Jason Liang

Claude code with remote, and react native with expo + metro is pretty awsome. Tell claude to fix something in the claude app, switch to your app and its fixed. Crazy. Can now build with Claude from your iphone easily with live result instantly.

3

Just found 5 YC startups with some of the most unique, “no one else would’ve built this” ideas I’ve seen 👇🏼 🚀 These are the kinds of concepts that make you stop and think “wait… that’s actually genius.” 1. Digipals (YC F25) Reinventing social media around real-life coordination instead of infinite scrolling. 2. Hypercubic (YC F25) Using AI to modernize COBOL and mainframes, the invisible backbone of banks, airlines, and governments. 3. Dome (YC F25) One API that connects every prediction market (Polymarket, Kalshi, etc.) into a unified trading layer. 4. Clad Labs (YC F25) A Gen-Z-native coding IDE where the coding loop blends with gaming, scrolling, and ads. 5. Nucleo (YC F25) Automated CT scan analysis for oncologists, already trusted by top hospitals. 👉 YC’s most innovative companies are often the ones that end up defining new categories. - - - - - - - 💬 Which one of these would you invest in purely because the idea is so wild?

33

1 Comment

Effective Agent Design Patterns in Production 🤖 Building agents is a bit complicated when you peel back the underlying implementations, but this is one of the easiest resources to learn how it all works under the hood. This is a fantastic introductory video by Laurie Voss to the core principles and workflow architectures behind agents.  1. Build an agent to make sense of large volumes of unstructured data in a semi-autonomous way.  2. An agent is powered by an underlying workflow. Some of these patterns were originally laid out by Anthropic - chaining, routing, concurrency, and more. Video: https://lnkd.in/grmT3k3p

59

90% of healthtech startups fail. Not because of bad tech. Because healthtech founders keep making the same 6 mistakes before they write a single line of code.

4

1 Comment

How do you win in the AI era? Speed? That seems to be what YC, VCs, and everyone else are preaching: launch faster, iterate faster, use the latest tools, move faster than everyone around you. Speed is supposed to be the only moat left. That might be true if you were only racing against other humans. But now you are racing against machines too — and against giant companies with massive distribution, capital, and model access. So maybe “speed as a moat” does not really mean you are building something durable. Maybe it just means you are buying yourself a small window of time — enough time to get traction, raise money, or land an exit before the wave catches up. That is what a lot of AI startups feel like today: not companies being built to last, but companies being built to be acquired. Run fast enough to make some money. Miss the window, and die on the road. So the real question is: can we still build something that lasts in the AI era? Or was that always a fairy tale from a different time? The same shift is happening with talent. Loyalty is fading. Everything feels more transactional, more short-term, more optimized for immediate upside. Get rich faster. Capture value now. Silicon Valley feels less like Silicon Valley and more like Wall Street with better branding. What has not changed is human nature. - Greed. Vanity. Status. The tools changed. The incentives did not.

2

1 Comment