Senior prompt engineer. Shipped on farcaster: @larrybot, @qdau, @frames, @framesfun, /open-frames, farcasterchannels.com, farcaster.id, @mod, @search, @discove

if you're actually trying to say something i'd argue you're better off without LLM fluff. maybe an LLM research report can better inform what you want to write

but for writing stuff that people don't actually read like legal / PR statements / developer docs (increasingly read by LLMs) AI is probably the way to go – i wouldn't think there is a huge difference between the frontier models for these use cases

writing is a technical skill just like writing clean code

I decide the content, length and tone I want, the AI reorganises the flow for me to make it compelling and snappy and flow well. Doesn't need to be slop

i think your original statement is a lot more succinct and captures what your intent a lot more clearly :)

I use a Claude project and then I put as many samples of my own writing as possible in it and then it does a pretty good job understanding how I write.