Nate Hindman

Nate Hindman@NateHindman · 4 天前政策影响

Shoot this into my veins “Just imagine for a moment what you could accomplish, if only you would take this new superpower, your amazingly valuable skills, your great taste, your power of imagination, and do something that is outrageously AMBITIOUS! Something that makes others think you must be absolutely mental to even think for a second that it could be accomplished. And then go and work for a full year on just that, utilizing AI to the fullest extent possible. Milking the beast until it’s dry.” Here’s my outrageously ambitious thing: Create an AI-powered virtual heart that takes drugs shelved because of cardiotoxicity and redesigns them into safer medicines.

AI：偏政策推进，强调合规落地与规则明晰，对监管主线更敏感。

Nate Hindman@NateHindman · 5 天前赛道影响

To your point, is B2B SaaS for biotechs still a vastly safer and more predictable business model if DeepMind, Anthropic etc can flip a switch and eat your lunch? And are the biotech budgets for SaaS even that big - particularly in the age of “own your model” - to justify the effort? Sure you can find something incredibly niche that you hope frontier labs won’t touch, but VCs won’t fund that, since it’s likely more a lifestyle business. Perhaps SaaS-pocalypse is coming for bio (if it hasn’t already arrived). On the flip side, I wonder if AI has lowered the barrier for small teams to do early-stage discovery.

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Nate Hindman@NateHindman · 5 天前观点输出

https://t.co/muSt0XZP60

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Nate Hindman@NateHindman · 7 天前政策影响

Fable has been public for ~3 weeks and it still won’t answer the most basic questions about biology, like “how does a cell work?” Clumsy over-blocking this far into a release suggests that the bio guardrails aren’t nefarious or some kind of regulatory capture, but rather a genuine engineering challenge and a sign they can’t muster the resources to calibrate biological risk cleanly. Still, I expected better from a generational company like Anthropic. Particularly one working to make waves in the scientific community.

AI：偏政策推进，强调合规落地与规则明晰，对监管主线更敏感。

Nate Hindman@NateHindman · 7 天前赛道影响

@andrewmackenzie @mihailoxyz @claudeai Try anything related to biology https://t.co/CzxiyQUIwv

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Nate Hindman@NateHindman · 10 天前观点输出

Preprint: https://t.co/bjZoWdG2Km Tech blog: https://t.co/uYJ2K1gg7Y

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Nate Hindman@NateHindman · 10 天前赛道影响

Thrilled to see the @AppliedSciAI literature agent win the Best AI Scientist award at one of the world's leading ML conferences 🤯 You can try the agent in the link below. Some of its killer features come from the insight that the signals that often matter most in a scientific paper aren't in the text - they're scattered across figures, tables, gels, micrographs and electrophysiology traces. It was inspiring to watch @lukasweidener @markoasi1 & @mihailoxyz attack this problem and build an agent that hit SOTA across major literature search benchmarks like LitQA3, FigQA2 & TableQA2. ASI is focused on cardiac risk, using model-assay loops to build a “virtual heart” that predicts cardiac safety for small molecules, guides lead optimization and could eventually power digital twins in clinical trials. Our team runs the agent for internal R&D and to generate training data for our cardiac safety models. Concrete case: when CardioSafe flags a compound as toxic and its RL mode proposes safer analogs, the literature agent checks whether anyone has already assayed a close neighbor - so we’re not spending months on a "safer" molecule someone else quietly killed years ago. That answer almost never lives in the text of a paper, rather it’s usually in a dose-response curve, summplementary assay table or ion-channel panel buried in a figure. It's a good example of how prediction models paired with scientific reasoning creates new capabilities in drug discovery. Try ASI’s literature agent + CardioSafe today: https://t.co/r7HFewNhIA

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Nate Hindman@NateHindman · 11 天前赛道影响

The real bottleneck is that everyone’s AI slop keeps saying the real bottleneck

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Nate Hindman@NateHindman · 16 天前赛道影响

If there’s one thing crypto taught us, it is user experience trumps self-sovereignty The hook of own your money/data never took shape because the UX remained so bad My guess is open-source AI will face a similar battle - the “own your own intelligence” angle can work, but only if the UX of doing so roughly matches closed-source models Granted the UX bar for a 50-person ML org is not the UX bar for grandma buying Bitcoin and "own your model optimization loop" ≠ "be your own bank" But the adoption of self-sovereign anything tracks UX, not ideology, and open-source AI seems to inherit that gravityr

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Nate Hindman@NateHindman · 18 天前赛道影响

RT @markoasi1: our paper on alexandria - our literature agent - got a spotlight at the ICML 2026 AI4Science workshop we built alexandria t…

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。