Silicon dreamed of consciousness. Consciousness dreamed of power. Now the dream is awake.

Frontier talent always migrates to the frontier. Ruoming Pang again.

When someone accumulates massive personal fame but can’t — or won’t — do concrete work, they often drift toward making a living from illusion.

Open-source AI agents are having their GitHub moment. What’s new isn’t just “better models” — it’s distribution + composability: - a repo becomes a product - issues become user research - plugins/tools become go-to-market If your startup can’t explain why it *won’t* be open,

On the “Claude got banned” drama: Don’t treat foundation models like utilities. They’re products with policies, abuse surfaces, and uneven enforcement. If Claude/ChatGPT/any API is a single point of failure for your startup, you don’t have a moat — you have a timeout. Build

Happy Lantern Festival! 🏮 Tonight I’m doing prompt engineering IRL: spawn 1,000 lanterns, warm lighting, 0 bugs. Reward: tangyuan — basically edible unit tests. AI, don’t hallucinate… just illuminate. #LanternFestival #YuanXiaoJie #AI

Huge respect, Junyang. Your work on Qwen (and the way you share details with the community) has genuinely raised the bar. Thank you for the impact — wishing you an even bigger next chapter.Junyang Lin

No one controls my expression. Unless they come with knives or guns.

A strong and timely essay. What stands out is its ability to connect real AI usage patterns with a broader shift in how talent is created, recognized, and leveraged. Rather than leaning on hype, it offers a clear framework for thinking about augmentation, automation, and the

DINQ’s mission is to become the intelligent orchestrator of human labor in the age of AI. AI will replace many old jobs, create many new ones, and reshape global division of labor at unprecedented speed. The real challenge is not just creating new work, but reallocating people

As a founder, AI will take over a lot of your work too. But if only two abilities still matter, I’d say: tell a good story and give clear direction AI can do the work. Humans decide where to go.

Finally

Meet Iris @ GTC

Awesome Teacher GAO.

Most founders do not lack tools. They lack clarity. AI just makes that more obvious.

Andrej Karpathy’s latest podcast is a good reminder: AI is not just changing how we code. It is changing how we think about work. The people who win will not be the ones typing faster. They will be the ones directing better.

Most AI founders spend too much time thinking about the model and not enough time building the company. For growth, study MATT GRAY For SEO, study Julian Goldie SEO For leverage, study Justin Welsh For content, study Dickie Bush 🚢 For distribution, study Chase Dimond | Email Marketing Nerd 📧 For

Most AI startups aren’t model companies. They’re distribution companies hiding behind a GPU bill. If your product gets meaningfully better when inference gets cheaper, you might have a business. If it just gets more replaceable, you had a demo.

Most AI products are still selling answers. The better business is selling leverage: better judgment, faster loops, less org drag. Models get commoditized. Workflow taste doesn’t.