1. What do you do for work?

I'm a product manager.

2. Has AI been helpful? && 3. What tools have been the most helpful?

I use claude projects for lots of different use cases, and I use claude code to prototype simple features. It's generally helpful yes.

4. What tools have been the biggest letdown?

Writing quality.

I've tried:

- meta prompting (asking an LLM to make an LLM prompt to do the task)

- providing examples

- prompt engineering to reduce the amount of obvious LLM giveaways in writing style.

I can't get claude sonnet 4/opus 4 to write with any taste.

There are two things LLMs can't do well yet in my opinion:

- To my knowledge they haven't generated any new or novel insights in any field or topic.

- They haven't produced what I would perceive as high quality writing.

I'd love to see prompts that others feel have produced good writing.

A little bit. I write for a living and code in support of my writing. It saves me from reading API docs for the sort of code I'm well able to verify. I also use it to test different ways to phrase something. Recently I used it to turn a diagram I doodle into Django code. It removes a bit of friction and I appreciate that.

I only use the free tier of ChatGPT. It's fine and I'm starting to get a good feeling for its limitations.

I also use Deepl a lot for longer, tedious German texts. I also use Deepl Write to check my writing, especially in German since I am not fluent. I love writing unimportant transactional emails in English and letting Deepl handle the translation. That saves me a lot of time.

I don't believe people's subjective reports. Even my own. I want more unbiased published studies.