Harry Y

I spent three hours asking Claude for startup ideas. Here's what I actually learned: LLMs are red ocean generators. Every token they output is sampled from the highest-probability options — which is just another way of saying "what most people on the internet say." Fluent = consensus = red ocean. So when an AI gives you "startup advice," what it's really doing is taking the average of every founder who's ever written about startups, and handing it back to you. But startups are fundamentally an anti-average game. If you want a blue ocean, you can't ask a consensus machine. The real opportunities are, by definition, not in the training data — because anyone who saw them went and built them, instead of writing about them. What AI is actually good for: judging whether the idea you came up with is already in a red ocean. It will mislead you on generation. It's reliable on discrimination. Stop asking AI what to build. Ask it what's already crowded.

ASO is the new SEO. His app makes $50K every month without running ads. These 14 mins might change your life:

@Bhavani_00007 Heard a rumor that complimenting and thanking it makes it smarter. Constantly yelling at it makes it dumber. Can't tell if that's true or if I've just started treating my AI like a person.

as a developer, have you ever said "Thank you" to Claude ?

@hiarun02 Claude actually told me to stop using Claude — said my work is people-heavy enough that I'd be better off just talking to humans. But I can't bring myself to waste the subscription. So I keep finding it things to build.

> NO Claude > NO Codex > NO Cursor > NO Copilot Just skill. Can you still ship?

@NoahKingJr I've been asking Claude to write prompts for me across different tools — not just ChatGPT but Kling, Seedance, Veo, Nano Banana, you name it. Rarely works. Mostly just burns credits.

Hey Claude, help me write a prompt for ChatGPT.

@DBirker78883 @petergyang Haha yeah, from my experience with Openclaw, the agents do genuinely improve over time based on how you work with them — but rarely in the direction you actually wanted. I end up spending more time tweaking prompts and md files than actually using the agent to get work done.

@Curiousyan08 @petergyang It self improves on its own when doing the same tasks.

I caved and downloaded Hermes to try. For those of you who have tried both Hermes and OpenClaw what difference do you notice? No shilling please, just want some honest opinions

@aryanlabde I always start by putting myself in the user's shoes — trying to figure out how they'd even discover my product. But halfway through that exercise I usually realize my product just isn't that appealing. Is this what a loser thinks like? lol

Most vibe coders have a GREAT product but a terrible distribution strategy.

@koderayush Just to say hello

X is cool but it gets 10x better when your timeline is full of people who actually build things coding • startups • AI • SaaS • design • products If you're in that zone, say hi 👋 let’s grow together

@DBirker78883 @petergyang Just started using it. Heard it can self improve but still have not figure out how

@petergyang Hermes is my go to. The fact that it can self improve and recall is amazing.

Ironically, AI raising everyone's productivity might be pushing us back to a fundamentally human era. When anyone can build anything overnight, the bottleneck is no longer execution — it's knowing the right people, understanding real problems, and having the connections to turn a solution into revenue. Network and relationships are becoming the ‘new’ moat.

If AI gives everyone the same output, what makes yours different?

@yashhq_22 I would love to learn from the guys who build their audience first because it matters 10x more

As a solo founder, what do you build first? - the product - the audience

There's another layer to this problem that nobody's talking about: the more you build, the deeper the self-doubt gets. With no real user feedback, you have no signal. Nothing telling you which product is actually worth committing to. So you keep moving, not because you're optimistic, but because staying feels worse than starting over. You end up with a pile of half-cooked products and a growing suspicion that maybe none of them will ever work — but you still can't tell which one deserved more time. It's not just a marketing problem. It's a feedback loop problem.

Talked to a vibe coder guy on Reddit today about his app. He launched his 9th app this year, but no customers. He says he enjoys building the apps so much because he is using AI and can build so much faster. But when he needs to get users, feel discouraged and does the next best thing, starts another app only to end up in the same place. What should he do ? Keep building until he hits a viral app that markets itself? or just continue with one app for 12 months and spend 9hrs a day marketing it

Spirit died overnight. 36,000 people on TikTok are crowdfunding to buy it back. Everyone clowned Spirit for years. Then the $39 cross-state flight disappeared and suddenly it hit — that ticket cost less than a bus ride. Wall Street looked at the books and walked. Regular people rushed in. They're not saving an airline. They're saving the idea that broke people can still fly. techcrunch.com/2026/05/03/wel…

Do you remember when you joined X? I do! #MyXAnniversary

ANTHROPIC刚刚发布了使用 Claude Code 构建公司的官方指南。 CEO：1 名人类。 员工：AI 智能体。 运营：全自动。 零员工公司不再是一个笑话。

Dumb but fun

Claude Security is now in public beta for Claude Enterprise customers. Claude scans your codebase for vulnerabilities, validates each finding to cut false positives, and suggests patches you can review and approve.

Today, we’re launching the @link wallet for agents. It lets you securely empower agents to spend on your behalf. Your payment credentials are never exposed and you approve every purchase. link.com/agents

Rip my last bit of energy

elon left OpenAI in 2018. built a whole competing AI company. mass-acquired twitter. runs 5 other companies. and he's still in court because a "non-profit" started making money. the man has infinite resources and zero ability to let go.