Mythical Man Month

I don't know what Dario Amodei says, does, how Anthropic is run or structured, or what kind of people work there, so I can't comment on that.

I do know about myself, though. The increase is very real, measured by the number of (Linear) issues resolved. No, I haven't changed how I open or close those issues, I've been using the system for years. During the first three months of 2026 I went through 12x more issues per month than during past years.

But I guess I am not an "average programmer": 35 years of experience means I can work with AI as I would with a small (but very skilled) team. I can architect systems, notice unnecessary complexity, and intuitively choose solutions that are more maintainable. And I am a single-person company, with no managers to report to, KPIs to achieve, presentations to make, etc.

I do not make "more money". It's a mature SaaS. Changes in revenue are over the long term, on a scale of years more than months, and implementing features is no longer enough: marketing is needed for more growth.

But, to be honest, I am tired of defending myself this way. It's not the first time I post this metric, I thought people would find it an interesting data point. Instead, I get downvoted (see my comment above which currently sits at -1 in spite of being objective and factual), and then get plenty of responses asking me to defend my statements.

Come to think of it, I'd rather not convince people about AI increasing productivity so much. I'm not really sure why I bother to post here anymore. I'd rather have everyone (including my competition) believe whatever they want to believe and not use AI.

In a mature SaaS features do not map into more revenue quickly, so the effect on revenue can't be measured easily.

I think the statement is generally true, but I also think there are specific tools that do vastly increase dev time for something that follows that specific pattern.

Program templates and also code generation have been used forever to simplify and speed up specific types of programs, like all of the CRUD programs in an ERP system for maintaining the hundreds to thousands of different master data like Customers, etc.

Those approaches can speed a specific activity that follows a specific pattern by some multiplier (e.g. 4x or 10x).

But the trade-off is time to build the template or code gen vs number of instances it will be used.

I think where "no silver bullet" has the strongest case is for something completely new where the patterns and templates have not been established.

When I think about LLM's and compare to past templates and code gen, I wonder what exactly is the nature of the LLM trade-off in the broad sense. For a template, it's pretty clearly up front design+dev compared to number of instances of usage of the template.

LLM's are more general and more broadly applicable, but still not the same as a human. Is the trade-off purely cost that was sunk into the model and tool that now needs to be recouped by the AI company? (e.g. charging the actual $1,000 to $1,500 that it really costs instead of $200

Or are the big trade-offs and costs going to show up in the future: the dwindling of human expertise due to reliance on the tool and the loss of understanding of complex systems.