Wow. So stimulating.

I love how the experience starts simple and gets worse and worse the further you go.

Same, “chat” and “conversation” is a compelling party trick played by the application layer and not the LLM.

Not as good as my moms, but they’re pretty!

My mom always made sticky buns. Gooey dough balls covered in cinnamon caramel and pecans.

It’s not Christmas without them for me ha!

Gonna try to replicate this year for my kids, but never as good as moms.

Correct, I’m aligned with that definition.

If it’s just a workflow that’s basically static, I’d call that more RPA with LLMs involved (maybe performing some step).

“Buying”. lol, I don’t want to become a game console

Perfect, thanks! This makes sense.

Thanks for the answer 🙂

I like the advice in this post for when agents make sense

www.anthropic.com/research/bui...

Who could ever possible save me?

If only there was a reliable guide with tons of cash to fund my venture, I’d surely make it then

Narrator: they don’t

Congrats! Incredible journey.

Im also a self-taught developer, and in building startups, I’ve found a love of building things and there’s always space to learn.

Not sure how you did it but you’ve just won parenting.

C) a quiet cabin in the woods by a lake

Did you get automatically enrolled in mandatory security training that you must complete before Friday?

Hornofgondor9@hotmall.com Guilty 🤪

(I don’t have it anymore but it was something like that ha!)

makes it such that prompts can be tested and developed robustly.

Then, generating prompt SDKs in various languages means enterprises can incorporate tested prompts into applications rather than ad hoc prompts.

Imagine …

npm install @company/prompts

Same for Go, Python, Java, etc

TL;DR is that the whole LLM application space has a lot of pick and shovel type problems in it.

But, the space is so young that it’s hard to know what’s gonna stick.