LLMs have only really gotten good at coding in the past 6 months or so and always will have gotten good enough just 6 months prior for years to come. You will always be using the wrong model and / or prompting it wrong. Deal with it, luddite
>>109686733Good.>work from home>spend half an hour making AI prompts to do all my tasks>spend the next 7 hours fucking around, watching anime, fucking my wife, playing with my kids, working around my house, shitposting>spend 30 minutes making pull requests>get paid full timeLife is amazing.
>>109686860What do you do to most efficiently compile requirements and turn them ino long running prompts? I use the Fable through the Claude API using the good boy points the company gives us, and while my work is much lighter compared to what it would be without AI, I am still having to check in every minute or few to verify actions, review dev progress from the AI, and deliberate what options are best in its planning feature. Very useful overall, but the more efficiency the better. Any help is appreciated, thanks!
>>109686860Enjoy while it lastsIn a year you are gonna be proompting 12 hours non stop to meet your KPIs
>>109688555This already happens in every tech companyI worked at a cyber company and with Claude Code, MCP’s and spec kit development you’re basically coding with prompts, and now have way more to do in a single sprint. Of course they’ll use that to suck you dry and get more work done
>>109686733Skill issue. I got a lot of use out of Qwen3-30B and Qwen3-Coder-30B a year ago, and I get even more out of Qwen3.5-35B now.
