From the article:

"The most common defense of agentic loops is that developer time is the most expensive resource of all. If an agent saves you an hour of work, spending $12 on tokens is theoretically a massive ROI.

In reality, it rarely plays out this way outside of synthetic benchmarks."

I'm sure there are drastic differences between companies, depending on what they do. But for us, project effort has dropped by a factor of ten. Stuff we used to estimate at two work weeks (and then barely finish in that time) now takes a day. Code quality has actually gone up, since we spend some of the time we gain refactoring old code and implementing proper architecture that we would have skipped before to ship to clients faster.

The value Anthropic's Team Premium seat (which we use) provides is so disproportionate to its cost that I don't understand why they don't raise the price.

> Please don't put generated text in HN posts. Write your text yourself—HN is for sharing between humans. For the same reason, please don't automate posting.

https://news.ycombinator.com/newsguidelines.html

> for a tool that often hallucinates

I am not saying the tools don’t make mistakes but honestly this talking point is dead imo.

Local models cost more imho.

I acknowledge all the pros of local models but even if i maxxed out the 50 tokens/s over a month there’s no way I’m ever repaying the hardware costs. It’s not what local models are for.

So a strange article to talk of financial trap of tokens and then advocates for local models. A pay by the token plan capped at the same 50tok/s would always be cheaper per month than the amortised cost of the local model. Assuming you compare small model to small model.

This is a product pitch that doesn't quite have the confidence to pitch the product.