Sure it’s different. But the point is early stage development is not the hard part. LLMs so far perform well within training data, which there is a ton for crud

Sure it’s different. But the point is early stage development is not the hard part. LLMs so far perform well within training data, which there is a ton for crud. We can’t say how it performs maintaining projects yet, it’s not clear it can be pattern matched in the same way.