I'm a cofounder of a Series A start-up, so work hours are hard to define. You are sort of always working or thinking about work and it is hard to put away. But

I'm a cofounder of a Series A start-up, so work hours are hard to define. You are sort of always working or thinking about work and it is hard to put away. But within the company, we've found lots of uses for LLMs to save time. One or two of the developers are fans of Github Copilot, but that is a relatively minor impact so far. The bigger impacts are around data cleaning and content. A lot of jobs that used to be "hire someone on Upwork for $20/hr to do a lot of boring clean-up work on thousands of records or write repetitive text about something" are now "write a 10 line python script with the OpenAI api to generate the result for $8 and then QA". There are also plenty of uses of boring old home-grown ML models and stuff like that to clean up data in cases where the problem is more specific and predictable. So far AI just makes our best developers faster and our best product people make a larger impact more quickly and at a lower cost. It hasn't resulted in less internal hiring or anything like that. But it has resulted in less Upwork-style one-off contracting, especially around data cleaning and content clean-up.