders, ‘Menu Data’ Becomes Competitive

Breaking down the logic behind how DoorDash’s AI agent selects restaurants reveals several important elements:

• Structured data for menus (dish names, prices, calories, allergens, cooking time)
• Text analysis results of reviews (frequency of keywords like “spicy,” “portion size,” “value for money”)
• Order completion rates and repeat rates
• Delivery time performance data

When dealing with humans, you could win with “eye-catching photos.” But against AI agents, the granularity and accuracy of data become the battleground.

What specifically changes?

For instance, there is a restaurant that simply states “Today’s Special” on its menu. A human might be curious and think, “I wonder what it is?” But the AI will skip it, judging it as “insufficient information.” On the other hand, a restaurant that states, “Charcoal-grilled domestic chicken thigh / 580 yen / 450 kcal / 12 minutes cooking time / gluten-free” is deemed “easy to choose” by the AI.

This is not just a story for large corporations; it is quite the opposite.

Large chains have fixed menus, and their data is already organized. The difference lies in whether small and medium-sized individual restaurants will organize this data or not. If they do, they can compete on the same level as chains. If they don’t, they will disappear from the AI’s options.

In terms of cost, organizing structured menu data can be completed in a day using tools like ChatGPT or Claude. Even outsourcing it would only cost a few tens of thousands of yen. There is no need to replace a 3 million yen POS system. Simply rewriting the text of the menu can change the probability of being chosen by AI.

Airbnb: AI Search Offers Opportunities for ‘Previously Overlooked Accommodations’

Airbnb’s AI search is fundamentally different from traditional filter searches.

Traditional: “Tokyo / 2 people / July 10-12 / under 15,000 yen” → Displays a list of matching conditions.
AI Search: “A rural old house where kids can catch insects, with a place for BBQ” → The AI interprets the meaning and makes suggestions.

What does this difference mean for small and medium-sized accommodations?

In traditional filter searches, they lost to large players based on price, location, and number of reviews. Large urban hotels and popular facilities with thousands of reviews dominated the top spots.

In AI search, the specificity of the ‘experience’ becomes a weapon.

“A 100-year-old old house,” “Breakfast with fish caught by local fishermen in the morning,” “An open-air bath with a view of a starry sky” — if such information is specifically written in the listing, the AI is likely to return this accommodation as a top choice for the query “a place where kids can experience nature.”

In other words, writing about ‘specific experiences’ that large hotels do not have in a way that AI can read becomes the greatest weapon for small and medium-sized accommodations.

Let’s talk about costs again. To rewrite the listing description, you need to inventory the charm of your accommodation and convert it into specific language that AI can interpret. This can also be done in half a day using AI tools, with almost no cost involved.

Because they are small, they have unique experiences. Making those experiences into data that AI can read will change the game.

Real Structural Change: ‘Discovery Costs’ Approach Zero

Let’s abstract the discussion so far to a higher level.

Until now, small shops and accommodations incurred ‘discovery costs’ to be found by customers. They had to advertise, work hard on SEO, rely on influencers, and subscribe to paid plans on review sites. This cost ranged from tens of thousands to hundreds of thousands of yen per month.

With the proliferation of AI agents, this structure is changing.

AI does not see advertising spaces (at least for now). AI looks at data. This means that restaurants with accurate and specific data have the potential to be chosen by AI without any advertising costs.

Conversely, establishments that previously paid for higher visibility through advertising will lose that advantage for orders coming through AI agents.

This presents a once-in-a-decade opportunity for small and medium-sized businesses.

The ‘right to be discovered’ that large corporations bought with advertising costs is being redistributed based on the quality of data. A world is emerging where businesses can compete based on the specificity and accuracy of information, not financial power.

So, What Should You Do Starting Today?

Putting aside the abstract discussions, let’s narrow down to three things that small and medium-sized restaurants and accommodations can do this week.

1. Rewrite Menu and Service Information in a ‘AI-Readable Format’

• Clearly state not just the dish names, but also ingredients, cooking methods, calories, allergens, and prices.
• For accommodations, write specific details about experiences (“What can be done,” “What can be seen,” “What can be eaten”) using proper nouns and numbers.
• Replace vague expressions (“heartwarming hospitality,” “carefully crafted dish”) with concrete facts.

2. Implement Structured Data (Schema.org, etc.) on Your Website

• Apply Schema.org markup recommended by Google to menu and facility information.
• This will become the “common language” for AI agents when retrieving information from the web.
• If you use WordPress, it can be done with free plugins. The technical barrier is low.

3. Use Review Responses as an ‘Information Addition Opportunity’

• Stop replying with just “Thank you.”
• Include AI-readable information in your responses, such as “Our XX uses ingredients from YY and is prepared using ZZ cooking methods.”
• Review responses are essentially a free channel to add information.

None of these involve significant costs. What is needed is simply a perspective shift to review your restaurant’s information under the premise that “AI will become the customer.”

‘Restaurants Not Chosen by AI’ Will Be Equivalent to Non-Existence

Finally, I must say something harsh.

The ratio of orders through DoorDash’s AI agent has rapidly increased within months of its release. Airbnb’s CEO, Brian Chesky, has suggested that it is only a matter of time before the usage rate of AI search surpasses that of traditional search.

This trend is irreversible.

In five years, when more than half of food orders are made through AI agents, restaurants not chosen by AI, no matter how delicious they are, will not be included in customers’ options.

“We compete based on the taste of our dishes” — I understand that sentiment. But before delivering that taste, you must first be found by AI.

Just as restaurants that did not take photos struggled in the age of Instagram, those that do not organize their data in the age of AI will quietly disappear.

What needs to be done is simple. Rewrite your restaurant’s information in a format that AI can read. Starting today.