Same here - completely relate.

One thing I’ve noticed though that actually coding (without the use of AI; maybe a bit of tab auto-complete) is that I’m actually way faster when working in my domain than I am when using AI tools.

Everytime I use AI tools in my domain-expertise area, I find it ends up slowing me down. Introducing subtle bugs, me having to provide insane amount of context and details (at which point it becomes way faster to do it myself)

Just code and chill man - having spent the last 6 months really trying everything (all these context engineering strategies, agents, CLAUDE.md files on every directory, et, etc). It really easy still more productive to just code yourself if you know what you’re doing.

The thing I love most though - is having discussions with an LLM about an implementation, having it write some quick unit tests and performance tests for certain base cases, having it write a quick shell script, etc. things like this, it’s Amazing and makes me really enjoy programming since I save time and can focus on doing the actual fun stuff

When I'm doing the coding myself, I'm at least making steady progress and the process is predictable. With LLMs, it's a crapshoot. I have to get the AI to understand what I want and may have try again multiple times, many times never succeeding, and end up writing a lot of text anyway. And in between, I'll have to read a lot of code that probably ends up being thrown away or heavily modified. This probably depends a lot on what kind of project one is working on, though.

But it's like you said, I like using LLMs for completing smaller parts or asking specific kind of help or having conversations about solutions, but for anything larger, it just feels like banging my head to a wall.

One of the core principles of my workflow (inspired by REPL development and some unix tools) is to start with a single file (for a function or the whole project). The I refactor the code to have a better organization and to improve reliability, especially as I'm handling more scenarios (and failure modes).

LLMs are not useful in this workflow, because they are too verbose. Their answers are generic and handle scenarios you don't even support yet. What's useful is good documentation (as in truthful) and the code if it's open.

This approach has worked really well in my career. It gives me KISS and YAGNI for free. And every line of code is purposeful and have a reason to be there.

For those discussions with the LLM do you just use Gemini chat or chat GPT etc... i.e. the chat interface?

Devs are starting to realize that the sweet spot for AI support in coding is on a small scale, i.e. extended code completion. Generating huge chunks of code is often not reliable enough except for some niches (such as simple greenfield projects which profit from millions of examples online).