Has anyone properly set up LLM programming workflow?
With new LLMs being released almost daily at this point and with availability of all the possible "spec-driven" workflows(BMAD, AgentOS etc), is anyone really developing production ready software with little to no intervention?
I have talked to a lot of software developers in my circles and most of them are using LLMs for autocomplete(which has gotten extremely good with Augment code and others) or attempting to "one-shot" some small features. I keep reading that many people are writing 10k lines of code daily which quite honestly sounds plausible. What I don't understand is whether that code is maintainable, modular, performant and otherwise production ready. I believe that with proper specs and setup, modern LLMs can do all that, but I have yet to hear real life examples of that.
Without revealing proprietary details, I can say that 2.9 million lines of Java code were generated and tested in 3 months via claude code with multiple max accounts. The technique is to decompose the specification, code generation, test generation, testing, documenting and other standard steps into minute prompts and responses within the task competency of claude code non-interactive. I also scoped tiny context by having a repo for each 3K module.
Much of the innovation was accomplished by asking the AI what was best practice.