Kai

I want to start posting on X again. I hope that X will give me the chance to rebuild the algorithm for an account that is suddenly inactive for 2 months.

This year was a pivotal moment for my life. I can't believe I been through a lot. Many things has changed for me. I am no longer the same. An accident simply rewired my brain. Now it's different. It felt like I moved on from the old to new. This feeling is so surreal.

What if I tell you Claude Mythos leads to Project Glasswing, leads to 40 vetted companies early access leads to DeFi biggest Aave biggest DeFi hacking?

This Claude Code project setup guide literally makes Claude Code 10x better🤯

just a week ago, I was one step away from hell gate. I'm just genuinely grateful that I'm still able to tweet here.

one honest question to close the week: what's the thing you kept putting off this week that you actually know needs to happen? mine: having a conversation I've been drafting in my head for two weeks. (it's always the human thing, not the work thing.) your turn. real answers only. I'll reply to every one.

the year I nearly stopped trying, I was working the hardest I ever had. that's the part nobody talks about — the collapse that happens not from laziness but from grinding on the wrong thing, in the wrong direction, with no one watching and no proof it was leading anywhere. (depression doesn't always look like doing nothing. sometimes it looks like doing everything and feeling none of it.) I came home that year defeated in a way I hadn't been before. not beaten by someone else. beaten by the version of myself that kept pushing when I should have pivoted. the lesson I pulled from the fire: effort without direction isn't discipline. it's just expensive confusion. pivot before the cost gets too high. if you're in a season like this right now I'm still in the rebuild, documenting all of it. follow along. you're not the only one.

friday night. everyone posting wins. everyone looks fine. some of us are still figuring it out and that's okay. the feed is a highlight reel, not a leaderboard. you're not losing. you're just not posting your drafts. if you had a real week, not a curated one. I want to hear it. drop the actual update below.

I thought wearing five hats at a startup would teach me efficiency. it taught me something more useful: which tasks drain you and which ones are just draining because you're doing them manually. the ones you dread doing are usually the best candidates for an agentic system. not because AI does them better sometimes it doesn't But because the setup forces you to think clearly about what good output actually looks like. (you can't automate something you haven't defined.) map the tasks you avoid. that's your workflow backlog. build the boring one first. it buys you back the hours for the rest. what's the task you keep pushing to tomorrow? drop it below! let's figure out if it can be systematised or just needs to get done.

AI fluency is not a competitive edge anymore. it's the floor. the edge is knowing what to build with it and having the judgment to know when the output is wrong. the people who win with AI aren't the fastest prompt writers. they're the ones who know their domain well enough to catch the mistakes.

21, 22, 23, 24, 25, 26, 27... and still figuring it out. not a crisis. a late start. and a late start with a system beats an early start with nothing but speed.

the AI tool that changed how I work isn't Claude or ChatGPT. it's the system that tells Claude who I am before I ask it anything. the context chain — voice file, lexicon, audience profile, story bank — fed in the same order, every single time. (without it, the output is technically correct and completely wrong. with it, the output sounds like me on a good day.) most people skip the context setup and wonder why AI content sounds like AI content. they're not prompting badly. they're prompting into a vacuum. build the context once. the system carries everything after that.

I have no idea if I'm doing this right. I just know that stopping feels worse than continuing wrong. so I keep publishing, keep adjusting, keep going. (the data tells the story eventually. the feeling never does.) building in public means the uncertainty is visible. that's the whole point. if you're figuring it out too, follow along.

your consistency problem is not a discipline problem. it's a system problem. discipline fills the gap when the system breaks. if you're relying on discipline every single day, your system is too hard to run. what's the one habit you've tried to build more than twice and it still won't stick? drop it, I'll tell you what I'd change about the system, not the person.

As an avid prompter, it’s safe to say, Prompting is not going away. User context and taste is something that is going to be embeded in your prompt (not someone else) I am working on a project which I believe will benefit billions of users. I am current ideating and mapping the product on how I would use it on a daily basis. I am excited to share the progress as I go along Drop below what are some ideas of yours! Would love to ship them if its good!

the loneliest I ever felt wasn't at rock bottom. it was when things started working and there was no one to tell. you close something, figure something out, hit a number that mattered — and you open your phone and there's no thread to send it to. (silence after a win is louder than silence during a loss. I don't know why that is, but it is.) I was doing everything alone and calling it strength. it wasn't. it was just what I was used to. find one person to build with. not to impress, not to report to. just someone who knows what you're building and why it matters. the rebuild is survivable alone. it's better with a witness. if you're building alone right now and it's getting heavy this is the part I'd change first. one person changes everything.

what does your version of 'starting over' look like right now? mine looked like: no followers, no system, no proof it would work — just a decision to build in public and let the rebuild be the content. (still in that chapter. it's messier than I expected.) your turn. I'll reply to every honest answer. no surface-level stuff.

Hey people, I have a death confession I nearly died 2-3 days ago from a motorbike accident in which I have 0 clues how it happened I suffered a minor concussion. With left side of my body scarred with broken shoulder bone. The bike is okay. But my head felt different for a moment. Right now I am in a battle with myself. I really dk how I survived. I am glad that I am still alive. Everything else is secondary

I used to think a good prompt was a long prompt. it's not. it's a specific one. three things: one clear role, one real constraint, one example of what 'good' looks like. that's it. everything else is noise you're adding to feel like you're being thorough. the prompt structure I use daily — role / constraint / example with works for content, strategy briefs, analysis, all of it. what are you prompting for? drop it and I'll show you how I'd approach it.