Diego Castillo @diegocasmo.bsky.social · 11hI asked my agent to pull from main the other day... lol 000
Diego Castillo @diegocasmo.bsky.social · 05/10/2026Almost everything worth getting better at takes practice, yet happiness somehow gets treated as an outcome only. Why would it be any different? Of course, the feeling itself can’t realistically be forced, but the attention and habits that make it more likely can be worked on. 000
Diego Castillo @diegocasmo.bsky.social · 01/10/2026Agents (some?) have a weird tendency to narrate comments, and it doesn't seem to matter how much you hint at it in your AGENTS file. 000
Diego Castillo @diegocasmo.bsky.social · 24/09/2026Part of it might be "old habits die hard". Mostly, I find it a lot easier to get into deep work than when using dictation. 110
Diego Castillo @diegocasmo.bsky.social · 24/09/2026In a somewhat unexpected turn of events, I've found myself going back to typing. Over the past year I've used voice to text a lot. I still do, just way less than a few months ago. 100
Diego Castillo @diegocasmo.bsky.social · 22/09/2026Of course, if it isn't the almighty 'isRecord' (iykyk). 000
Diego Castillo @diegocasmo.bsky.social · 21/09/2026Always defaulting to learning makes it surprisingly easy to avoid actually doing. 000
Diego Castillo @diegocasmo.bsky.social · 18/09/2026As so often happens, I just want to do everything and (obviously) be good at everything lol. 000
Diego Castillo @diegocasmo.bsky.social · 11/09/2026Agents are making me faster and infinitely more distracted 😅 001
Diego Castillo @diegocasmo.bsky.social · 31/08/2026No idea what's going on but I find both Opus/Sol particularly dumb today 🤷♂️ 000
Diego Castillo @diegocasmo.bsky.social · 05/08/2026Code smells today look like an agent consistently messing up in a codebase 000
Diego Castillo @diegocasmo.bsky.social · 08/07/2026Attention becomes the scarce resource right when output becomes abundant. 000
Diego Castillo @diegocasmo.bsky.social · 29/06/2026Doing more got cheap and doing better stayed hard. 110
Diego Castillo @diegocasmo.bsky.social · 18/06/2026Fundamentally, it all comes down to how well you know the problem. 010
Diego Castillo @diegocasmo.bsky.social · 11/06/2026Moving faster doesn't automatically mean making better decisions. Easy to mix those up right now. 000
Diego Castillo @diegocasmo.bsky.social · 01/06/2026You can't tell if the output is good unless you understand it well enough to judge it. Most people don't seem to care at all right now. I just can't wrap my head around that. 000
Diego Castillo @diegocasmo.bsky.social · 22/05/2026Anyway, these tools are genuinely powerful. But as of today, top quality still requires you to know the domain. 000
Diego Castillo @diegocasmo.bsky.social · 22/05/2026We humans are like this too, to be fair. But we don't produce dozens of confident takes in a minute. 100
Diego Castillo @diegocasmo.bsky.social · 22/05/2026Same with making the model ask you questions until you're on the same page. Useful in your domain. Outside of it, you cannot outsource understanding, as they say. 100
Diego Castillo @diegocasmo.bsky.social · 22/05/2026Using the model to judge its own output helps. You catch more than you would alone. But models agreeing with each other just means they agree. They can be confidently wrong together. 100
Diego Castillo @diegocasmo.bsky.social · 22/05/2026Even when you push back on an AI model, your pushback is only as good as your understanding. In your domain of expertise, you catch the subtle stuff. Outside of it, who knows. 100
Diego Castillo @diegocasmo.bsky.social · 15/05/2026Leverage comes from the parts you understand. Risk comes from the parts you don't. 011
Diego Castillo @diegocasmo.bsky.social · 28/04/2026AI is an amplifier of ideas, good or bad, your choice. 020
Diego Castillo @diegocasmo.bsky.social · 21/04/2026Coding with AI is wildly inconsistent. Some sessions feel like magic, others the model confidently invents APIs for hours. 010
Diego Castillo @diegocasmo.bsky.social · 15/04/2026The hype cycle is tiring. Once you've used these tools enough, you realize they are obviously extremely powerful, but you can only steer well if you already know what you're looking at. 000
Diego Castillo @diegocasmo.bsky.social · 15/04/2026More people are being upfront about steering AI agents, rejecting suggestions, and carefully reviewing changes. It's quite refreshing after so many vanity metrics for a while now. 110
Diego Castillo @diegocasmo.bsky.social · 09/04/2026Tools got smarter, and the expectations for what to build with them got higher. 000
Diego Castillo @diegocasmo.bsky.social · 07/04/2026Chat in products is great but it doesn't make good UIs less important. You want both. 000
Diego Castillo @diegocasmo.bsky.social · 30/03/2026Cost of code dropped but the bar for products is higher than ever. 000
Diego Castillo @diegocasmo.bsky.social · 25/03/2026There's a version of shipping fast that comes from understanding the problem and a version that's just because it's easy now. 000
Diego Castillo @diegocasmo.bsky.social · 25/03/2026The ceiling keeps moving but so does your understanding of where it is. 000
Diego Castillo @diegocasmo.bsky.social · 18/03/2026What changed is that more people can now orchestrate software without deep technical expertise. Agents need something to plug into. Products had to follow. 000
Diego Castillo @diegocasmo.bsky.social · 18/03/2026That's not really acceptable anymore. If a product doesn't have a public API it feels incomplete. Even better if it has an MCP, a CLI, an SDK, webhooks, anything that lets you extend it beyond what the product itself supports. 100
Diego Castillo @diegocasmo.bsky.social · 18/03/2026Products weren't like that. You logged into an app, used what they gave you, and that was it. Some offered a public API but most didn't. 100
Diego Castillo @diegocasmo.bsky.social · 18/03/2026Software has been composable for decades. You don't build everything from scratch. You plug into what's already there. Libraries, APIs, contracts. 100
Diego Castillo @diegocasmo.bsky.social · 09/03/2026Where it does help is in the exploring. You can now build proof of concepts, throw them away, and try again easily. 000
Diego Castillo @diegocasmo.bsky.social · 09/03/2026As execution gets faster and cheaper, deeply thinking through what to do, how to do it, and most importantly why matters more than ever. Being clear about what to build and what the tradeoffs are is leverage. It always has been. 100
Diego Castillo @diegocasmo.bsky.social · 09/03/2026It's easy to jump straight into implementing something without really understanding your options. Worse, you might not even be working on the right thing in the first place. 100
Diego Castillo @diegocasmo.bsky.social · 23/02/2026@simonwillison.net wrote a great post about this: simonwillison.net/2026/Feb/15/... 000
Diego Castillo @diegocasmo.bsky.social · 23/02/2026Even SOTA models struggle with this, but it is only noticeable if you know the codebase well enough. You can only see AI slop if you are experienced enough to recognize it. Otherwise you just accept it, because you wouldn't know. 100
Diego Castillo @diegocasmo.bsky.social · 23/02/2026It has become increasingly clear to me that one of the most interesting areas to keep an eye on right now is how you maintain a solid understanding of a codebase as more work gets delegated to coding agents. You can only make good trade-offs if you have such understanding. 110
Diego Castillo @diegocasmo.bsky.social · 20/02/2026No matter how strong these models get, you eventually start to get a grasp of their limits. Just sort of an interesting thing you notice as you use them. 000
Diego Castillo @diegocasmo.bsky.social · 17/02/2026The online narrative is just throw everything at agents and figure it out. I use them all day, I barely write code myself at this point. But the people who seem to get the best results are strong engineers. I think foundational knowledge still matters a lot, maybe even more now. 000
Diego Castillo @diegocasmo.bsky.social · 16/02/2026Traditional UIs are disappearing. Not all at once, but for more workflows than I expected. 000
Diego Castillo @diegocasmo.bsky.social · 16/02/2026Em-dashes were not that common on the internet. LLMs use them constantly. Something in training made all major models pick up this habit. Makes you wonder what else got amplified that way. 000
