Ludwig Pettersson

@bdc lets gooooo

Back to work.

NYC Compute

bonus of working on an image/video agent: I thoroughly enjoy skimming through the eval results

Also the irony that LLMs aren't that great at writing system prompts (tbc probably a skill issue)

Every time I catch myself trying to do clever system prompts + deterministic guardrails (old world thinking) and instead rethink it into a few simple primitives, my little agent loops get like 2-3x more capable and 90% less brittle Also: evals. Evals evals evals.

@DamiDina something new I'm experimenting with

@ludwig Origami or you’re making a new app?

spaghetti

@fredzaw something new!

@ludwig woah whats goin on here? origami in...something??

the curse of building a sandbox is it's quite fun in the sandbox itself

@benjitaylor same tbh

(too complicated)

I just had my first instance of an LLM using a tool in an unexpected way but _it produced a valid output_ so yeah you could say I know what it's like to see your child take their first steps

working with llms is mostly building scaffolding around this giant magnificent green alien blob that appeared one day and continously poking it with a stick to try to figure out why it did what it did and it's all very exciting

Teeeeechnically OpenAI had a custom font ~day 1, @benbarry asked the foundry to serif the capital I, see v1.nonlinear.co/openai

@benblumenrose @bnj @bdc @typochondriac yes my personal favorite is definitely the K1 (pictured) but naturally I have a soft spot for the K6 as they were installed on such a large scale (and who can forget the widely panned K8, pictured, where they went a little bit too Apple with those big panes of glass!!)

@bnj lol - to be fair he's been around some of the best designers in the business e.g. @ludwig, @bdc, @typochondriac, and many others. All lovers of historic phone booths too i'm sure ;)

Patrick is not a designer founder but he's not NOT a designer founder ;)

Yes yes we all know it’s very good yada yada but it just keeps replacing functions and impressing me 2 years on — everything from interpreting an all Japanese menu, telling me how to order, to teaching me about basic physics, every day still

I am obviously biased, but whatever, after literally years of consideration I think ChatGPT (especially o1 pro) is the most impressive thing since the iPhone