I think you'd be the first one, so I highly doubt that. If it's true, good on you.
I use Opus daily. It can take some typing off my hands, as long as I keep it to highly specific, limited, straight-forward things. And as long as I spend a long time preparing everything in the most minute detail. Then I have a chance of it being a slight efficiency gain. Veer slightly outside of those preconditions and the output is invariably impressive at first glance, garbage at second glance. Not to mention that it's barely any help for the 80-90% of the job that isn't writing code.
Even Jensen Huang expects the LLM efficiency boost to be about 30% for software engineers. Jensen freaking Huang, who has every reason to exaggerate the benefits! So I'm realistically taking that as an upper bound.
Run rate is easy to get while not meaning much (which is why GenAI vendors love talking about it). Report back in a year about how things are going, and how much of your code you had to rewrite from scratch.
I think you'd be the first one, so I highly doubt that. If it's true, good on you.
I use Opus daily. It can take some typing off my hands, as long as I keep it to highly specific, limited, straight-forward things. And as long as I spend a long time preparing everything in the most minute detail. Then I have a chance of it being a slight efficiency gain. Veer slightly outside of those preconditions and the output is invariably impressive at first glance, garbage at second glance. Not to mention that it's barely any help for the 80-90% of the job that isn't writing code.
Even Jensen Huang expects the LLM efficiency boost to be about 30% for software engineers. Jensen freaking Huang, who has every reason to exaggerate the benefits! So I'm realistically taking that as an upper bound.
Run rate is easy to get while not meaning much (which is why GenAI vendors love talking about it). Report back in a year about how things are going, and how much of your code you had to rewrite from scratch.