Started subscribing to Cursor around April last year. Happy human-machine collaboration for half a year. Got swept into CC in early January and quickly upgraded to the $200 plan.
All of January was pure joy from the massive productivity boost: Chrome plugins, websites, desktop apps. Reaching 99% was effortless. By the end of January, I broke up with Cursor.
The Limits of Vibe Coding
My own limitations became glaringly obvious through vibe coding:
- Self-control: Going from 0 to 1 is too easy. Too tempting to open projects indiscriminately. Not the best ROI choices.
- Limited bandwidth: Too many threads and context switches caused massive prefrontal cortex overload. Exhausted yet wired — it disrupted my daily life.
From Vibe to Value
In February I deliberately changed the rhythm of vibe coding: resisting the urge to start new projects. Constantly reminding myself not to reinvent the wheel. Focused on going deep with the finance agent app. Or just building glue apps to maximize ROI at minimal cost. Soon I felt my gaps in product value thinking. 99% is easy, the last 1% is the hard part.
Daili Liu
In early February, openClaw appeared out of nowhere. During a break at work, I gave it a name: Daili Liu (刘呔礼). You need to say "Dai" with a rising, spirited tone. Still immature, but it clearly shows the future: LLM → agent → openClaw (AI assistant or new OS).
More and more friends are starting to use coding agents. Riding the wave together, keeping things light and clear.
P.S. My favorite joke this month: Han Han's Pegasus is so good — how is he great at everything? Because Han Han is always "sure, sure, sure" (a pun on his name in Chinese).