Just a reminder at how cool immersion cooling can look like. This past week at Computex I've had a couple of conversations about it, if it's good or relevant. The short answer is that even with the PFAS removed, and it being on display for 10+ years, no-one wants to buy it.
@thetreygoff related: I have found Codex to be horrendously bad at writing copy for web apps. Like it doesn't understand that the copy is made for people using the website, not me.
Podcasts are a great source of context, but they're rarely available in text form.
That's why I launched Stenobird: the fastest way for agents to transcribe podcasts. $0.01 per minute of audio.
@mweinbach A real assistant would have all the context needed if you opened the app and said "the usual". This would be faster than the app.
OpenAI's approach seems like they're trying to build an app inside a chat which is never going to beat a purpose-built app.
A lot of prolific Claude/Codex users are talking about running 3-10 sessions at a time, sometimes spread out across different projects.
The science seems pretty clear that in most cases, this leads to poor performance across all tasks.
And yet the temptation is so strong.
I think this workflow feels a lot like playing a management-style game like Roller Coaster Tycoon, Game Dev Story, Pocket Planes, etc.
someone made the most ADDICTIVE game to learn DATA CENTER networking
its called Data Center, $6 game, you start with bare floors, buy racks, mount servers, route every cable by hand
the INSANE part, every customers traffic shows as colored balls rolling through your cables... you literally see bottlenecks in real time
180 reviews in 48 hours, people with RTX 4090 rigs are HOOKED on a $6 cabling sim