Conclusion

Let’s get straight to the point: The moment something is labeled as “AI,” it becomes unusable in the field.

McDonald’s has integrated AI into its drive-thru operations, while Microsoft has effectively scrapped the “AI PC” brand.

At first glance, these two pieces of news seem unrelated. However, they share a common root.

The one that doesn’t say “It’s AI” wins; the one that does loses.

This isn’t just a story about large corporations. It’s the very wall that small and medium-sized enterprises (SMEs) in local areas face when they try to implement AI.

—

What McDonald’s Did — “Fast Drive-Thru” Instead of “AI”

McDonald’s has introduced a system that incorporates AI into the drive-thru ordering process. It uses voice recognition to take orders, instantly informs customers of any out-of-stock items linked to inventory data, and suggests menu items.

The key point is that from the customer’s perspective, it simply feels like “Wow, today is smooth.”

There are no signs saying, “AI will take your order.” What customers experience are results like “shorter wait times” and “fewer order mistakes.”

Looking at McDonald’s goals through numbers makes it clearer:

• Average wait time at drive-thrus: about 6 minutes (industry average). If this can be reduced by 10-15%, the number of cars processed per hour at each location increases by several.
• Assuming an average customer spend of 700 yen per car, if 20 more cars are processed in a day, that results in an additional daily revenue of 14,000 yen per store, or 420,000 yen per month.
• Considering the global scale of 40,000 stores, even a few percentage points of improvement could translate into impacts in the hundreds of millions of yen.

Moreover, this isn’t about “creating new revenue”; it’s about increasing the turnover rate of existing operations. There are no additional labor costs; in fact, they are likely to decrease. This makes it easier to forecast a return on investment.

This is crucial. The cost of implementing AI continues to decline. With cloud-based voice recognition APIs, the cost can be just a few yen per request. The additional hardware needed at stores is increasingly just a tablet and a microphone. Systems that once cost tens of millions of yen can now operate on monthly SaaS fees of just a few tens of thousands of yen.

As a result of falling costs, we have entered a phase where the issue is no longer whether to implement AI, but rather the cost of not implementing it.

—

Why Microsoft Abandoned the “AI PC” — Labels Got in the Way

On the other hand, Microsoft has effectively pulled back its prominently launched “AI PC” and “Copilot+ PC” brands in 2024.

Why? Because consumers didn’t buy it.

To be precise, the label “AI PC” did not encourage purchases. According to surveys, many consumers didn’t understand what would be convenient about an “AI PC.” Just the information that “it has AI” wasn’t enough to justify a price premium of 50,000 to 100,000 yen.

Interestingly, the features included in Copilot+ — such as image generation, text summarization, and real-time translation — received positive feedback from users. The features were good. But the label “AI” unnecessarily raised expectations and simultaneously stirred vague anxieties.

As a result, Microsoft decided to shift its branding away from emphasizing “AI” and instead focus on the intrinsic value of the features themselves.

This is highly suggestive.

When trying to sell based on the name of the technology, the substance of the technology fails to get across.

“Promoting DX,” “Utilizing AI,” “Digital Transformation” — the moment these labels are applied, people in the field think, “Here we go again with something from above,” leading to a halt in thought. I have seen this scene repeatedly in local SMEs.

—

What’s Happening in SMEs — It Stops the Moment You Say “It’s AI”

Let’s talk about a real case.

I once supported a local manufacturing company (with 30 employees) in automating its estimation process. Based on 500 past estimation data points, we created a system that provides rough estimates when conditions are input. Behind the scenes, an AI regression model is running. The development cost was about 400,000 yen, and the monthly operational cost is around 5,000 yen.

During the introduction phase, I initially explained, “We will automate the estimation process using AI.” The response from the field was:

• “There’s no way we can present an estimate generated by AI to a customer.”
• “Who will take responsibility if it’s wrong?”
• “There’s no way AI can replace my 30 years of experience.”

All of these concerns were valid. However, I changed the explanation.

“This is a tool that searches past estimation data and displays reference amounts for similar cases.”

The action being taken is the same. But the moment I said this, the seasoned veterans in the field thought, “Oh, that might be convenient,” and began to use it.

As a result, the time taken to create estimates was reduced from an average of 45 minutes per estimate to 12 minutes. Since the company produces about 60 estimates a month, this resulted in a reduction of approximately 33 hours per month. In terms of hourly wages, that’s about 100,000 yen worth of labor saved. The initial investment of 400,000 yen was recouped in four months.

The essence of this story isn’t about the “accuracy of AI.” It’s the fact that “the field used it because it wasn’t called AI.”

—

Another Example — A Company That Failed by Calling Itself an “AI Chatbot”

There’s also a reverse case. A small e-commerce company implemented an “AI chatbot” for customer support. The aim was to automate FAQ responses with a service costing 30,000 yen per month.

They placed a button on the site saying, “Ask the AI chatbot.”

As a result, the usage rate was below the expected 20%. Customer surveys revealed many concerns such as “I’m worried about entering personal information into AI” and “It’ll probably give irrelevant answers.”

Three months later, they changed the button text to “Search Frequently Asked Questions” and replaced the chat interface with a simple search box. The underlying mechanism remained exactly the same.

The usage rate tripled. The number of inquiries decreased from 120 per month to 45 per month. The support staff’s response time was reduced by 40 hours a month.

The only changes made were to the label and the UI. The technology itself didn’t change at all.

—

What’s Happening Structurally

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

The patterns are clear.

The value of AI lies not in “being AI” but in “what becomes faster, what becomes cheaper, and what becomes easier.”

Furthermore, the label “AI” currently hinders rather than helps in conveying value. It either unnecessarily raises expectations or stirs vague anxieties.

This is also the fate of technology. Once, “cloud” went through the same path. There was a time when people said, “It’s dangerous to store data in the cloud.” Now, no one says, “This service is in the cloud.” It has become commonplace. AI will follow the same path.

When we stop saying, “We use AI,” we can truly say that AI has become widespread.

—

So, What Should SMEs Do?

I’ll say three things.

1. Call it “Business Automation” Instead of “AI Implementation”

Both internally and externally, it’s more effective to avoid using the term “AI.” Simply rephrasing it as “a tool that automatically provides reference values for estimates” or “a sheet that extracts trends from past data” can eliminate resistance in the field.

2. Start with “Cost-Reducing Tasks”

Using AI to increase revenue is challenging. However, utilizing AI to reduce costs has high reproducibility. Tasks like estimate creation, meeting minutes, invoice processing, and FAQ responses — these are ideal targets for automation. Tools costing a few thousand to tens of thousands of yen per month can save dozens of hours of labor. ROI can be realized in a few months.

3. Test Small and Make Decisions Based on Numbers

There’s no need to think about “company-wide implementation.” Start with one task, one person, and a one-month trial. If the numbers are good, you can expand; if not, you can stop. The criteria for judgment should be based on whether “time has decreased, mistakes have decreased, or costs have gone down,” not whether it’s AI or not.

—

In Conclusion

McDonald’s made its drive-thru faster without saying “AI.” Microsoft failed to sell when it labeled its product as “AI.”

This contrast teaches us that the name of the technology can determine its survival or demise.

For SMEs, utilizing AI isn’t about chasing the latest technology. It’s about quietly, reliably, and cheaply automating the tasks at hand. And it’s about not calling it “AI.”

The AI that survives in the field doesn’t call itself AI.