Dream 100 · Public registry
Contribution before contact.Long-term relationships, earned in public.
These are the people and organizations whose work changes how we build. The aim is not proximity for its own sake. It is to notice carefully, understand accurately, make something useful, and let collaboration become a possible consequence.
4 × 25
balanced fields of attention
20
priority relationships this cycle
5 stages
observe to collaborate
Priority cohort
Twenty relationships receiving the clearest attention now
Priority means higher research cadence and more deliberate contribution. It is not a judgment of status or value.
Registry · snapshot 2026-08-16.2
One hundred sources of consequential work
Four fields prevent the list from collapsing into model-company attention alone. Each row records why the work matters and how closely it is being followed.
01 / 04
Frontier intelligence
Model labs and research groups changing the capability, safety, and deployment frontier.
- 01Anthropic Now
Anthropic can change the capability or deployment assumptions underneath an AI system.
Signals - 02Mistral AI Now
Mistral AI can change the capability or deployment assumptions underneath an AI system.
Signals - 03OpenAI Now
OpenAI can change the capability or deployment assumptions underneath an AI system.
Signals - 04SpaceXAI Now
SpaceXAI can change the capability or deployment assumptions underneath an AI system.
Signals - 05DeepSeek Active
DeepSeek can change the capability or deployment assumptions underneath an AI system.
Signals - 06Google DeepMind Active
Google DeepMind can change the capability or deployment assumptions underneath an AI system.
Signals - 07Meta AI Active
Meta AI can change the capability or deployment assumptions underneath an AI system.
Signals - 08Microsoft AI Active
Microsoft AI can change the capability or deployment assumptions underneath an AI system.
Signals - 09NVIDIA Active
NVIDIA can change the capability or deployment assumptions underneath an AI system.
Signals - 10Safe Superintelligence Active
Safe Superintelligence can change the capability or deployment assumptions underneath an AI system.
Signals - 11Thinking Machines Lab Active
Thinking Machines Lab can change the capability or deployment assumptions underneath an AI system.
Signals - 12AI21 Labs Horizon
AI21 Labs can change the capability or deployment assumptions underneath an AI system.
Signals - 13Alibaba Qwen Horizon
Alibaba Qwen can change the capability or deployment assumptions underneath an AI system.
Signals - 14Amazon AGI/AWS AI Horizon
Amazon AGI/AWS AI can change the capability or deployment assumptions underneath an AI system.
Signals - 15Apple ML Horizon
Apple ML can change the capability or deployment assumptions underneath an AI system.
Signals - 16Baidu ERNIE Horizon
Baidu ERNIE can change the capability or deployment assumptions underneath an AI system.
Signals - 17ByteDance Seed Horizon
ByteDance Seed can change the capability or deployment assumptions underneath an AI system.
Signals - 18Cohere Horizon
Cohere can change the capability or deployment assumptions underneath an AI system.
Signals - 19Liquid AI Horizon
Liquid AI can change the capability or deployment assumptions underneath an AI system.
Signals - 20MiniMax Horizon
MiniMax can change the capability or deployment assumptions underneath an AI system.
Signals - 21Moonshot AI/Kimi Horizon
Moonshot AI/Kimi can change the capability or deployment assumptions underneath an AI system.
Signals - 22Nous Research Horizon
Nous Research can change the capability or deployment assumptions underneath an AI system.
Signals - 23Sakana AI Horizon
Sakana AI can change the capability or deployment assumptions underneath an AI system.
Signals - 24Tencent Hunyuan Horizon
Tencent Hunyuan can change the capability or deployment assumptions underneath an AI system.
Signals - 25Z.ai/GLM Horizon
Z.ai/GLM can change the capability or deployment assumptions underneath an AI system.
Signals
02 / 04
Agentic infrastructure
The platforms and primitives that turn models into reliable products and working systems.
- 01Canva Now
Canva shapes how AI work is built, operated, reviewed, or distributed.
Signals - 02Cloudflare Now
Cloudflare shapes how AI work is built, operated, reviewed, or distributed.
Signals - 03Cursor/Anysphere Now
Cursor/Anysphere shapes how AI work is built, operated, reviewed, or distributed.
Signals - 04Hugging Face Now
Hugging Face shapes how AI work is built, operated, reviewed, or distributed.
Signals - 05Lovable Now
Lovable shapes how AI work is built, operated, reviewed, or distributed.
Signals - 06n8n Now
n8n shapes how AI work is built, operated, reviewed, or distributed.
Signals - 07Replit Now
Replit shapes how AI work is built, operated, reviewed, or distributed.
Signals - 08Supabase Now
Supabase shapes how AI work is built, operated, reviewed, or distributed.
Signals - 09Vercel Now
Vercel shapes how AI work is built, operated, reviewed, or distributed.
Signals - 10Cerebras Active
Cerebras shapes how AI work is built, operated, reviewed, or distributed.
Signals - 11GitHub Active
GitHub shapes how AI work is built, operated, reviewed, or distributed.
Signals - 12LangChain Active
LangChain shapes how AI work is built, operated, reviewed, or distributed.
Signals - 13LlamaIndex Active
LlamaIndex shapes how AI work is built, operated, reviewed, or distributed.
Signals - 14Notion Active
Notion shapes how AI work is built, operated, reviewed, or distributed.
Signals - 15OpenRouter Active
OpenRouter shapes how AI work is built, operated, reviewed, or distributed.
Signals - 16v0 Active
v0 shapes how AI work is built, operated, reviewed, or distributed.
Signals - 17Composio Horizon
Composio shapes how AI work is built, operated, reviewed, or distributed.
Signals - 18CrewAI Horizon
CrewAI shapes how AI work is built, operated, reviewed, or distributed.
Signals - 19Fireworks AI Horizon
Fireworks AI shapes how AI work is built, operated, reviewed, or distributed.
Signals - 20Groq Horizon
Groq shapes how AI work is built, operated, reviewed, or distributed.
Signals - 21Modal Horizon
Modal shapes how AI work is built, operated, reviewed, or distributed.
Signals - 22Pydantic AI Horizon
Pydantic AI shapes how AI work is built, operated, reviewed, or distributed.
Signals - 23Replicate Horizon
Replicate shapes how AI work is built, operated, reviewed, or distributed.
Signals - 24Together AI Horizon
Together AI shapes how AI work is built, operated, reviewed, or distributed.
Signals - 25Trigger.dev Horizon
Trigger.dev shapes how AI work is built, operated, reviewed, or distributed.
Signals
03 / 04
Leading minds
Builders, researchers, and interpreters whose decisions and explanations shape the field.
- 01Guillermo Rauch Now
Guillermo Rauch offers decisions and explanations worth testing against real systems.
Signals - 02Andrej Karpathy Active
Andrej Karpathy offers decisions and explanations worth testing against real systems.
Signals - 03Andrew Ng Active
Andrew Ng offers decisions and explanations worth testing against real systems.
Signals - 04Arthur Mensch Active
Arthur Mensch offers decisions and explanations worth testing against real systems.
Signals - 05Chip Huyen Active
Chip Huyen offers decisions and explanations worth testing against real systems.
Signals - 06Clément Delangue Active
Clément Delangue offers decisions and explanations worth testing against real systems.
Signals - 07Dario Amodei Active
Dario Amodei offers decisions and explanations worth testing against real systems.
Signals - 08Demis Hassabis Active
Demis Hassabis offers decisions and explanations worth testing against real systems.
Signals - 09Ethan Mollick Active
Ethan Mollick offers decisions and explanations worth testing against real systems.
Signals - 10Fei-Fei Li Active
Fei-Fei Li offers decisions and explanations worth testing against real systems.
Signals - 11Jensen Huang Active
Jensen Huang offers decisions and explanations worth testing against real systems.
Signals - 12Sam Altman Active
Sam Altman offers decisions and explanations worth testing against real systems.
Signals - 13Simon Willison Active
Simon Willison offers decisions and explanations worth testing against real systems.
Signals - 14Thomas Wolf Active
Thomas Wolf offers decisions and explanations worth testing against real systems.
Signals - 15Amjad Masad Horizon
Amjad Masad offers decisions and explanations worth testing against real systems.
Signals - 16Aravind Srinivas Horizon
Aravind Srinivas offers decisions and explanations worth testing against real systems.
Signals - 17Daniela Amodei Horizon
Daniela Amodei offers decisions and explanations worth testing against real systems.
Signals - 18Harrison Chase Horizon
Harrison Chase offers decisions and explanations worth testing against real systems.
Signals - 19Ilya Sutskever Horizon
Ilya Sutskever offers decisions and explanations worth testing against real systems.
Signals - 20Jerry Liu Horizon
Jerry Liu offers decisions and explanations worth testing against real systems.
Signals - 21Logan Kilpatrick Horizon
Logan Kilpatrick offers decisions and explanations worth testing against real systems.
Signals - 22Mira Murati Horizon
Mira Murati offers decisions and explanations worth testing against real systems.
Signals - 23Nathan Lambert Horizon
Nathan Lambert offers decisions and explanations worth testing against real systems.
Signals - 24swyx Horizon
swyx offers decisions and explanations worth testing against real systems.
Signals - 25Yann LeCun Horizon
Yann LeCun offers decisions and explanations worth testing against real systems.
Signals
04 / 04
Generative culture
Studios, tools, and artists expanding what generative creation can look, sound, and feel like.
- 01Bilawal Sidhu Now
Bilawal Sidhu expands the practical language of generative creation.
Signals - 02Curious Refuge Now
Curious Refuge expands the practical language of generative creation.
Signals - 03ElevenLabs Now
ElevenLabs expands the practical language of generative creation.
Signals - 04Refik Anadol Now
Refik Anadol expands the practical language of generative creation.
Signals - 05Runway Now
Runway expands the practical language of generative creation.
Signals - 06Suno Now
Suno expands the practical language of generative creation.
Signals - 07Adobe Firefly Active
Adobe Firefly expands the practical language of generative creation.
Signals - 08Black Forest Labs Active
Black Forest Labs expands the practical language of generative creation.
Signals - 09Claire Silver Active
Claire Silver expands the practical language of generative creation.
Signals - 10HeyGen Active
HeyGen expands the practical language of generative creation.
Signals - 11Holly Herndon Active
Holly Herndon expands the practical language of generative creation.
Signals - 12Krea Active
Krea expands the practical language of generative creation.
Signals - 13Luma AI Active
Luma AI expands the practical language of generative creation.
Signals - 14Midjourney Active
Midjourney expands the practical language of generative creation.
Signals - 15Paul Trillo Active
Paul Trillo expands the practical language of generative creation.
Signals - 16Dave Clark Horizon
Dave Clark expands the practical language of generative creation.
Signals - 17Freepik Horizon
Freepik expands the practical language of generative creation.
Signals - 18Higgsfield Horizon
Higgsfield expands the practical language of generative creation.
Signals - 19Ideogram Horizon
Ideogram expands the practical language of generative creation.
Signals - 20Kling AI Horizon
Kling AI expands the practical language of generative creation.
Signals - 21Niceaunties Horizon
Niceaunties expands the practical language of generative creation.
Signals - 22Pika Horizon
Pika expands the practical language of generative creation.
Signals - 23PJ Ace Horizon
PJ Ace expands the practical language of generative creation.
Signals - 24Shy Kids Horizon
Shy Kids expands the practical language of generative creation.
Signals - 25Sougwen Chung Horizon
Sougwen Chung expands the practical language of generative creation.
Signals
Operating rule
Notice consequential work. Verify it at the source. Translate it into something useful. Contribute before asking.
Observe → understand → contribute → converse → collaborate. No automatic outreach and no pay-to-rank placement.