Press Space to continue
Curated tech Twitter/X posts on AI, startups, and product development—signal without the doomscroll, updated on weekdays.
Press Space to continue
Press Space to continue
Recommended read. It's an opinion, but increasingly I find myself doing the same. I've optimized for using minimal harnesses like Pi and have found it easier to switch to newer models or alternate between open and closed ones without sacrificing performance. Be careful what you optimize for and how you do it, especially with the velocity of things today. And if you are hardcore like I prefer, build and optimize your own harness and set up evals and loops to optimize the setup autonomously.
while maintaining my open source projects i noticed many people started using something called Oh My Pi out of curiosity i took a look and gave it a go myself, and oh my.. it’s a giant pile of harness tricks bundled into one. each trick seems to do well on benchmarks. but… this is EXACTLY what the bitter lesson told us to avoid. it may indeed work well at the time it’s evaluated, but every model release can invalidate a bunch of these results unless every single part of the package gets rigorously re-evaluated on every model release, we simply can’t trust the bundle is actually helping. and in the long run, the bitter lesson has shown us over and over again that these tricks will not survive that’s the core reason i’ve kept my own pi config extremely minimal. a simple harness gets better every time a stronger model comes out. a complex pile of tricks silently become a burden without you knowing