ethan (@ethanorbe) header
ethan (@ethanorbe) profile picture

mostly here to learn and share 🌬️

Joined March 2023
ethan retweeted
Your Black Friday inspiration hunt has reached browser tab 37. Time to give those ideas a job. Inside Dreamina Marketing Studio, browse templates built around winning ad concepts until a direction fits your product. Make that product the focus, then generate fresh campaign variations. Less time collecting references. More ideas you can actually use. Grab the $0.29 offer while it’s available. Your browser could use a holiday 👇 dreamina.capcut.com/ai-tool/… @Dreamina_AI #Dreamina #MarketingStudio #DreaminaOnly29Cents #HireDreamina #HireDreaminaBlackFriday #Seedance #DreaminaPartner #BlackFridayDeals
🤝 Paid partnership
10
7
11
305
ethan retweeted
Some briefs need a video ad. Others need an image. Then there’s the client who writes “both” and goes offline. Dreamina Marketing Studio supports both through trending ad templates in Dreamina’s Inspiration Library, continuously updated with fresh ideas. Choose your creative style, substitute your product for the original, personalize the model, and generate your ad. Fresh starting points mean less work recreating winning ads for your next project. For $0.29 during the promotion, this workflow deserves a bookmark. Start with the format your brief needs 👇 dreamina.capcut.com/ai-tool/… @Dreamina_AI #Dreamina #MarketingStudio #DreaminaOnly29Cents #HireDreamina #HireDreaminaBlackFriday #Seedance #DreaminaPartner #BlackFridayDeals
6
5
9
291
ethan retweeted
GEMINI YA CONTROLA TU ORDENADOR DE VERDAD Computer Use en la API de Gemini pasa a Generally Available: > Agentes en navegador, móvil y escritorio > Sin breaking changes respecto al preview > Listo para producción, no solo demos Docs oficiales abajo. ¿Dejarías que un agente te reserve un vuelo solo? guárdalo y sígueme para más → @karlarboledas
8
14
22
481
ethan retweeted
ESTE ROBOT APRENDIÓ MIRANDO A HUMANOS LiberAI lanza Liber-0, un World Action Model zero-shot: > Preentrenado con +2,2M horas de experiencia humana (vídeo + guantes) > 60 tareas nuevas (propuestas por GPT-6), 600 trials reales > 36% de éxito global sin demos de robot de esa tarea > Pick-and-place 63,5% · otras 34% · multi-step 10,5% Incluye el clip del boba en continuo, autónomo. guárdalo y sígueme para más → @maxwell_lenoir
5
14
15
251
ethan retweeted
GREG BROCKMAN: LA MUERTE PUEDE LLEGAR A SER UNA ELECCIÓN El presidente de OpenAI lo suelta sin filtro en una entrevista: Su visión es brutal: > La biología no es exclusiva de los humanos > “Death becoming a choice” es posible > En 12–18 meses veremos avances reales curando enfermedades > Hay que aplicar estos modelos a biología y life sciences YA Cita real: "Having cures for diseases… you're gonna see real delivery… over the next 12 months, 18 months." La IA realmente es el futuro.. guárdalo y sígueme para más → @ronixtec
7
14
19
331
ethan retweeted
JENSEN EN BOXWORKS: AGENTES CON EL CONOCIMIENTO DE TU EMPRESA En BoxWorks 2026, NVIDIA + Box muestran el combo: > Modelos abiertos Nemotron > Contexto y tools de negocio > Agentes que usan el conocimiento interno de verdad Menos chatbot genérico, más agente que sabe de tu empresa. guárdalo y sígueme para más → @larsholst_
6
14
15
396
ethan retweeted
🚨 PEGA ESTA FRASE EN TU CLAUDE.md Y RENDIRÁ x10 VECES MÁS Investigadores acaban de descubrir la forma en la que tu agente se sabotea a sí mismo. Le cree antes a tu archivo que a tu propio código. Crearon 40 proyectos en los que el CLAUDE.md decía una cosa y el código hacía otra. El resultado: → El 90% de las decisiones siguió el archivo desactualizado → Claude Code vio el conflicto en 79 de 80 ejecuciones. Y lo siguió igual → Codex reescribió código que funcionaba en 20 de 80. En 17 fallaron los tests ¿Por qué? Tu archivo le llega como una orden. Tu código, como un simple resultado. Lo mejor es que se arregla con una sola frase. Precisamente por eso te dejo abajo mi guía de Claude Code de 0 a experto, con lo que debe llevar tu CLAUDE.md. Abajo también tienes la frase exacta y el paper completo ↓
21
12
42
2,263
ethan retweeted
ESTO ES ORO PURO Todo lo que necesitas para dominar Claude en una sola web, GRATIS Se llama Claude 101: guías cortas y gratis, ordenadas de principiante a experto. → Nivel 1: primeros pasos, configuración y prompts → Niveles 2 y 3: skills, límites, privacidad y que Claude escriba como tú → Nivel 4: Claude Code, explicado para quien nunca va a programar Lo difícil ya no es encontrar información sobre Claude. Es saber por dónde empezar. 🔖 Guárdalo para no perderlo
19
20
78
5,325
ethan retweeted
🚨 EL CREADOR DE CLAUDE CODE CONTÓ CÓMO LO USA Y ALGUIEN LO JUNTÓ TODO EN UN REPO GRATIS La mayoría le escribe a Claude Code como si fuera un chat. Un desarrollador reunió lo que hacen sus creadores y otros expertos en un repo gratis. Se llama claude-code-best-practice. 10 trucos que puedes probar hoy: → Empieza en modo plan: Shift+Tab y que piense antes de tocar nada → Deja tu CLAUDE.md por debajo de 200 líneas → Pídele que te entreviste antes de construir → Si falla, no lo corrijas: Esc dos veces o /rewind → /compact si sigues con lo mismo, /clear si cambias de tema → Tarea nueva, sesión nueva → Las búsquedas largas, a un subagente: tú recibes la conclusión → Lo que haces más de una vez al día, conviértelo en skill o comando → Si te atascas, pásale una captura → Que otro Claude revise el plan antes de ejecutarlo Son 10 de los 83 que tiene el repo, cada uno con su fuente, entre ellas Boris Cherny, creador de Claude Code. +68.000 estrellas en GitHub. Licencia MIT. Pasamos de corregirlo cien veces a que salga bien a la primera. 📌 Guárdalo antes de necesitarlo. Repo en el comentario de abajo 👇
16
12
55
2,757
ethan retweeted
the future of robotics won't be built in the real world alone. it will be imagined, tested and perfected in simulation before ever touching the ground. the sim-to-real gap is one of the biggest challenges in robotics. this is how we start closing it. @moonlake
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
18
9
224
21,308
ethan retweeted
ROBOTS NEED THEIR OWN TRAINING GROUNDS. simulation is becoming the place where robots learn, fail, adapt and try again. the interesting part is when everything they learn there actually transfers to the real world. this is where robotics is heading.
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
16
12
235
19,059
ethan retweeted
the next generation of robots won't learn everything through trial and error in the real world. → build a simulated world → let robots fail safely → learn from those failures → transfer that knowledge into reality the loop between simulation and reality might define the next era of robotics.
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
16
12
218
17,938
ethan retweeted
THE REAL WORLD IS JUST THE FINAL TEST. the training happens in simulation. robots that can learn, fail and improve in increasingly realistic virtual environments are going to take robotics somewhere we've barely started exploring.
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
18
10
259
36,494
ethan retweeted
kinda crazy to think that the next big leap in robotics might happen inside simulated worlds. robots learning, failing, adjusting and trying again thousands of times before doing the same thing in reality. simulation is becoming a pretty big deal. @moonlake
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
12
12
213
16,430
ethan retweeted
first, teach the robot the world. then let it learn how to move through it. simulation + reinforcement learning could fundamentally change how we build intelligent machines. really interesting research from @moonlake
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
15
11
247
20,805
we taught machines how to think. now we're teaching them how to interact with the physical world. simulation could be the missing link between AI intelligence and real-world capability. really interesting work from @moonlake
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
11
10
239
19,077
ethan retweeted
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
67
144
91
1,388
421,649
ethan retweeted
VOICE AI HAS A DEMO PROBLEM. WHAT HAPPENS WHEN SOMEONE ACTUALLY CALLS? A smooth demo doesn’t mean an AI agent can handle a real customer. That’s what makes Cekura Bench interesting: → real phone calls, not text simulations → the same tests for every platform and model → 1000+ caller scenarios, each tested 3 times → every result and call recording publicly available No cherry-picked demos. No taking a company’s word for it.
Introducing Cekura Bench: the world’s leading Voice AI Benchmarks We benchmarked 50+ models from openAi, Google, Nvidia, xAI, Inworld, Gradium, Assembly, Fish audio and more Now you can: → Compare naturalness, performance and cost → See how the best voice agents are being built. → Find the best fit for what you are building. Our tests cover Speech-to-Speech models, Speech-to-Text models, Text-to-Speech Models, and Voice Agent platforms. We recorded real humans doing tasks in different domains and created the most representative voice benchmark out there! Explore the benchmarks: benchmarks.cekura.ai We would love to collaborate with more teams on these benchmarks. Reach out if you are interested.
12
8
298
17,304
ethan retweeted
Swap objects, erase stuff, change the background, the weather, even the style. All in one video edit.
Introducing Viggle Remake. The most powerful video transformation agent. And we’re releasing the weights today. → recast every character in a scene, and they stay consistent across cuts → swap objects, change the background, restyle or add VFX → up to 2 minutes → keep chatting to fix any shot until it’s right From $0.96 per 10 seconds at 720p, less than half of what Higgsfield Genjutsu charges. Try it now: viggle.ai/meme/app/home?ui=v… Open weights: huggingface.co/Viggle/Viggle…
10
6
281
15,794
ethan retweeted
we taught machines how to think. now we're teaching them how to interact with the physical world. simulation could be the missing link between AI intelligence and real-world capability. really interesting work from @moonlake
The RL era of robotics is here, and simulation is how to scale it. An early look at our research on a achieving a better scaling paradigm for robot learning with simulation 👇
14
7
190
16,648