Nobody Wants to Get Better at Work Anymore

A conversation with a PM at a public tech company in New York, and what my team is struggling with

The Claude Blue post I wrote on Brunch a few days ago went completely viral. It spread through KakaoTalk (Korea's main messaging app) group chats and social media accounts I'm not even part of. A reporter from a Korean daily newspaper got in touch, and so did a researcher at Harvard and developers at big tech companies overseas. I woke up to messages from strangers on LinkedIn and KakaoTalk open chats asking for coffee chats. I think they're curious whether I have some kind of answer, and I worry they'll be disappointed that I don't. Still, I figure we can trade insights as we talk, so I'm trying to meet nearly everyone.

At the same time, I was even more curious how people would react in the US, where the AI trend started. So I posted the same piece on Reddit, where I'm a regular, and it blew up there too: 630,000 views, 961 upvotes, and 300 comments, with the debate still going. Of course, if you search for "Claude Blue" you won't find much. (I made the term up, borrowing from "Corona Blue," the Korean term for pandemic-era depression, haha.) Even so, AI depression seems to be real, especially in Silicon Valley, and it looked like it was gradually spreading to other big American cities. Korea tends to lag a bit behind, so to get a read on the situation early, I've kept interviewing people beyond Reddit: friends in the US and people they introduce me to.

So this time I caught up with a friend I hadn't talked to in a while, a PM at a public tech company in New York. AI has definitely moved into the workplace fast. But contrary to what I expected, not everyone is reacting the same way. I wanted to understand why companies outside big tech and startups are slower to respond. Above all, I wanted to know whether the AI wave still makes people want to get better at their jobs, or whether the impact is so big that the urge to just run away is stronger. (We're in an era where nothing we took for granted can be taken for granted anymore.)

That made me think even harder about what we should be building for knowledge workers.

1. The reality of a PM at a big New York company

This person works as a PM at a public tech company in New York. The title is PM, but at this company it's closer to a mini CEO. They own the P&L and run an entire small business unit, from revenue and marketing to technical support and sales.

The company's writing culture is extremely strong, to the point that it reminds you of Amazon's six-pagers. A PRD runs 18 pages at minimum, and everyone reads every last word. If you get something wrong, you get torn apart. They use Claude and Cursor in this environment, but handing things to AI and shipping them as is would never fly. They have to go over everything personally, from early planning to the final check. AI is a second brain. They haven't handed over the brain itself.

What I found interesting was how they handle data. The company dashboards are hard to read and it's hard to find the data you want, so they dump entire tables into Claude, have it learn the data structure, and use that one instance to answer every data question. "That's my data engineer." They no longer need to go into the dashboards at all.

But this person is in the minority at the company. By their own account, very few PMs use AI actively, and managers don't even know how to use it.

2. Security isn't what's holding back AI adoption

I had assumed the real reason big American companies are slow to adopt AI was security, or maybe tech literacy. Their answer was much simpler: "There's no time."

There's a mountain of work that needs doing right now, and quarterly targets to hit. With layoffs in the air lately, people are actually working more than before.

Stripe is a good example of a company that tackled this head on. I heard they set aside an entire week company-wide so every employee could do nothing but learn AI. But most companies can't afford to give up that week. They need results now, they have customers to handle, and they can't absorb the cost.

We went through something similar at our company. We declared ourselves AI Native last year, but honestly, we weren't native at all. So starting this January, we made it mandatory: every Thursday, everyone downloads Claude Code and spends time using it themselves. Forcing it actually worked. But we can do that because we're a small startup. I can't imagine how you'd coordinate it at a company with thousands of people.

3. In the AI era, people no longer want to be good at work

Here's what they said, word for word: "Who wants to spend their time after work thinking about how to make more money? It's exhausting." It's a very realistic thing to say. When I think back to my own years as an employee, honestly, I was busy just keeping up every day. I didn't have much room or desire to build my skills after hours or on weekends. I didn't put any effort into LinkedIn, social media, or networking either. I didn't need to.

These days AI is having the opposite effect from what you might expect. You'd think it would push people to get better at work so they can beat it. Instead it's amplifying AI depression and anxiety, the feeling that "I'm worthless now." That's the Claude Blue phenomenon I wrote about. So people are less interested in "using AI to do better at my company" and more preoccupied with a different question: "I'm going to get cut eventually anyway, and someday I'll probably have to start my own business whether I want to or not. What would that be?" This PM was also thinking they'd have to start their own business eventually, and said they were exploring options now. The New York PM I talked to today and the Silicon Valley big tech engineer I talked to before were both looking in the same direction.

4. The rise of the CPTO and the breakup of job roles

The conversation naturally turned to the future of the PM role. They told me about an interesting trend: more companies are creating a CPTO position, combining the CPO (chief product officer) and CTO (chief technology officer). Their company already has one.

The reason is simple. With AI, a CTO can handle product, and a CTO can take on the CPO role too. The lines between roles are collapsing. As Jensen Huang has put it, we now live in a world where the difference between a "good AI user" and everyone else has become the measure of skill. Marketers run SEO analysis in Cursor, PMs touch code, and non-developers run agents. Sam Altman's "one-person billion-dollar startup" is looking less and less like a joke.

Startups like Ramp have reportedly connected everything through MCP, so their workflows themselves now run on AI. Those companies have already realized they don't need low-level PMs, and they're keeping only the people who can set strategy and execute in the field. Big companies are much slower. Even within the US, the speed gap between startups and big companies is now measured in years.

5. So what should we build?

Up to now, we've been building tools that help knowledge workers work more efficiently with AI. But honestly, the more interviews I do, the more I doubt that people today still have any desire to "do better work with AI."

Maybe we should flip it and build a service that helps people get out of the office, instead of one used in the office. The thought crossed my mind, half joking and half serious. I happened to see a post on Instagram saying high-rise office buildings are emptying out and could soon turn into slums. If employers stop hiring people, they won't need offices either.

I also feel anxious that in a few months, the services we're building now could become meaningless. Judging from the interviews and public sentiment in the US, that view clearly exists, and I think it's only going to grow.

People who became YouTubers before YouTube took off got the early-mover advantage, and I think moving first during this transition could matter in the same way. I just don't know which direction to move yet. I suspect this is the honest worry of everyone in software right now, all over the world, from the bosses to the employees.

To find an answer, I'm going to keep meeting people on the ground, especially office workers in the US. Once I have even a rough sense of the answer, I'll be sure to share it.

I'm not an AI expert, but I'm always up for a coffee chat. If you'd like to share your story or hear more about other people's experiences, reach out on LinkedIn.

Originally published in Korean on Brunch, March 22, 2026.