Hangyul Lyna Kim
7 followers · 17 following · 948 views
on the atlas — 265
- The Millennium Problems for Biology1 savers
- Introducing System One Models and Jev - TypeSafe AI Blog16 savers
- How to Unclench | Jonny Miller17 savers
- Mercury2 savers
- Persimmon | humans&7 savers
- On Navier–Stokes | OpenAI11 savers
- Robot-use agents5 savers
- vas on X: "Applied AI Doesn't Work" / X1 savers
- VC isn’t VC anymore — understanding the rise of Cancer Capital - Anil Dash1 savers
- Project Fetch: Phase two \ Anthropic2 savers
- Backsliding, by Meghan O’Gieblyn10 savers
- Go Big or Go Broke – The Middle Ground is Gone | LinkedIn2 savers
- Together Alive | LinkedIn1 savers
- Project Fetch: Can Claude train a robot dog? \ Anthropic1 savers
- Join Us in Building the AI Stack for Industrial Engineering at Mistral | LinkedIn1 savers
- Previewing the Model Hardware Standard \ Anthropic10 savers
- FLUX 3 x mimic: The Next Generation of Video-Action Models | Black Forest Labs1 savers
- Woz | Douglas Qian1 savers
- Atlas Motion Emerges from Stealth with $11.5M in Funding1 savers
- Cortentia1 savers
- Raising a person in a culture full of types | The Outline9 savers
- Building Worlds That Train Robots | World Labs1 savers
- Ankit Gupta on X: "Introducing Atlas Discovery: Predicting Human Response to Drugs" / X1 savers
- Hyunwoo Kim1 savers
- An OpenAI model has disproved a central conjecture in discrete geometry | OpenAI10 savers
- Ando: Building Slack from Scratch — New Ontologies7 savers
- Learn to love the Moat of Low Status - by Cate Hall5 savers
- Maybe you’re not Actually Trying - by Cate Hall3 savers
- The Nerdy Escorts Cashing In On Silicon Valley’s AI Boom1 savers
- Things you learn dating Cate Hall - by Sasha Chapin1 savers
- 2026 Q2 update - by vincent huang - a slice of my mind1 savers
- The Winning Essays for the Big Questions About AI2 savers
- Can You Just Do Things?—Asterisk16 savers
- Jeff Bezos’s Blue Origin Nears a Big Fund-Raising Round - The New York Times1 savers
- Deep-Tech.com: The Best Deeptech Funds1 savers
- imbue5 savers
- To Improve The Real World, Simulate It | LinkedIn1 savers
- [2606.25076] Machine learning is revolutionizing weather forecasting -- the next step is a change in how we work1 savers
- Zero Shot Fund | Investing in the Post-AGI World1 savers
- A Beginner's Guide to Robotics Hardware | Interlatent Blog2 savers
- Spark Climate Solutions – Accelerating progress in emerging, high-impact climate fields1 savers
- Is Anthropic Losing Goodwill With AI Researchers?; AGI House Alums Raise $25 Million For New Fund — The Information1 savers
- Meet the 22 Investors to Know in Robotics and Physical AI - Business Insider1 savers
- [2402.03300] DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models2 savers
- Long Humans3 savers
- 2025 하반기 SNAAC 활동 소개1 savers
- About TwelveLabs: Video-Native AI Company1 savers
- outlier projects1 savers
- What failure looks like - LessWrong12 savers
- MUSICA Background | Atmospheric Chemistry Observations & Modeling1 savers
- Benchmark raises its first-ever growth fund as part of $2B capital haul | TechCrunch1 savers
- Gokul Rajaram on X: "The Three Humans Left in a VC Firm Fred Wilson, co-founder, Union Square Ventures, interviewed by Michael Mignano (USV) [I post one executive summary daily of an interview I enjoyed and learnt from. I loved this interview that @mignano did with @fredwilson, who I've learnt a" / X1 savers
- What’s New in WeatherMesh-61 savers
- Reshaping a Broken Industrial Robot Market: Introducing the Era of Physical AI + Robotics as a Service with Verne1 savers
- DINOv31 savers
- DINOv31 savers
- Atlantic_meridional_overturning_circulation3 savers
- Planetary Intelligence - Will Marshall1 savers
- Bessemer Predicts: Robotics and physical AI - Bessemer Venture Partners1 savers
- VC Fund of Funds: The Market - by Pavel Prata1 savers
- Building bet-taking machines (debt vs. equity), with William Godfrey1 savers
- llm-wiki6 savers
- [1706.04599] On Calibration of Modern Neural Networks4 savers
- Introducing AIMIP: The AI weather and climate model intercomparison project | Ai21 savers
- Abrar Anwar1 savers
- Percepta | A General Catalyst Transformation Company1 savers
- Generalist - The Dark Matter of Robotics: Physical Commonsense2 savers
- The Robot Is Ready. The Factory Isn't. | LinkedIn1 savers
- 🇺🇸 The American Dream 🇺🇸 | LinkedIn1 savers
- Ben Levin | LinkedIn1 savers
- Cofounder6 savers
- Day One: Why I'm Joining Eclipse1 savers
- Learning dexterity | OpenAI1 savers
- [1910.07113] Solving Rubik's Cube with a Robot Hand1 savers
- 9. Classes — Python 3.11.3 documentation4 savers
- bookbear express | Ava | Substack14 savers
- Economics of Orbital vs Terrestrial Data Centers3 savers
- Ultra Instinct | Eric Jang5 savers
- Economics and AI | Tom Cunningham – Tom Cunningham9 savers
- Our purpose - Substrate2 savers
- Class Dismissed - Colossus2 savers
- Dexterous Robotic Foundation Models - YouTube2 savers
- Flow Matching Policy Gradients4 savers
- Deep Checks — Deep Tech Seed Investors2 savers
- Stevey's Blog Rants: Done, and Gets Things Smart4 savers
- LLM Settings | Prompt Engineering Guide2 savers
- Models | Community Earth System Model3 savers
- The implausibility of intelligence explosion | by François Chollet | Medium3 savers
- Problems for Self Study - Harvard Review2 savers
- Understanding Layout Algorithms2 savers
- in praise of uselessness - by Ava - bookbear express15 savers
- Your Smile2 savers
- The 'Don't Look Up' Thinking That Could Doom Us With AI | Time2 savers
- How does in-context learning work? A framework for understanding the differences from traditional supervised learning | SAIL Blog3 savers
- Curius / Onboarding2621 savers
- Theory of Change (Aaron Swartz's Raw Thought)83 savers
- You and Your Research77 savers
- What Should You Do with Your Life? Directions and Advice - Alexey Guzey73 savers
- Advice · Patrick Collison67 savers
- A blog post is a very long and complex search query to find fascinating people and make them route interesting stuff to your inbox66 savers
highlights — 475
Her story reminded me of a Greek term I like, techne, which means craft grounded in practice. A founder builds a particular shape of company out of their techne, all their experiences, personal craft and taste that they bring to the work.
Ando: Building Slack from Scratch — New OntologiesYou’d have to succeed here, for there was no further West to go.
Ando: Building Slack from Scratch — New OntologiesYou’ve re-entered the state of childlike wonder where you don’t have adult concepts to mediate reality, what William James called “blooming, buzzing confusion.”
Learn to love the Moat of Low Status - by Cate HallI think that really good relationships can help people develop agency, in part because they give people a lot more freedom to experiment. So, at different points in our relationship, both Sasha and I have benefited from the ability to have one partner who is the anchor — financial anchor, emotional anchor, or something else. That gives the other person more ability to roam and be less constrained. For nine months I was really unsure what I was doing professionally, and I had the ability to let myself be aimless during that period. I have likewise supported Sasha through many explorations, like…
Can You Just Do Things?—AsteriskSo psychedelics, desperation, exposure to high-agency people. I think those are the standard things. And then there is just grace. Sometimes people wake up one day and they're like, oh, I don't like the way that I'm living. And that happens. But it's less reliable for me.
Can You Just Do Things?—AsteriskMy first instinct is that you're seeing some sort of selection effect, where sociopaths tend to do both. People who tend to view others in transactional terms are also people who are high agency, in the sense that they have never bothered to learn social scripts. They are very low in conscientiousness. And so, naturally, without any study, they are able to exude high-agency instincts. A large part of learning high agency is learning not to be so constrained in your view of the world and of what comprises possible action. The people who, for whatever reason, never learn those things in the firs…
Can You Just Do Things?—AsteriskI think that LARPing as a high-agency person by following the playbook of a tech founder seems unlikely to be a true exercise of agency, and therefore is unlikely to confer the benefits of “true agency:” a meaningful life that, upon reflection, you are happy to have lived.
Can You Just Do Things?—AsteriskOther than this, we make two methodological improvements on the original horizon paper. First, the original paper defined task length as geometric mean human time conditional on success; here, we account for the fact that a lower success rate makes tasks more difficult by estimating the task duration as when human success rate data is available.
How Does Time Horizon Vary Across Domains? - METRyour attention samples the environment at a rate that exceeds the environment’s novelty generation
What 50 Milliseconds Feels Like - by SomewhereSuddenly, board meetings were no longer about judgment. They were about being productive. “If you think of your board members as working for you, there’s no reason to lie or fudge things to look a certain way or prove your competence,” he says. “If you do those things, your board can’t help you effectively.”
The Secret to Making Board Meetings Suck Lessit wasn’t out of line to give board members action items,
The Secret to Making Board Meetings Suck LessHas the market changed since we last met? If so, did it affect us negatively or positively? Has the team changed? For better or worse? Has our position in the market changed? Did we do what we said we would?
The Secret to Making Board Meetings Suck Lesssustainability drives 4-7% EBITDA uplift over the holding period, through cost reductions and new revenue. The idea that private firms are less committed appears outdated-private markets are leading on climate action and value creation.
JefferiesEcoVadis data shows 50% of US executives kept sustainability investment unchanged in 2025, with many increasing spend but not publicising it (otherwise known as “greenhushing”). Few have reduced investment, and pushbacks are seen as temporary.
JefferiesMost of us are choosing Option 2. We name them Siri, Alexa, Cortana, Samantha. Honest, Helpful, Harmless… Her. We slap cute faces on them, give them backstories, debate endlessly about the best way to finger their latent spaces. We feminize what threatens us, anthropomorphize what we don’t understand, and obsess over how to “use” what we can't admit exists.
God is a Parasite - by Jason Yuan - Letters from MarsWith personal intelligences powered by exponentially scaling model capabilities, what we gain in individual power we lose in shared reality and collective sense-making.
Gossip is Good - by Jason Yuan - Letters from MarsWhen our ancestors weren’t spilling blood, they were spilling tea.
Gossip is Good - by Jason Yuan - Letters from MarsWhat if the thing that truly makes us exponential is… each other?
Gossip is Good - by Jason Yuan - Letters from MarsMercury is designed, first and foremost, to be an experience inclusive of people with limited executive function and cognitive load.
Introducing Mercury OS - A speculative vision of the OSthe name that softens their voice,
when I say I don’t believe in marriage, I lie | by R. Khoirotun | Nov, 2025 | MediumWhen VC is at its best, it deploys its capital against world changing bets, even in the face of uncertainty.
Meditations: A Requiem for Descartes Labs | by Mark Johnson | Mediumfired in January 2020
Meditations: A Requiem for Descartes Labs | by Mark Johnson | MediumThe company was burning too much cash. The sales process was run poorly. (i.e., the process of selling the company)
Meditations: A Requiem for Descartes Labs | by Mark Johnson | MediumTrained on 200M+ observations drawn from The Multimodal Universe Collaboration et al. (2024) and five major surveys, Legacy Survey, Hyper Suprime-Cam (HSC), SDSS, DESI, and Gaia.
PolymathicIt implements a two-step process: bespoke tokenization strategies that homogenize diverse scientific data, followed by multimodal masked modeling that learns how different observations relate, encouraging the model to learn the underlying physical relationships in the data.
PolymathicShe scrupulously prepared for her interview with Thiel. She workshopped her answer for his famous question: “What important truth do very few people agree with you on?” But instead, he asked her: “What worries you?”
Cyan - by Kevin Gee and Dan Scott - Cloud ValleyBanister was so smitten with Buttigieg that she tried pitching Founders Fund, the elite VC firm founded by Peter Thiel, to open an office in South Bend.
Cyan - by Kevin Gee and Dan Scott - Cloud ValleyI love short-lived habits, and regard them as an invaluable means for getting a knowledge of many things…
Cyan - by Kevin Gee and Dan Scott - Cloud ValleyBeing bored is an insult to your intelligence -- My sister.
Cyan - by Kevin Gee and Dan Scott - Cloud ValleyAfter leaving IronPort, she founded a company called Zivity that paid photographers and models to post pin-up photography online—an OnlyFans before its time. She even posted nude photos of herself to promote the platform, earning both ire and respect.
Cyan - by Kevin Gee and Dan Scott - Cloud Valleyabout her fascination with Japanese sex clubs than she is about her success. Banister lives life on her own terms.
Cyan - by Kevin Gee and Dan Scott - Cloud ValleyThis brings us to a sheep behavior pattern: the Bellwether. In a flock, a Bellwether assumes the position as a leader. They're often the ones who first venture into new territory, who find the fresh pastures, who spot the dangers ahead. The rest of the flock follows not because they're forced to, but because following that particular sheep has proven beneficial.
Our Mascot: Lore of the Bellwether — Long Journey VenturesEach of these actions required you to be a "sheep" – to trust in the expertise and good faith of countless others. And that's the point: civilization itself is built on our ability to follow, to trust, to believe in others' expertise.
Our Mascot: Lore of the Bellwether — Long Journey VenturesThe real question isn't whether to be a sheep, but how to be a smart one. How do we choose which shepherds to follow?
Our Mascot: Lore of the Bellwether — Long Journey VenturesI’ll be honest — the moments I’m least proud of running this fund are when I’ve let my inner skeptic rule the day.
What's a Second Believer Really? — Long Journey VenturesTraining costs follow a power-law growth pattern during scaling epochs, experience sudden drops with breakthrough architectures, and then resume scaling — forming a sawtooth pattern where each reset is followed by even greater demands. In contrast, R&D compute costs rise steadily — and often exponentially — until the product reaches sufficiency (see: microwave oven) or obsolescence (see: compact disc).
The Only AI Moat is Hardware. And Compute is the Upper Bound for… | by Murat Onen | Mediumnovel topologies undoubtedly provide a moat
The Only AI Moat is Hardware. And Compute is the Upper Bound for… | by Murat Onen | MediumAnthropic's “thinking” cap. OpenAI's Dev Day token usage awards. Base's newspaper. Creative teams unconstrained by convention, driven by pride, care, and a bit of fun. As Zach said in our interview, "You don't want to compete with a team having fun."
Willem on X: "Three Weeks to Print: Base Pace in Action" / XThe firms that will thrive over the next 15 years are those that actively redefine the role of a VC — bringing not just individual expertise but the full power of their team and network to help founders succeed through every market cycle.
The Long Game: How Salesforce Ventures is Reimagining VC | Salesforce Ventures“Or you know, ‘Patrick and John have been really great for the last decade, but actually in the next decade maybe they’re not the right ones to do it.’ All this stuff. I just remember how Patrick and John were feeling. The world had loved them, and now suddenly the same people were basically saying they didn’t believe in them. There’s nothing that rallies a killer founder like that.”
The New World - ColossusToday, system complexity has exploded. It’s not just changes in one domain - it's deeply cross-functional engineering.
Announcing Flow’s $23M Series A Led by SequoiaBotQ is Figure’s dedicated manufacturing facility designed to scale robot production.
Introducing Figure 03internal development of our first-generation tactile sensor, guided by three principles: extreme durability, long-term reliability, and high-fidelity sensing.
Introducing Figure 03Each hand now integrates an embedded palm camera
Introducing Figure 03new camera architecture delivers twice the frame rate, one-quarter the latency, and a 60% wider field of view per camera
Introducing Figure 03What will it take to create AI assistants with much faster multi-modal reaction times, that have the ultra instinct?
Ultra Instinct | Eric JangDespite the AI gold rush that we are now in, it is still very non-obvious how to turn 10M USD worth of GPU-hours into 10M USD+ in incremental margin (besides something like mining crypto)
All Roads Lead to Robotics | Eric Janggeneral-purpose humanoid robots become a read/write API to physical reality
All Roads Lead to Robotics | Eric JangWe make the most money when we are right and contrarian. If we are right and consensus, since everyone knows about it, we will make some money, but if we are right and contrarian, we should be able to make a lot of money. Some people focus only on being contrarian, but being contrarian and wrong means we will lose money and look stupid. Being consensus and wrong, we will lose money and find safety in numbers, but what is the glory in that?
Investing is Hard - Outlier’s PathRather than being an inevitable loser, we should remember that business is an infinite game, and we shouldn’t use finite game approaches.
Finite and Infinite Mindsets - Part II - Outlier’s Path