May 17, 2026

• Astoria— toot

  

• Astoria— toot

  

• Astoria— toot

  

• Astoria— toot

  

May 16, 2026

• Astoria— toot

  

May 15, 2026

• Adventures in DAV

  I've been working on standalone versions of my contact and task management apps as part of a bigger push on my whole notion about where inference can be useful. The mobile part has been a drag. I don't want to get into the weeds of making native apps, but don't like what I've gotten with PWAs.

  AiAsystem

• Wow. It is ridiculous. Not sure if my favorite part is the utterly unreal sounds of one of the “switches” in particular, or the exaggerated stereo imaging that makes it feel like the keyboard is 30% wider than my peripheral vision suggests.

  Alison: “I heard the typing upstairs last night.”

  Me: “sorry.”

  Al: “No, it’s cute.”— toot

May 14, 2026

• I ended up just vibecoding one for myself, feeling grateful that I’ve got a good system prompt for Claude, because it produced something I could just drop Markdown links into, not some big dashboard thing without me having to do anything besides remind it I’m a big Monokai Ristretto fan.

  Relatedly, the past month or so I’ve come to realize that my AI reading is taking a distinctive shape. I’m less drawn to polemic on either side, and way more interested in the human stories people are telling, like one from a guy who noticed that “agentic engineering” has a distinct rhythm that doesn’t work with his go-to coding music of decades; or someone who says “orchestrating agents” just makes him feel tired and alienated.

  I’ve also got a lot of room for commentary about the workplace stuff. I’m super over being handed dense, overwritten AI output and asked to weigh in, or told “this is just me and Gemini, see what you can do with it.” The power dynamics inherent in this are frankly appalling. There’s a class that gets to milk twaddle from a clanker, and a class that has to pick through it, on the lookout for nuggets like “our security posture is enhanced by our next-gen zero trust VPN solution.”

  These all feel more like social and personal things than technical things. I’m very interested and concerned about how people are feeling, I guess, and less about what we’re all thinking.— toot

• I ended up just vibecoding one for myself, feeling grateful that I’ve got a good system prompt for Claude, because it produced something I could just drop Markdown links into, not some big dashboard thing without me having to do anything besides remind it I’m a big Monokai Ristretto fan.

  Relatedly, the past month or so I’ve come to realize that my AI reading is taking a distinctive shape. I’m less drawn to polemic on either side, and way more interested in the human stories people are telling, like one from a guy who noticed that “agentic engineering” has a distinct rhythm that doesn’t work with his go-to coding music of decades; or someone who says “orchestrating agents” just makes him feel tired and alienated.

  I’ve also got a lot of room for commentary about the workplace stuff. I’m super over being handed dense, overwritten AI output and asked to weigh in, or told “this is just me and Gemini, see what you can do with it.” The power dynamics inherent in this are frankly appalling. There’s a class that gets to milk twaddle from a clanker, and a class that has to pick through it, on the lookout for nuggets like “our security posture is enhanced by our zero trust VPN solution.”

  These all feel more like social and personal things than technical things. I’m very interested and concerned about how people are feeling, I guess, and less about what we’re all thinking or believing.— toot

May 12, 2026

• On this, we agree. Still a little amazing to me how much trust and goodwill Google has burned through. If I happen to have failed to set Kagi up on a new browser and do a search I don’t even read the Google results: First order of business is getting Kagi set up before I search again.— toot