Claude Bloom Singapore

What was discussed

We opened with the story behind the phrase "Claude Blue." It came out of an interview with an engineer at Meta who said that AI had made him more productive and yet he kept getting more depressed. The cause turned out to be a confusion about identity: he no longer knew whether he was a prompt engineer or just a person typing. The piece written about that conversation traveled much further than expected, in Korea and abroad, and out of it came a frame, from the first rush of excitement, through anxiety (Blue), to adapting with a sense of purpose (Bloom).

Junho started out analyzing government proposals and turned that work into a company. At first he was anxious that AI would replace his own business model, but he said everything shifted once he changed the question. Instead of asking what was safe from AI, he asked what could grow because of AI. That is why he moved from a subscription model to revenue sharing. He also mentioned that his ten-year-old son now builds 3D games with Claude instead of playing them. The cost of building has collapsed, he concluded, but only people still know what users actually want.

Yier said the past year had brought every emotional stage in turn. Once the phase where coding was fun had passed, the anxiety of having to learn endlessly set in. Eventually the answer was accepting that mastering every tool is impossible, and switching to learning in small pieces, one project at a time. Burning a lot of tokens, Yier added, is not a sin. A career that ran from product to strategy to finance and on to AI architecture was the proof: experience compounds at the points where domains overlap.

Max runs an AI-native startup where three people do the work that would once have taken 10 to 15. The feeling of building things with your own hands is a real reward, he said, but it comes at a price. Every time a new tool appears, he feels an almost addictive compulsion to keep optimizing and keep switching. He admitted plainly that he is tired of chasing the latest tools. He also had an explanation for why Anthropic does not run online events: the company believes the connections people make face to face are simply different.

The numbers came up too. In a recent survey, productivity was up, but burnout had climbed from 45% to 56%. One attendee confessed to an obsession with keeping agents running 24 hours a day, setting alarms in the small hours to queue up new tasks. The attendee called it AI-induced madness, and in the same breath said the satisfaction of watching a complex system run on its own outweighed the sleep.

Keziah led the meditation session. It was an experience called "The Cave," built with Claude Code and Gemini. Someone who had never coded before used AI to visualize the idea: particles and sound standing in for the stream of consciousness, and a focal point that holds your gaze while you practice staying in the present moment. Constantly worrying about whether AI will replace you means living outside the present, she said, and what you are left with in the end is the muscle of noticing that your attention has scattered and coming back to now.

The small groups picked up the same threads. A machine learning engineer with more than 10 years of experience had burned out chasing every new framework, and fixed it by cutting the options at each layer down to two. People from banking talked about how the organization's expectations reset every time a new feature ships, and said they had found Bloom in the slow, long-horizon change projects rather than the quick wins. Then came the demos: a quiet social platform stripped of ads and engagement hooks, a tool that turns the notes of a fragrance into a bottle shape, and a project that ties the physics of a double pendulum to typography to generate handwriting out of deterministic chaos.

What stood out was how closely Singapore's response mirrored Korea's. The approaches were different: Singapore is the more cautious of the two, Korea the one that moves first and figures it out later. But the anxiety people feel in front of AI does not care about geography. The conclusion we were left with was to change the question. Not how do I survive AI, but how do I use AI to grow the work I am trying to do.