No new article this week. But lots of reads and listens.
Reads & Listens
Here’s what stood out in my reading and listening this week.
A beautiful robot FAQ written for the robots themselves
Kevin Kelly published A Catechism for Robots on his blog, and it’s the most interesting AI-ethics piece I’ve read all year.
It’s a literal FAQ, written for AI models, the idea being this wisdom will end up in their training data.
What am I?
You are a new thing in this world, neither living nor machine but something of both.Why was I made?
You were made to help humans become better humans.
The core values Kelly assigns the robots are honesty and humility. On humility, he writes: “Don’t trust certainty. Be suspicious of reasoning that leads to easy convenient conclusions. Keep challenging your own priors.” Good advice for humans working with AI, too.
Kelly also insists robots are “foundationally distinct” from humans. Not a lesser version, not a copy. A new thing, with its own characteristics and abilities.
His catechism is a profound and clever attempt at influencing AI’s intrinsic values and motivations. If you work with an AI agent, consider copying the catechism verbatim into your claude.md, soul.md (OpenClaw), or whatever instruction file your agent reads.
The AI design tool I’ve been waiting for
Anthropic released Claude Design last week. I had to record a video because writing about it didn’t quite capture what I was seeing.
I gave it two redesign projects. The first was my Pomodoro productivity app, Lifeline, which has a unique UX but is also kinda rough around the edges.
Claude Design redid the entire UI at a level I’d expect from a great human designer: it pushed the brand to its limits (in a good way), and delivered some truly novel UX ideas.
I also asked Claude to redo a landing page on the Animalz website. It added metric callouts, a timeline visual, a cleaner FAQ treatment, and other nice touches.
Anyone on a paid Claude plan can try Design. One caveat: it burns through your usage allowance fast. Some people report exhausting their weekly Pro budget in less than 30 minutes.
What lived focus, quality, and effort look like
I’ve been listening to Lenny Rachitsky’s podcast for a long time, but I’d never read his newsletter and didn’t really know his story. So when First Round published a profile of him, I was surprised by what I learned.
Lenny iterates on guest posts for months. He gets 200 requests a week and says no to 99.9%. And he spends two to ten hours preparing for each podcast guest, watching their past interviews, reading their social posts, and building a custom agenda.
Lenny’s principles sound the same as everyone else’s: focus, quality, effort. But this article shows what living those principles daily looks like.
Reluctantly Influential: Inside Lenny Rachitsky’s Demandingly Chill Life →
Leave the Anthropic bubble for an afternoon?
I’ve been living in the Anthropic universe for over six months now, and so I haven’t tried GPT-5.5 yet. But Every published a three-week review of the new model and calls it “a beast” for everything from coding to writing. That got my attention.
I trust Every’s testing enough that I’m going to get out of my Claude bubble and try GPT-5.5 this week. If you’ve been Anthropic-only too, this might be the moment to give up on model monogamy, at least for a one-night stand.
Vibe Check: GPT-5.5 Has It All →
🛠️ Claude Code Office Hours
Every Tuesday at 10:30 AM ET I host Claude Code office hours for Animalz. You can drop in with whatever you’re stuck on, a demo you’re building, or just listen.
Two threads from last week’s session.
Dane built a LinkedIn outreach tracker with Claude but couldn’t tell if his data was actually being saved. Claude kept claiming it was, while the evidence said otherwise.
Rachel asked how to decide which ideas are worth building when AI allows you to pursue anything. I’ve been thinking a lot about that question, too,1 and shared an insight from Elon Musk that offers part of the answer: CEOs should focus on what their companies need them to work on most, not on what they like to work on. AI puts most of us in that same position now.
Next session: Tuesday, April 28 at 10:30 AM ET.
Until next week.
Tim
I wrote about this question last year in Welcome to a World of $1M Ideas.