rewind

@ScottyBeamIO this looks way beyond template work

CHATGPT 5.6 + HERMES AGENT JUST BUILT A WEBSITE BETTER THAN WHAT MOST $20,000 AGENCIES DELIVER – THE ONLY PROMPT WAS: “BUILD SOMETHING INSANE” → Full animated sequences, scroll-triggered transitions → A concept built around “the ages of AI agents” – work, sleep, forget, repeat → Then a second act: design, act, multiply – with its own animation to match From a pure UI and animation standpoint, this isn’t a template. This is the kind of design work you’d pay a studio for. But building fancy websites isn’t even the interesting part. The same setup can be scheduled to run tasks automatically, every single day, at the exact time you want – like sending you a morning brief at 6am before you’re even awake. This is what happens when you stop treating AI as a chatbot and start treating it as an assistant that actually executes. Bookmark this. Full demo in the video below.

@Nozelcode open source design wins

HAN CLONADO CLAUDE DESIGN Y PUEDES USARLO GRATIS Y SIN LÍMITES Se llama Open Design, un proyecto open source que te deja usar Claude para workflows de diseño sin pagar. Sin suscripciones. Sin límites (como la versión oficial). Acceso total. Esto es lo que puedes hacer: → Generar diseños UI/UX con Claude → Convertir prompts en diseños reales → Sustituir herramientas de diseño caras en muchos casos → Personalizarlo completamente (es open source) Está hecho para devs, indie hackers y creadores que no quieren quemar dinero en herramientas. De esos repos que pasan desapercibidos hasta que de repente todo el mundo los usa. Si usas AI + diseño, tienes que probarlo Enlace abajo👇 (guárdalo antes de que explote)

@wandermist marketing forgot the bottleneck

DGX SPARK 128GB VRAM AND IT LOSES TO A QUAD 3090 RIG BY 30X someone actually ran the numbers instead of trusting the marketing qwen3 32b q8, same model, same quant quad 3090 rig hits 104 tokens per second dgx spark hits 3.5 that's not a typo, that's a 30x gap gemma 327b q8 tells the same story, 25.4 vs 4.5, dgx spark barely limping along while a $4k rig built from used 3090s eats it alive the reason is boring and it's the one thing nobody puts in the marketing slide, bandwidth dgx spark ships with ~276 gb/s of memory bandwidth, a single 3090 alone hits 936 gb/s vram tells you what fits, bandwidth tells you what actually runs 128gb of memory means nothing if it can't feed the gpu cores fast enough, and this benchmark makes that painfully visible the trap is thinking more vram equals more speed it doesn't, it just equals more stuff sitting there loading slowly full bandwidth breakdown and gpu ranking in the thread below ↓

@kingwilliam_ the builders will pass everyone still collecting courses

ANDREW NG JUST GAVE AWAY A $1,000 AI ENGINEERING COURSE FOR FREE the man who taught half of silicon valley machine learning said it on camera, plainly - building agentic ai systems is the most valuable skill in ai right now three hours, start to finish - agents built from scratch, where ai engineering is actually headed, the full prompting course, then a working app in 30 minutes most people will bookmark this and never open it again the ones who actually sit through it walk away knowing how to build the exact systems companies are paying six figures for right now the ones who skip it will be explaining their job to someone who didn't here's the guide that turns everything in that video into a system you can build today

@noisyb0y1 the income claim is doing heavy lifting

Stanford leaked for free a 60-minute AI course lecture that paid universities sell for $75,000: I sent this to my friend - 3 months later his AI product makes $60k a month: 00:10 - AI automates your entire life 23:52 - Claude cuts your 8-hour workday down to 20 min 49:34 - you don't need a degree to make $60k a month This 60-minute talk replaces a year of studying AI at Stanford for $90,000 a year. Watch it today, then read the full breakdown in the article below.

@leopardracer custom assessments beat generic playbooks

the second brain enterprise team ran 100 companies through AI transformation and leaked exactly what worked: 00:03 - claude tailoring business assessments live instead of hardcoding them for every industry 12:11 - one prompt telling you exactly which two channels to focus on based on your real unit economics 44:52 - the 20% who become AI operators and end up running five people's jobs solo this conversation gives you for free what most consultants charge five figures to figure out over months watch it today, then see who in your team could be that 20%

@0xwhrrari observability is the hidden layer

Anthropic just leaked their 2026 agent roadmap In a 22-minute breakdown from the Claude team They walk through tools, memory, observability, and the parts most builders are still 12 months behind on The last 3 minutes alone are worth the watch Bookmark and watch it Then read the article below

@AnatoliKopadze describing the animation instead of building every layer feels like a huge shift

just think about the era we're living in you can now make layering for motion design by simply describing what you want to see for pros it speeds up the workflow, for beginners it's finally a way in, all inside After Effects powered by Claude, it can design, animate, script, write expressions right in your project and everything stays fully editable more information you can find below

@ridark_eth one agent was never enough

Paying developers by the hour is the old way. $15,000 in one month, and the team was Claude agents. Not one agent. A pipeline. One plans the work, one writes the code, one tests it, one checks for breaks. Each agent runs its piece. They share context. Nothing waits on a human in the middle. The billing stack matters. Basic tasks route to cheaper models. Complex tasks go to the capable ones. That gap cuts API costs significantly. RuFlow handles the routing automatically. 60 agents can run at once. It has 14,100 stars on GitHub and costs nothing to run. The developer pairs it with Obsidian for notes and task structure. Prompts go in, output comes out, the pipeline handles the rest. Month one on this setup: $15,000. That number is from one person's workflow, not a case study. The Claude subscription stays. The $200 platform fees go. The difference is who controls the stack. Open source means you fork it, adjust it, run it on your own infra. No vendor owns the ceiling.

@heyfredds 🙌🤝

@rewind02 saved thanks!

@0x_fokki one hour beats studio timelines

🚨 Disney spent over $150,000,000 making a Zootopia movie a 23-year-old American student recreated its entire red-carpet premiere on a laptop for $50, and the clip is pulling millions of views. Margot Robbie as Judy Hopps. Dwayne Johnson as a horse in a tuxedo. Shakira as Gazelle. none of it real. these celebrity premieres are the format taking over your feed right now. > Research scores the outliers: views ÷ channel median, over 30 and you copy that format this week > Claude reskins the winning shape onto a new subject and writes the shot list: 20 minutes > CapCut generates every celebrity, every reveal, exports 9:16: 1 hour > Make posts to 3 platforms and reads the view counts back in 48 hours 4 tools. $50/month. Disney pays a studio to premiere one film. this student ships a new one every week. first people to run the format eat the whole trend. the whole factory is in the article above👇

@gippp69 worktrees keep context clean

THIS F**KING 7,300 STAR REPO TURNS CLAUDE FABLE 5, CODEX, AND CURSOR INTO ONE AI TEAM THAT KEEPS WORKING WHILE YOU SLEEP 00:01 he opens Omnigent, an open source layer that puts multiple coding agents inside one workflow while sandboxes and policies control exactly what they can touch. setup takes minutes. Fable acts as the supervisor, cheaper models handle execution, and premium tokens are reserved for decisions that actually need judgment. each task gets split across separate agents and git worktrees, then a fresh model reviews the final diff before anything can merge. a $5 daily cap, 150 line approval limit, automated signoff tests, and a 95% pass rate keep the whole team from running wild. bookmark this. one repo, four agents, and roughly 15 to 20% of the usual token cost.

@heyfredds rust rewrite sounds promising

CAPCUT GRATIS, SIN MARCA DE AGUA Y SIN SUSCRIPCIÓN. YA EXISTE. Un grupo de devs se cansó y construyó la alternativa. Ya va por 64.000 estrellas en GitHub. Se llama OpenCut. Editor de vídeo completo, sin marca de agua ni paywalls. → Corre en tu navegador, nada se sube a ningún servidor → Compatible con web, escritorio y en camino a móvil → Open source con licencia MIT: úsalo también para trabajo comercial → Se está reescribiendo el núcleo en Rust para ir más rápido CapCut procesa tus vídeos en los servidores de ByteDance. Esto los procesa en tu ordenador. Es lo que CapCut debería haber sido desde el principio, sin la letra pequeña. Repo abajo.

@alphabatcher yeaah g

@rewind02 building second brain from Claude Fable 5 -> smart move

@angeldot_ less code fewer bugs

ESTE REPO HACE QUE CLAUDE ESCRIBA HASTA UN 94% MENOS DE CÓDIGO Fable 5 es el mejor modelo de código del mundo. También el más caro. Cada línea que escribe de más, la pagas tú. Ponytail lo arregla con una idea absurdamente simple: convierte a tu agente en el senior vago del equipo. Ese que mira tus 50 líneas y las reemplaza por una. Antes de escribir nada, el agente se pregunta: → Esto necesita existir? → Ya está en el codebase? → Lo hace la librería estándar? → Lo hace el navegador de forma nativa? → Cabe en una línea? Solo si todo falla, escribe el mínimo que funciona. Resultado: • hasta 94% menos código • un 20% menos de factura • un 27% más rápido Sin recortar seguridad ni validaciones. Se instala en 2 comandos: /plugin marketplace add DietrichGebert/ponytail /plugin install ponytail@ponytail 83.000 estrellas en GitHub. Gratis y open source. El mejor código es el que nunca escribiste. Enlace al repositorio abajo👇

@antisadh yeeeah

@rewind02 karpathy running the exact 4-layer second brain stack is insane

Andrej Karpathy just showed off his "second brain" and it reveals all the hidden connections the idea: every thought is linked to everything else - decisions, research, observations, past experience instead of starting from zero every time, he stores everything in one place and uses AI to: - find the info he needs - surface hidden connections between ideas - resurface old thoughts exactly when they're relevant - help him make decisions based on his entire accumulated context that's the whole point of a "second brain." it's not just about storing information - it's about turning it into a system that gets smarter and more useful over time most people build a pile of notes Karpathy built an operating system for his own thinking so how do you build a "Second Brain" for Claude Fable 5? 👇

@0xSlyth fact

@rewind02 Karpathy's second brain is next level ops

@0x_fokki yess

@rewind02 Clean review

@undefinedKi hybrid search is the baseline

This one rule runs every major AI company out there, and it's not what you think It isn't the model. Everyone calls the same API. The rule is: whoever controls what the model looks at, wins. That's RAG, and an IBM lecturer just laid out how to actually do it. Run keyword and semantic search together. Semantic catches intent, keyword nails the exact terms it fumbles: product codes, clause numbers, names. Hybrid retrieval is the baseline now, not an optimization. Rerank before anything reaches the model. The retriever pulls candidates fast and messy. The reranker reorders them for real relevance. Skip this and you're feeding the model whatever cosine similarity happened to like. Rewrite the query before you search it. The user's phrasing is rarely the best search string. Expand it, clarify it, then retrieve. Let the agent decide when to retrieve at all. This is the actual shift. It picks where to look, what to ask, when it has enough. It compares sources, validates claims, rewrites its own queries, and loops. Retrieval becomes a tool it reaches for while reasoning, not a fixed step it walks blindly. This isn't theory. Glean is worth $7.2B doing exactly this over company knowledge. Harvey is at $11B doing it over case law and contracts. Sierra at $16B doing it over support docs. None of them trained a model. They all built the retrieval layer. Full breakdown in the article below. Bookmark this

@rostikdeni retrieval keeps the brain usable

A ONE LINE INDEX CAN CUT A CLAUDE SECOND BRAIN'S TOKEN COST BY 40% The retrieval path starts with code, it strips a question down to keywords, scores the workspace without opening every file, reads the one relevant section, then gives Claude the question plus the evidence it actually needs In one implementation, that structure cut token use by 40% versus default Claude Code while keeping a memory layer with tens of thousands of files practical to query The durable asset is the routing tree: rules, skills, project context, and source files that can move to the next model when the current one changes Full architecture below