AI is all the rage these days so I spent the weekend building a @tferriss AI to answer questions addressed to him or any of his past guests. tim.nem.ai

Why does Astra just give up so often?

It's kind of funny how AI companies are basically double dipping because they created LLMs to hack everyone and defend everyone

The amount of work AI creates is at least an order of magnitude larger than the amount of work it removes.

at least for those that build software directly (but I suspect its much wider) the appetite for AI's execution speed is wildly mismatched to the portions being served I hate waiting 20+ minutes for agents to be done

There’s no such thing as a solved problem. All solutions are temporary and striving to solve things nobody else has tackled is overrated. Solving existing problems with better solutions can work just as well. @loundai is fighting in a saturated space, slow and steady.

AI is slow on task 1 so instead of waiting I start task 2 which uses even more of my cpu/ram so AI is even slower so I wait for 2 tasks instead of 1

man I'm a Codex power user at this point and Claude Cowork is so much better in a lot of small ways

We went from model intelligence mattering the most to harnesses mattering the most and I guess we'll next move on to specialized mini models next?

Opus 5.5 is addicting like Opus 4.8 when it launched and the fact that it's dirt cheap in comparison to Fable etc is just 🧑‍🍳 💋

Facebook marketplace is a fun place 😂

Astra might be solving fancy math problems but Opus 5.5 is absolutely crushing it on practical coding and problem solving tasks in my experience

Opus 5.5 should've been Opus 6

i dont know how to reconcile Astra solving milenium problems and also struggling with basic infra for days on my journaling app

Hey @github why isn't there a "Merge when checks pass" type of check?

Making your journal feel a little more yours - starting with @loundai's wardrobe :)