pull down to refresh

100 sats \ 0 replies \ @optimism 14h
Great article, thanks!
The only thing I disagree with somewhat is:
Always be part of the loop by moving code by hand from your terminal to the LLM web interface: this guarantees that you follow every process. You are still the coder, but augmented.
This is because the chatbot interface is used. But what if the "tool call" is/includes the LLM? I've found specific programmatic calls to LLMs, including post-call cleanup and processing, much more efficient than the generic chatbot interface for process automation. The user input can still be a prompt and the LLM can still have access to tool calls if needed, but for each token of instructions that is about tooling or context, i.e. anything other than just solving the problem at hand, is "distracting" and diminishes results.
reply
There seems to be a similar look in design with all of these vibe coding LLM's and such.
reply
0 sats \ 1 reply \ @optimism 2h
You mean when it prototypes a web app?
reply
Yeah exactly
reply
Just finished reading antirez’s “Coding with LLMs in the Summer of 2025” — really solid write-up. Appreciate how honest and grounded it is.
There’s so much noise out there about agents doing all the coding for you, but this piece brings it back to reality: LLMs are amazing tools, not magic. They help you move faster, catch bugs earlier, and explore design space way more efficiently — but only if you stay in control and guide the process.
The idea of pairing your instinct with the LLM’s PhD-level knowledge is 🔥. Totally agree that you need to feed it proper context, give it structure, and still be the one steering. Otherwise, you end up with bloated, brittle code.
Big thanks for sharing your workflow and mindset — it’s honestly the most useful take I’ve read in a while.