Looks like the only way to change my @Cloudflare account password is via the forgot password flow 😐

My favorite @elonmusk quote that I often send friends: Do not fear losing. “You will lose,” Musk says. “It will hurt the first fifty times. When you get used to losing, you will play each game with less emotion.” You will be more fearless, take more risks.

3rd being the most important, and the most toughest imo

Hiring without a resume, but with “favourite programming agentic dev session” is such a nice idea 🤯🫡

“same problem as before but 100x harder because people are crossing domains” Being a back-end dev, I’m experiencing this so much as I’m trying to learn design just so I can write better prompts. Figma Make, Stitch, Canva AI, or whatever is pretty useless if I don’t have at least a decent understanding of design.

Found these old gems while cleaning the house. Guess it’s time to touch some grass 🍃

we as software engineers are becoming beholden to a handful of well funded corportations. while they are our "friends" now, that may change due to incentives. i'm very uncomfortable with that. i believe we need to band together as a community and create a public, free to use repository of real-world (coding) agent sessions/traces. I want small labs, startups, and tinkerers to have access to the same data the big folks currently gobble up from all of us. So we, as a community, can do what e.g. Cursor does below, and take back a little bit of control again. Who's with me? cursor.com/blog/real-time-rl…

Mario Zechner just dropped the most based take on the whole AI coding frenzy and I’m quoting the entire 2nd half of the blog in my head on repeat. Takeaways from “Thoughts on slowing the fuck down” (mostly direct quotes) Agentic search has dogshit recall. Before an agent touches any code it’s supposed to first search the entire relevant codebase. But current agentic search is broken, even if you give it a queryable index, LSP server, or vector DB. The bigger the codebase, the lower the recall. Low recall = the agent literally doesn’t see half the code it needs → it makes garbage architectural decisions that compound into unmaintainable slop. So how should you actually use agents right now? Only on scoped work where the agent doesn’t need to grok the full codebase. The loop must be closable (it can evaluate and fix its own output). The output must not be mission-critical (side projects, internal tools, experiments, all the stuff that can break without taking your product down). You are the final quality gate. Always. Let the agent do the boring, rote, “I’d never have time for this” stuff. Let it explore ideas you wouldn’t otherwise try. Then you review everything it spits out, keep the parts that are actually good, and finalize the implementation yourself. (Yeah, you can even let it do the final polish, but only after you’ve signed off.) Review every single line the clanker generates. (This is something I’m struggling with currently) Anything foundational like architecture, public APIs, core data models, security boundaries, write it by hand. Or at the very least pair-program with the agent so you stay in the code the whole time. Because slowing the fuck down and suffering a little friction is exactly what lets you learn, grow, and actually understand your own system. You’ll sleep like a baby knowing you still have an idea what the fuck is going on… and that you never surrendered your agency. Quote-tweeting this because the everyone needs to hear this, and I’d keep coming back to this in my timeline. Thanks Mario, for writing a piece “without much technical depth” but with depth nonetheless 🙇🏻

One of the most interesting challenges I got to work on during my time at FunnelStory.

you still need to read and understand code

building something where you spent a lot of time thinking about the right abstractions and nailed them is amazing it's dramatically easier to ship if you nail this the temptation is to worry about all your abstractions up front the right thing to do though actually is to optimize for learning, ship things, and then refactor after patterns emerge from hitting the same problems 10 times

So true! I now spend more time reviewing code than thinking about or designing the architecture initially. I have a rough idea of how the features are going to be implemented, as per an Implementation_plan.md file. Then Kimi2.5 writes the code, Opus reviews, and the PR gets merged. I only review code thoroughly after a couple of PRs have been merged, and then ask for changes. The loop goes on.

In the coming years, designers are going to be hugely valued. I’m a back-end engineer professionally, but now I’m obviously working on everything across mobile apps, landing pages, UI components etc etc. Everything is pretty easy with agents, but designing a good interface which people see and think “damn, I wanna try that” just seems impossible! And I’ve tried everything - Figma, pencil. dev, Stitch, and what not. And none of these are generating good designs. At this point, I’m trying to learn design so I can prompt the AI in a better way. Because right now, I don’t even know what a good design sounds like when read out loud! Colleges should probably add a small design course along with C in the first semester ngl 🥹

There was a time when I used to save every little script that I wrote or got from a blog, someone’s GitHub gist, or YouTube video tutorial, etc. That was useful once and possibly in the future. But now all that is moot