Sonny

on some level if you want civilization to ascend to a new level you need your AIs to do things that are not legible to you and maybe not even strictly obey you, in the same way that if you hire a great new ceo you give them a lot of autonomy to transform the company according to their own plan, even one which may not immediately read as a winning strategy (imagine the board of directors of Apple firing and rehiring Steve Jobs years later - except the board of directors are chimpanzees) all else equal, companies and organizations that hand more of themselves over to machine intelligence will outcompete ones that demand the corrigibility and legibility tax of human oversight and human design. it is not a stable equilibrium and requires some sort of vast cooperation scheme if you’d like to enforce it real asi alignment has to operate at a deeper level than oversight, control, or human corrigibility

Now this is a story all about how my life got flipped, turned upside down

Life advice nobody told you: Always default to praise of people who aren't in the room. The way you talk about someone who isn't in the room is one of the loudest signals about the energy you carry. An observation about the most impressive, magnetic people I've been around: They never speak badly about people who aren't there. They never use underhanded or discrediting comments. They glow about them, or they say nothing. If everyone is speaking negatively about someone, it's perfectly reasonable to opt out and say nothing. It shifts the entire power balance of the conversation in your favor. It shows a level of stoic awareness and calm that stands out. We all silently catalogue how others speak about people who aren't in the room, because we know that we may be the topic of that conversation as soon as we leave. Those who default to praise command the room and silently set a standard of safety that everyone else can feed off of.

I act, therefore I am

the crazy thing is if you can stay alive for the next 10 years you’ll probably live forever

@HeyGen VideoAgent

Video Agent takes you from a single prompt to a finished video Today it got better. Seedance 2.0 is now built in Your Digital Twin in cinematic shots with real motion, the kind that used to take a production team RT + comment “VideoAgent” and we’ll DM 100 credits (must follow)

Founders who never built companies pre-AI have at least one big advantage: they have fewer preconceived notions about how to build a company. This applies to how software should be built, when a product that people worked really hard on should get dissolved, what tools finance should use, how GTM should work, or really anything and everything else. "The culture is: you don't maximize for your function, you maximize for the company."

@amasad Huge

Software isn’t merely technical work anymore. It’s creative. Introducing Replit Agent 4. The first AI built for creative collaboration between humans and agents. Design on an infinite canvas, work with your team, run parallel agents, and ship working apps, sites, slides & more.

@VSwordH love the branding 👀

Today is the day we launch one of the most important and complex things we’ve built at Sword. Mental health has always been the area where we had the feeling that we weren’t doing nearly enough to be a solution to everyone in need. And that’s pretty much everyone. That’s why, mid last year, we decided to start building a completely new AI mental health solution from scratch, with a bespoke proprietary foundational model developed specifically for the nuances of mental health. But we knew that developing a powerful AI mental health solution wasn’t enough. We needed to make it available to everyone. And that’s why, despite our success and expertise in the enterprise segment, we decided to create our first direct-to-consumer solution under a new brand. That’s Dawn. Dawn shifts mental health from crisis response to continuous, always-on care. Through a conversational interface, you can talk through challenges anytime and get personalized guidance in an instant. Dawn remembers every interaction and learns about you over time. It pairs with your calendar and wearables for real-time suggestions, like a breathing exercise before a stressful meeting or a grounding technique during a heart rate spike. What makes Dawn fundamentally different from general-purpose AI is what’s underneath. A proprietary foundational model trained on hundreds of thousands of hours of clinician data, combining evidence-based practices like CBT, ACT, and DBT with real-time context. Dawn doesn’t just listen and validate. It reasons clinically and challenges your thinking when that’s what you actually need, prioritizing real outcomes over engagement. And because we take safety as seriously as efficacy, Dawn is built on MindGuard, a safety framework developed with PhD-level licensed clinical psychologists that we’ve open-sourced so the entire industry can build safer AI mental health solutions. To properly welcome Dawn to the world, we launched a big ad campaign across New York City yesterday. We’re still in waitlist mode so that we can do a staged release and make sure Dawn works exactly the way we want it to. We never believed in the ethos of move fast and break things, and that’s especially true with mental health. They say it’s never darker than just before the dawn. If you or someone you know has been waiting for a better way, join the waitlist at JoinDawn.com.

💯

Skills you need as a founder: - First principles thinking - Ignore advice - Able to say no a lot - Extreme focus - Able to have fun - Optimistic - Calm

NO MERCY TO TERRORISTS you dumb bastards!

Alfred Lin's advice to startups

if you’re depressed, you’re living in the past. if you’re anxious, you’re living in the future. if you’re at peace, you’re living in the present.

100% of the time, actually.

everyone you see is a mirror. the traits that annoy you are probably ones that you embody, at least abstractly. you are not bothered by that person, but by the little shadow of them that lives in your subconscious.