Professional Software Developers Don’t Vibe, They Control: AI Agent Use for Coding in 2025 99 respondents, surveyed from 8/18 to 9/23, about three months ago. That was before GPT-5.2 and Claude 4.5, so it can only reflect trends as of three months ago, but it’s well worth a look.
Professional Software Developers Don't Vibe, They Control
ChiChieh HuangFounder · AI engineer Translated from the Chinese original · Read the original
LocatorVibe Coding Valley
2 wks
Overview
End of the trail
51 words, and you made it to the end.
Newsletter
Get the next essay by email.
One email when a new essay goes up, nothing else. Unsubscribe in one click.
← Previous · 2025.12.18After Observability, the Next Step Is Evals Next · 2025.12.28 →Rethinking Prompt Caching Through ngrok's Article
Nearby cairns · the closest essays on the map
Vibe Coding Valley2 days apart After Observability, the Next Step Is Evals Observability lets you see what an agent did; evals force you to define what success means. Notes on Anthropic's Demystifying evals for AI agents, from pass@k versus pass^k to grading outcomes instead of paths. Vibe Coding Valley8 days apart Rethinking Prompt Caching Through ngrok's Article ngrok's article on prompt caching made me realize the real cost isn't the price per token. It's recomputing the same opening prompt hundreds of times a day. Vibe Coding Valley15 days apart How Google Came Back Google looked stuck in the innovator's dilemma in 2022–2023, then caught up and briefly pulled ahead in 2025. Notes on a WSJ piece about the three things it got right.
Surveyor
ChiChieh Huang
I build generative AI products and write about them, first in Chinese. Lately I’ve been researching agent memory and testing the ideas in Cairn.