Fechu

ChatGPT 5.5 ya puede crear Motion Graphics. Animaciones de texto, transiciones y gráficos en movimiento que antes requerían After Effects y años de experiencia. 28 minutos de tutorial. Pasos simples. Aquí ⬇️

Google adopted it. AWS adopted it. Microsoft adopted it. Oracle adopted it. LangChain, Mastra, TanStack, LlamaIndex — all of them. CopilotKit's AG-UI Protocol quietly became an industry standard while everyone was debating which agent framework to use. $27M raised. 40K+ GitHub stars. Fortune 500 in production. the agentic frontend just got its defining company.

Un cliente se registra. Usa tu producto. Después disputa el cargo. Ahora tienes dos opciones: Aceptar la pérdida. O probar lo que pasó. La evidencia normalmente está regada por todos lados. Fecha y hora del registro. Dirección IP. Facturas. Uso del producto. Correos de soporte. Términos aceptados. Política de reembolso. Capturas de pantalla. Línea de tiempo de los eventos. La mayoría de los equipos no pelean cada disputa porque armar la evidencia es un dolor de cabeza. Por eso liberamos AgentGuard™ Chargeback como código abierto. Un SDK gratuito y de código abierto en TypeScript que convierte las disputas de Stripe y la línea de tiempo de tu producto en evidencia Visa CE 3.0, en el formato que espera la Disputes API de Stripe. Sin LLM en el pipeline de evidencia. Sin datos inventados. Plantillas estáticas. Por defecto submit:false. Bitácora firmada con Ed25519 y encadenada por hash. El servidor MCP funciona con Claude Code, Cursor, Codex, o cualquier LLM local (Ollama, LM Studio, Qwen). Licencia MIT. Auto-hospedado. Funciona sin conexión a internet. Tu evidencia se queda en tu repositorio. Los datos de tus clientes se quedan en tu infraestructura. Un humano aprueba antes de que algo se envíe a Stripe. Hecho para comercios cansados de perder disputas que podían haber ganado. 100% código abierto. Instalación y GitHub en los comentarios.

JavaScript tip: Use the `speechSynthesis.speak()` method to add voice to your web applications:

Hey devs, which paywall will hit $10K MRR quickly? - Hard paywall - 7-day free trial - 3-day free trial - Free + paid (freemium) - Soft paywall

A complete breakdown on how to use AI to become the most recognized founder in your industry in just 45 minutes a week:

the prompt was: "github.com/Leonxlnx/taste… Based on the skill above, generate images for a website for an AI agency. The design should include eight sections, with one image per section, for a total of eight distinct images. The website is for a creative AI company focused on research in creativity and design. Because of that, I want the visuals to feel highly original, playful, and art-directed, with text integrated thoughtfully into the design. Make it feel ultra-creative and intentional, like an Awwwards SOTD-level website in both concept and execution. Please go beyond standard layouts. Do not rely only on simple text-left, image-right compositions. Explore more experimental and varied layouts. Feel free to go completely wild, but keep it purposeful, not random. I want different section structures, including horizontal images, fullscreen sections, full background imagery, and more minimal sections with beautiful colors and a strong sense of motion or animation. Please use full background images or strong full-background color compositions, not just plain white sections. Keep it in light mode. Overall, try to stay somewhat consistent across the site while still making each section feel distinct. I want it to look crazy creative, thoughtful, and visually impressive, with strong UX and a clear sense of purpose. Generate 8 different images total. Do not combine them into one image. Each image should represent one section of the website."

The future of AI video is here, and it is completely mind-blowing. 🤯 OpenArt's new Smart Shot feature takes the "GPT-image 2 + Seedance 2.0" workflow and pushes it to the absolute extreme. The generated storyboard isn't just a simple sketch anymore. It is a highly detailed director's shooting guide, complete with precise camera lenses, framing, and lighting setups. One single prompt. A full multi-cut cinematic sequence. Zero editing required. Let me show you how to use this game-changer in the thread below 👇

Someone just built replica of Claude Design. That works Claude Code, Codex, Cursor or any other coding agent. 100% Opensource.

Claude Code + Higgsfield MCP is f*cking cracked 🤯 I built an entire DTC ad campaign inside Claude Code using the new Higgsfield MCP. One product URL → hero static, animated hero shot, 2 UGC clips with a creator wearing the product. 5 assets. One Claude conversation. 3 Higgsfield models. All inside Claude Code. Perfect for DTC brands and agencies who need full campaign packages without booking a shoot or briefing a designer. If you're spending hours every week generating statics in one tool, briefing a motion designer for the hero clip, then chasing a UGC creator for the talking-head shots — this MCP eliminates the entire pipeline: → Drop a product URL into Claude Code → Claude pulls the brand brief — voice, hero SKUs, visual style, target customer → Generates the hero static with ChatGPT Images 2.0 → Animates it into a 5-second cinematic opener with Seedance 2.0 → Generates a UGC creator with GPT Image 2 → Drops her in the product and generates 2 native UGC video clips with Seedance 2.0 No tab-switching between tools. No copy-pasting prompts between platforms. No briefing 3 different vendors for one campaign. What you get: → A complete campaign package — static, animation, UGC — from one product URL → Brand-specific outputs that pull from a real brief, not generic AI slop → Claude making creative decisions between every step (which variation wins, which creator fits the persona, which clip needs a re-spin) → A repeatable pipeline you can run for any product in your catalog Built 100% in Claude Code with the Higgsfield MCP. I recorded a full walkthrough showing exactly how this works: the MCP setup, every prompt, every model, the full campaign output. Want the full video walkthrough? > Like this post > Comment "MCP" And I'll send it over (must be following so I can DM)

Introducing Website ↔ App. Turn any website into a mobile app. Just paste a URL. Claude Opus 4.7 will code, design, launch and translate a mobile app inspired by the original website and vice-versa. We’ve been using this internally a ton @shipper_now!

The agent decides whether a prompt needs text, a tool call, or a full visual component. When it picks visual, the cool stuff happens-- it generates real HTML/SVG/JS (charts, 3D scenes, simulations, whatever) and passes it to the frontend via a useComponent hook. That code gets rendered inside a sandboxed iframe, so it has no access to the parent window, cookies, DOM, or app state. The agent's code can only touch what's inside its own frame. Theming passes through automatically, so visuals match light/dark mode. And it all works inside your existing React/Next.js app. Standard HTML/JS in, sandboxed UI out. Test it out in our Open Generative UI app: opengenerativeui.copilotkit.ai Any questions?

Querés que tu agente deje de responder absolutamente todo? Para eso existen los "guardrails". Hay varias formas de hacerlo, algunos métodos más efectivos y otros menos... por ejemplo si lo dejás solo en el system prompt del agente es MUY fácil saltar porque al final el agente va a querer cumplir con la petición del usuario. Yo en Susi implementé este método y creo que es muy bueno y en la balanza de efectividad / complejidad es ideal: 👉Antes de que el modelo principal vea el mensaje, un modelo chico (Haiku 4.5) clasifica si el texto es de negociación o no. Si no lo es, el modelo principal solo dice "che, yo solo ayudo con negociaciones" y corta. 1. Llega un mensaje al chat → interceptamos antes de invocar al modelo principal. 2. Le pasamos el texto a Haiku con un schema rígido de Zod, con un system prompt cortito con lo que puede dejar pasar y lo que debería bloquear. Haiku te devuelve un allowed: true o false y nada más. No hay consumo de tokens de relleno, etc. El AI SDK (@vercel) tiene un generateObject + Zod te garantizan un output estructurado que sirve para este caso de uso. 3.a. Si allowed: true → el flujo sigue normal, entra al modelo principal con todas las tools de Susi. 3b. Si allowed: false → ni tocás al modelo grande. Spawneás un streamText cortito (max 100 tokens, temperatura 0.5) que solo dice algo tipo "Soy Susi, te ayudo con negociaciones. ¿Qué querés negociar?". Y se cierra. Haiku como modelo me pareció ideal para este trabajo porque tiene una latencia super baja que el usuario ni se entera del filtro, además es una tarea barata si se construye de esta forma. Se reduce el riesgo de prompt injection y mantiene el contexto limpio. Como último detalle, existe también un "fail open" que sirve para cuando el clasificador no funciona por cualquier motivo externo... esto deja pasar el mensaje igual y listo.

Google tiene un endpoint publico para obtener el favicon de cualquier dominio en PNG. Sin API key. Sin registro. Gratis. Lo usa Chrome internamente. Lo saqué del código fuente de Chromium 🧵

想用自己专属域名收发邮件，需折腾搭建完整邮件服务器，用第三方服务又担心数据隐私泄露。 最近 Cloudflare 团队开源的 Agentic Inbox，一个完全跑在 Cloudflare Workers 上的自托管邮箱客户端，还自带 AI 助手。 一键即可部署到自己的 Cloudflare 账户，就能用自定义域名收发邮件，每个邮箱独立隔离存储，附件存在对象存储里。 GitHub：github.com/cloudflare/age… 内置的 AI 邮件助手，收到新邮件会自动阅读并生成回复草稿，但不会擅自发送，必须你确认才行。 还能搜索历史邮件、整理对话，每个邮箱可以设置不同的提示词。 如果你手上有 Cloudflare 托管的域名，想要一个轻量、数据完全自主的邮箱方案，部署一下就能用。

🚨 Meta released their Ads MCP and CLI today – if you use Claude or ChatGPT you should install this asap (resources in comments). What makes this announcement so interesting is that it gives AI tools direct, authorized access to help manage your Meta Ads account through natural language. 1. Comprehensive reporting Pull detailed reports, surface performance trends, and quickly understand what is happening across campaigns. 2. Campaign management Create and edit campaigns, ad sets, and ads without manually clicking through Ads Manager. 3. Catalog management Create product catalogs, add product data, and troubleshoot feed issues faster. 4. Signal diagnostics Access signal health and quality insights so you can prioritize the parts of your setup that need attention. This is a huge step forward in agentic media buying. Will be testing this rest of the week!

You cannot convince me we don't already have AGI. Claude Code + ahrefs API is automating my SEO research in 10 minutes when it used to take months.

Siendo que estoy todo el tiempo probando tools de AI, solo tuve pocos momentos 'wow': - Cursor (Feb 25) - Claude Code (Ag 25) - Cloud Agents + SDD + Automations (Sep-Oct 25) y sin dudas de las últimas semanas tengo uno nuevo: - Hermes + Obsidian + LLM Wiki (skill de karpathy)

Meet Bilal. He used Bud to build an Instagram account to 3M views and 10k followers in 7 days: > Bud would post 20 curated posts per day across IG, YT, TikTok > Would then gather data and metrics to self improve the content > Had Bud also build a site and monetize the brand Try today bud.app

You can build interactive applications with gpt-realtime-1.5, so users can control app state more naturally with voice. Hi Chappy 👋