From the very beginning everyone tells us “you are using the wrong model”. Fast forward a year, the free models become as good as last year premium models and the result is still bad but you still hear the same message “you are not using the last model”… I just stopped caring to try the new shiny model each month and simply reevaluate the state of the art once a year for my sanity. Or maybe my expectation is clearly too high for these tools.
Are you sure you haven't moved the goalposts? The context here is "agentic coding" i.e. it does it all, while in the past the context was, to me anyway, "you describe the code you want and it writes it and you check it's what you asked for". The latter does work on free models now.
When one is not happy with LLM output, agentic workflow rarely improves quality --- even though it may improve functionality. Now, instead of making sure that LLM is on track at each step, it goes down a rabbit hole, at which point it's impossible to review the work, let alone make it do it your way.
This discussion is a request for positive examples to demonstrate any of the recent grandiose claims about ai assisted development. Attempting to switch instead to attacking the credentials of posters only seems to supply evidence that there are no positive examples, only hype. It doesn't seem to add to the conversation.
There's people spending 5k a month on tokens, if you're work generates 7-8 figures per year, that's peanuts and companies will happily pay for that per engineer
Which is what it is by describing it as a tool needing thousands of dollars and years of time in learning fees while being described as "replaces devs" in an instant. It is a tool and when used sparingly by well trained people, works. To the extend that any large statistical text predictor would.
