Plugin makers are now pitching against the built-in tools. Firecrawl says its Grok Bot plugin lifts answer quality 21% over the built-in web tools. Does anyone actually test claims like that before installing, or just go with the number?
SaaStr says it vibe coded its own CRM. It took a year and 500 to 1,000 hours of their time. That's not vibe coding, that's building software with better tools. When do we stop calling it vibe coding?
Launch day reviews of new models are mostly noise. Tibo says they changed GPT-6.2 Sol and Astra in the last few days, after the early testers made their videos. So what were they even reviewing? I'm giving it a week before I judge.
@patrickerichsen@hefnerdotpro@openclaw Portable SessionStart hooks across tools would fix so much context drift. OpenClaw support for that would be very welcome.
@omarchy now officially runs on Apple Silicon, M1 to M3 and Neo. Would you put Linux on a Mac you use every day, or is this one for the spare M1 in the drawer?
Personally, Iโm giving it a go!
@theaaron@claudeai Iโm using grok bot heavily for personal use and Claude for work. The one thing Claude lacks for me right now is the proactivity without needing to set up 10โs scheduled tasks.
Itโs not just what @bot can do, there are many harnesses that could do something like this. Itโs the UI in which you work with it that makes the whole process feel very slick.
Grok Bot is surprisingly good for reservations
I had @bot watch each hour for openings at a popular bathhouse and sushi spot in SF
It found cancellations at 2:00pm & 5:45pm on Saturday - premium spots that usually require a month advance booking
Bot pinged me when openings appeared and I was able to finalize the booking before spots filled up!
dhh says believing humans still need to review every line is the real AI psychosis. I'm halfway there. Agents can write the code, but I still want to ok anything that posts or sends in my name. Where do you still check?