Get live statistics and analysis of Mihnea de Vries's profile on X / Twitter

I hear ppl complaining about this phrase while completely missing the point. “Just build something people want” is not advice. It’s a destination. The advice is: figure out how to get there.

Reddit approach to not get banned: You gotta contribute first. Show up, be helpful, drop comments over time so people start recognizing you. Like, “Oh yeah, I’ve seen this dude around, he’s chill.” Then when the time comes, don’t drop links. Just casually mention you’re building something that solves [insert pain point]. If it’s interesting enough, people will as, or check your profile (where you have the link) Let them pull the link from you not the other way around. That’s how you play the long game on Reddit.

Day 12 of my “build an app in 30-day” challenge. Spent all day tweaking the system prompt. It’s better, but still not there. Close but no cigar. Half of the outputs feel solid. The other half? Meh. Unusable. And now I’m wondering, Is this the ceiling of the model? Or just the ceiling of my prompt? Upgrading to a better model would help, but it’s way too expensive to run at scale. So I keep tweaking. Tuning. Hoping it’ll click. Weekend grind mode: ON. Still building. Still believing. Hello from Vietnam to all the indie hackers out there.

is there a way to disable this “premium feature” i find it very distracting.

Everyone thinks: “Just vibe code with AI to $10k MRR.” Month 1, 2, 3 hype. Month 6, quit. 99% have no clue about systems, architecture, security, or how to keep a prod app alive. They’ll fail and vanish AI coding isn’t dying, but this gold rush is. prove me wrong

Jobs said people will want it after they see it. That’s still building what people want, just before they know it. The want is real, they just can’t name it yet. Same rule, only earlier in the process.

Why non-technical vibe coders fail often? they never heard of atomic commits and hard roll backs They don’t prompt as devs do, using that “coder logic” Coders see everything as a system, a function, or a potential bug which is basically the core skill in prompt + context engineering

one mind. one machine. one mission. the rest is AI. code. market. support. autonomous. relentless. no team. no funding. no permission. just signal. no noise. makers os built in the shadows. dropping soon. follow to jack in.

Serious be How long did I ignore Claude? Thought it was some fancy french journaling app. Didn’t even open it for months. Turns out, I missed out on way more than poetry and croissants.

Used a now-dead viral hack on X. Impressions jumped from 1,000 → 450,000 in 2 days, yes about 500x ! You’d think that kind of reach would explode your followers, right? Wrong. Follows barely double. 👇 Here’s what I learned from chasing virality: Even when I hit half a million impressions in a day, the follow count didn’t scale. I went from a couple a day… to 17. That’s it. Meanwhile, engagement dropped, and people scrolled on without caring. A few days later, I stopped chasing viral posts. Just went back to writing a few thoughtful posts and doing 20-30 solid replies per day. And guess what? I was back to 5- 8real followers daily, same as before, but with way less chaos. Here’s the punchline: Viral impressions are passive and bringing the wrong crowd if they follow real follows are active trust. Big difference. Virality looks like growth. But real growth? It’s built through consistent signal and actual connection. Post by post. Reply by reply. I’m done chasing spikes. Give me the slow burn. That’s what compounds.

Everyone’s freaking out about AI stealing jobs. Cool. But the real threat? Not layoffs. think someone with $50K-$100k, buys a CRISPR kit, and Grok 4 Heavy and other LLMs as personal PhD doing real damage in a garage. Not theory. Possible reality. Grok 4 Heavy and maybe Kimi can explain gene editing better than a PhD. It’ll walk you through everything, design, protocol, troubleshooting, sourcing. You don’t need credentials. Just grit. The gear? Buy it online. No (many) questions asked. Biohazards? One error away. But sure, let’s keep arguing about AI taking accounts jobs. Meanwhile, anyone with motive can now weaponize biology with AI as their co-pilot. We’re not ready. And no one’s paying attention.

Every 1,000 organic impressions saves you $5–6 in ad spend. And that same 1,000 impressions can earn you $0.50 in creator rev. Post like a creator. Leverage like a founder.

Product designers had to go through devs to make anything real. It wasn’t a frictionless process. Now? Not so much. With vibe coding, those same idea people can build their own stuff. Fewer middlemen. Faster feedback. Real results.

Everyone’s like “switching from Cursor to Cloud Code.” Cool, but… why not both? Cursor still wins for quick edits, inline refactors, and version control that doesn’t need you to git blindly in terminal. Cloud Code can run pushes, sure, at the cost of tokens and its slow. Cursor’s got autocomplete. Cloud Code? Use the right tool for the job. Stack ’em. Whats your flow?

there are 10 kinds of people here, the one who understands binary and the ones who don’t.

i built and app without thinking of a domain yet, is’t is silly? Now i have to find a .com becouse anything else is ngmi :-)

I hear ppl complaining about this phrase while completely missing the point. “Just build something people want” is not advice. It’s a destination. The advice is: figure out how to get there.

Used a now-dead viral hack on X. Impressions jumped from 1,000 → 450,000 in 2 days, yes about 500x ! You’d think that kind of reach would explode your followers, right? Wrong. Follows barely double. 👇 Here’s what I learned from chasing virality: Even when I hit half a million impressions in a day, the follow count didn’t scale. I went from a couple a day… to 17. That’s it. Meanwhile, engagement dropped, and people scrolled on without caring. A few days later, I stopped chasing viral posts. Just went back to writing a few thoughtful posts and doing 20-30 solid replies per day. And guess what? I was back to 5- 8real followers daily, same as before, but with way less chaos. Here’s the punchline: Viral impressions are passive and bringing the wrong crowd if they follow real follows are active trust. Big difference. Virality looks like growth. But real growth? It’s built through consistent signal and actual connection. Post by post. Reply by reply. I’m done chasing spikes. Give me the slow burn. That’s what compounds.

Everyone thinks: “Just vibe code with AI to $10k MRR.” Month 1, 2, 3 hype. Month 6, quit. 99% have no clue about systems, architecture, security, or how to keep a prod app alive. They’ll fail and vanish AI coding isn’t dying, but this gold rush is. prove me wrong

The Accelerator path: Idea → MVP → Marketing → Feedback → Launch The Indie hacker way: Idea → Stripe link → Proof → Feedback → PH → Cold email → $10K MRR One is slow burn. The other’s rapid fire. which one you cookin w today?

Serious be How long did I ignore Claude? Thought it was some fancy french journaling app. Didn’t even open it for months. Turns out, I missed out on way more than poetry and croissants.

If you want to bend reality, you need to rewrite the laws you live by. Stop quoting common limits like facts. Start treating them like experiments. The world doesn’t need more people following rules. It needs more people questioning them.

Most startups copy. The strongest ones dont. They go from zero to one, not from one to n. Compete and die you will, Monopoly you must become. Hidden truth? The greatest businesses look boring… until they own the world. Read Zero to One again. You missed the point.

App idea: every time Claude the Coder says “you’re absolutely right” i want my laptop to screenshot it and automatically send it to my wife

Reddit approach to not get banned: You gotta contribute first. Show up, be helpful, drop comments over time so people start recognizing you. Like, “Oh yeah, I’ve seen this dude around, he’s chill.” Then when the time comes, don’t drop links. Just casually mention you’re building something that solves [insert pain point]. If it’s interesting enough, people will as, or check your profile (where you have the link) Let them pull the link from you not the other way around. That’s how you play the long game on Reddit.

i built and app without thinking of a domain yet, is’t is silly? Now i have to find a .com becouse anything else is ngmi :-)

If you cant measure ROI, it’s not an investment. It’s a donation. Use tools that show you clear metrics. Use channels with trackable growth. Without numbers, you're not testing. You're guessing. Thats a terrible way to build.

ChatGPT, Claude, and Grok walk into a bar. Bartender goes, “Is this a joke?” "Yes, on us!" They hit the booth like corrupted JSON, spilling secrets, ready to roast the freakin’ userbase like it's Stack Overflow in 2014.

“Stay consistent” is NOT the worst advice. But it’s close. Without a reason, without feedback, without a goal, it becomes a ritual. Like jogging in place and wondering why you're not at the beach. It’s not chaos. but it is a waste of effort.🧵

Most SaaS founders start with something similar with this: ”Bro, i just discovered n8n, let’s build any SaaS and print" Wrong direction. Here’s why starting with a real personal problem leads to better products, and fewer wasted months. A thread:🧵

Something dreadful is coming. Not in 10 years. Not even 5. You’ve got 1000 days until the most valuable thing you own, your skills, your status, becomes obsolete. And no one’s ready for it. Here’s is how to navigate trough the AI disruption. 🧵

Everyone says “validate your idea” Been in this scene for months. Big communities. I post. I help if asked. Only two people ever sent me a Google Form or anything asking for feedback. So be honest: Are you actually validating? Or just yapping about it? 👇 Prove me wrong