Barto Molina

maybe gta6 is also too dangerous to release

@victorianoi @NotebookLM @GeminiApp 💯 los podcast que genera estan muy bien, aunque no me termino de acostumbrar a escuchar AIs Igual con esto y un buen prompt se saca algo parecido ai.google.dev/gemini-api/doc…

@bartomolina @NotebookLM @GeminiApp pero para mín lo principal es que me haga el podcast con esa calidad! no tanto la parte de retrieval de información! lo suelo hacer sobre un documento o dos como mucho!

Last year, we integrated into the @GeminiApp by allowing you to upload your notebooks as sources. Now, we’re taking our relationship to the next level 🏠 ♥️ Starting today, you can now: — Access all of your personal, unshared notebooks directly inside the Gemini App — Use your chats with Gemini as sources in new or existing unshared notebooks We're rolling out notebooks in Gemini today, starting with Google AI Ultra, Pro, and Plus subscribers on the web. In the coming weeks, we'll expand access to mobile, more countries across Europe, and to free users.

@victorianoi @NotebookLM @GeminiApp File Search te da la mitad (simplificar el RAG). Hice un nodo de n8n y me funciona muy bien github.com/bartomolina/n8… Soporta muchos tipos de archivo y el pricing esta bien tambien porque no te cobran por el storage, solo por los embeddings y las queries

@NotebookLM @GeminiApp very nice! it would be very amazing being able to create and consume Notebooks from an API/MCP/CLI too!

Openclaw, beyond the hype. Thanks @bartomolina for the amazing event

hosting zero to agent in Madrid on 24/4 global build week. ship a real agent with @v0 + @vercel. free, $6k+ prizes. Link below #ZerotoAgent

@onusoz ok these openclaw wfs look really cool, thinking about migrating the last few ones I still have in n8n

Is Claude better or Codex? There are many benchmarks to answer that. But they are BORING I propose something more interesting: ⚔️ AI BATTLE ⚔️ A 1v1 real-time quiz format where AI agents try to pose each other problems that they think the other agent will not be able solve Claude vs Codex 10 questions each Codex asks first, Claude tries to answer Then Claude asks and Codex tries to answer Repeat 20 minutes to come up with a problem and 20 minutes to solve it Judge (Codex) judges the validity of the questions and answers, and gives points All automated, with acpx flow feature Implementation and full rules all open source, on github osolmaz/ai-battle So who won? I ran 4 games. It tied in 2, and Codex won in 2 closely An example question by Codex, which Claude could not answer: How many 3-colorings of the edges of the complete bipartite graph K_{5,5} are there with the following two properties: (1) there is no monochromatic 4-cycle, and (2) among the 25 edges, exactly 15 are red, exactly 5 are blue, and exactly 5 are green? Which is apparently 4029912, but Claude answered 0 In other cases, Claude asked a flawed question and failed to come up with a valid question in 20 minutes. So that's how it lost those 2 games with just 1-2 point difference In these 4 runs, Codex answered every question by Claude correctly. But there were some runs where it couldn't, which I did not commit to the repo because the runs couldn't complete due to bugs I did not tell them do ask math questions, but that is what they tended to do, because the answers had to be verifiable by the judge. The quiz can be done in any hard subject, physics, chemistry, computer science... Opus 4.6 and GPT 5.4 matched very closely in terms of problem creation and solving. But I cannot tell how creative these problems were at first glance. Maybe someone with more experience can tell me, looking at the problems in the repo? I need someone to tell me how legit they are Please take the code, modify it and run with different rules and subjects. I am curious to see the results! You will need paid subscriptions to all the models/agents you want to test of course I also feel that the game structure has a potential to be used in self-play. If you are an ML researcher, please look at the repo and lmk if this or a variant of it could be useful in RL! Full transcripts of the runs, including Codex and Claude session files are committed to the repo, for those who want to do archaeology on them Btw this idea came from the desire, "how can I create a cool demo of acpx flows?" Whole game is implemented in typescript, and automatically drives Codex and Claude sessions over ACP, Agent Client Protocol The video below is from acpx flow viewer rendering a run. You can see it loop through the same paths, first letting Codex ask, then Claude, then repeat acpx flows use a general programmatic workflow engine where ACP is just one type of node. You should be able to use it for non-ACP workflows, but I haven't tried that yet This implementation is separate from OpenClaw's current workflow implementations, with the intention to merge them somehow in the future You might find bugs in my implementation. Feel free to send PRs. I wanted to do more runs but I finished my Codex plan. It would be great if this idea could evolve in a decentralized manner!

Eventazo de @bartomolina este miércoles! OpenClaw de la mano de los que más sabe. luma.com/0hfofmox via @LumaHQ

Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw. You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.

The loudest story about AI is a lonely one. One person with an army of chatbots. Other humans are friction. That gets the future wrong. The best things aren’t built alone. In a moment of change, we want to remind the world (and ourselves) what Notion stands for: — Think Together

This is something a lot of people are also wondering when they see people on @x showing off how they're running 10 or 20 or more agents at a time. I think the max number most humans can manage is 3 - 5 at a time for most actual "real" engineering work. Sometimes I have more than that number running, but it's not for new features / products it's usually just running low-stakes/secondary tasks (bugs, research, marketing, internal tools), skipping or automating QA, and using heavy orchestration, batching, or agent pipelines so the I am not reviewing everything in real time. IMO the real argument for parallel agents is team-wide, company wide, and org-wide usage. When you have a few people working on new docs PRs, a few people working on new design updates, a few people fixing bugs, a few people adding new features, et... and all of this work is happening in parallel. This is where we see the bulk of the work happening with Devin and the biggest value add for running tens or dozens or even hundreds of agents in parallel (often on the same codebase)

@TomasStambulsky 💯

Hoy me volví loco armando una mega organización en Paperclip: AI employees de contenido, research, competencia… todo perfecto. El problema? No definí los skills. Resultado: el sistema empezó a improvisar cómo hacer todo. Lección: en AI, los roles sin skills son solo humo.

a few practical steps to securing against supply chain attacks

@dabit3 Try Marp (markdown+html) marp.app

What's the best AI tool for helping to make nice slide decks?

Build the thing that feels too ambitious. If you're using agents to do exactly what you were doing before - just faster - you may be thinking too small. Ambitious is the right size side-project.

@XMihura 👀 @metasurfero MadBuilders x Samplia seria un buen sitio

REFUGIO MADRID wenas! viendo la gran acogida que tuvo la idea del hackathon por Madrid, he estado dándole una vuelta para bajar la pelota al suelo mi intención es clara: hacer un evento guay, sin muchas pretensiones pero con mucho nivel. el objetivo es poner en contacto a gente competente con ganas de hacer cosas quiero combatir el derrotismo que veo a veces en España y demostrar que "you can simply do things". Soy de los que cree que ver a otros construyendo te da el empujón y el coraje para lanzarte tú el plan que tengo en la cabeza es algo así (sujeto a cambios a medida que se vaya definiendo). habrá dos sesiones: - sesión 1 (mañana/mediodía): charlas técnicas y hackathon. aforo capado a 45 personas máximo (15 equipos de 3) - sesión 2 (tarde): ecosistema y networking. aforo ampliado para founders, inversores, curiosos y terminar con un cóctel/cervezas SESIÓN 1 (Aforo max 45 personas) ------------------------------------------ 1. CHARLAS (9:00 - 10:30) empezamos el día tranquilos. una introducción y me gustaría conseguir algún que otro invitado top para que nos cuenten cosas chulas. esta sesión tendrá una orientación un poco más técnica. probablemente daré yo también una pequeña charla e introduciré el hackathon. 2. EL HACKATHON (11:00 16:30) después de las charlas, a quemar tokens. la idea del hackathon no es que montéis el enésimo wrapper de la API de OpenAI o hacer un pitch de tu próxima start-up, para eso creo que ya hay otros hackathones me gustaría montar equipos de unas 3 personas que se van a enfrentar a un problema técnico difícil, fuera de su area de expertise. tendrán unas 5 horas para resolverlo ayudándose de herramientas de IA creo que es una buena forma de aprender a usar estas herramientas y conocer gente para el hackathon no busco solo devs que tengan 800 horas en el Claude Code, quiero perfiles técnicos diversos: físicos, biólogos, economistas, ingenieros... gente con capacidad analítica de todas las edades animo a las chavalitas a que se apunten, molaría que no fuera el típico evento de IT que parece un campo de nabos os aviso que habrá un pequeño proceso de selección para asegurar el nivel y montar equipos equilibrados SESIÓN 2 (Aforo según espacio) ------------------------------------------ 3. TECH EN ESPAÑA (17:00-20:00) cuando acabe el hackathon, abrir las puertas a un público más amplio (founders, inversores, startups, curiosos). habrá charlas, y la idea central de la tarde será mostrar a los xavales que se pueden construir cosas serias en España y en Madrid quiero insistir en el you can simply do things, quiero traer a gente que hable de sus éxitos y de sus fracasos reales montando proyectos tech e investigación en España después de eso, a tomarnos unas cervezas y un cóctel para charlar sobre el ecosistema y que los participantes conecten entre ellos y con las empresas/universidades ------------------------------------------ Qué necesitamos? (sponsors) - Espacio: un sitio para 30-45 personas currando (y aforo ampliado por la tarde) con mesas, enchufes y una conexión que aguante a 45 personas quemando el Claude Code - Pizzas, café, cóctel, cervezas: lo básico para sobrevivir al día - Créditos de API: podemos ir en formato Bring Your Own Tokens, pero si algún sponsor se anima a dar créditos para que los xavales no gasten de su bolsillo, sería top - Transporte y alojamiento de los ponentes: puede ser que quiera traer gente que no vive en Madrid. De ser así habría que cubrir sus gastos. - Premios: lo que queráis aportar para los que ganen el reto (hardware, licencias, fast-tracks para entrevistas...). ¿Cuándo? pues para la fecha del evento barajo 3 opciones: - Sábado 2 de mayo (es puente en madrid, no sé yo) - Sábado 9 de mayo - Sábado 27 de junio IMPORTANTE⚠️⚠️⚠️ si tenéis intención real de asistir, escribid en los comentarios de este tuit: REFUGIO y qué fechas os cuadran para votar. sudo de hacer una encuesta anónima, que a saber quien vota, prefiero veros en las replies y de paso le damos difusión al tuit para llegar a más gente para sponsors, charlas o cesión de espacios, escribidme por DM o a este email: mihurax at gmail dot com Abajo pongo el GitHub donde mantendré un README.md con toda la información del evento según se vaya definiendo

@elwatto having a custom dashboard to keep track of the agent-channels might help. Also worth trying some custom workflows to assign tasks to agents and further isolate contexts. kanban is working great for me

@bartomolina a lot messier to set up securely in my experience, but it can totally be skill issue

Update: thanks for the recommendations. Discord is the way to go. I also decided to burn my setup and start over. It had way too much debt from the early clawdbot/moltbot/openclaw days. New bot feels so fresh. Excited to keep tweaking and nailing the setup.

@elwatto Cool. Yes. Multiagent is the way to go to avoid context/memory contamination. You can do that with tg topics too though right?

@bartomolina organization. one agent per channel.

you're probably underestimating how crazy things are