CODE IN TOKYO

The Day My Wife Casually Said "Claude" at Dinner. Is Copilot Really in Trouble?

Claude Has Reached Ordinary Companies

Something surprised me the other day.

My wife works at a regular business company. She is not in the IT industry, and she is not the kind of person who follows every new development in AI.

While we were having dinner, she casually said:

“My company is going to test Claude.”

Wait.

Claude?

I did not expect to hear that name from an ordinary business company.

Until recently, Claude was something mostly engineers knew about. Now it seems to have reached the point where regular companies are testing it.

AI has really moved into everyday business.

My Wife Was Already Frustrated with Copilot

When I think back, my wife had often complained about Copilot at work.

“It is not useful.”

“It does not give me the answer I want.”

“In the end, it is faster to do it myself.”

I usually listened from the side and thought, “Yes, that sounds about right.”

Even among engineers, Copilot does not always have a great reputation. I also use Claude more often in my own daily work, so I understand the feeling.

That is why I was a little surprised when I heard that a business company was going to try Claude.

It felt like AI adoption had reached another stage.

Copilot and Claude Are Not the Same Kind of Thing

There is one thing that bothers me, though.

Online, I often see people ask:

“Which is smarter, Copilot or Claude?”

But this comparison is a little off.

Copilot is not the name of an LLM.

It is the brand name for Microsoft’s AI environment and AI services.

Which LLM is used inside that environment is a separate question. Depending on the use case, it may involve OpenAI models, Microsoft’s own models, or other model configurations.

Claude, on the other hand, is an LLM developed by Anthropic.

So if we want to compare things accurately, the comparison should be between LLMs such as GPT and Claude, not between Claude and Microsoft 365 Copilot.

To put it another way, the more relevant question might be:

“What would happen if Claude became selectable inside Copilot?”

Of course, from a user’s point of view, this distinction may not matter much.

The AI tool used at work changes from Copilot to Claude.

That is the visible reality.

The Real Test Comes After the First Excitement

What I am really interested in is what happens a few months later.

At first, the reaction will probably be:

“Claude is amazing.”

“It is smarter than Copilot.”

I also rate Claude highly, especially for Japanese writing and for understanding what the user is trying to do.

But that does not mean AI suddenly makes every work task go well.

If the request is vague, AI gives a vague answer.

If the assumptions are missing, AI fills them in on its own.

If the purpose is unclear, AI may simply produce something that looks plausible and stop there.

This does not change just because the model is Claude.

So if, six months from now, my wife says:

“Claude is not useful either. It keeps lying.”

I probably will not be surprised.

And I already know what I will say:

“Maybe the problem is not Claude. Maybe you are still asking it in the same way you asked Copilot.”

Better Models Do Not Automatically Create Better Users

Models will continue to change.

After Claude, an even smarter LLM may appear.

But the way people ask AI to do work will not automatically update itself.

In the end, if the way we use AI does not change, the only thing that changes may be the name of the tool we complain about.

First it was Copilot.

Next it may be Claude.

The real issue may not be which model is smarter.

It may be whether our way of giving work to AI has improved at all.

Contact

I provide system development and technical consulting using AI, AWS, and Claude Code.

Contact form: https://holly-money-e94.notion.site/390ff30cf18c8086a676fe630d171873