Salah Khenissi

Hey No bullshit, ClawdBot is about to completely change how you run your business (and your life). Welcome to the era of ultra-accessible, private AGI. The last 24 hours have been insane: viral tutorials, dirt-cheap setups, wild skills dropping left and right. Here’s the best stuff I saw : 👇🧵

Introducing the all new vibe coding experience in @GoogleAIStudio, feating: - One click database support - Sign in with Google support - A new coding agent powered by Antigravity - Multiplayer + backend app support and so much more coming soon! x.com/GoogleAIStudio…

Get featured for Google I/O! 🚨 We're putting together something special for @GoogleAIStudio users at I/O and we want to feature you and your creations! The ask: just reply with the best apps you've vibe coded and share your story The surprise will be worth it.

Thanks @OfficialLoganK for the hard work and the team behind it. Quick questions, how come the free quota has been reduced drastically. After less than 10 chats a message pop up saying quota limit exceed for the day😭 The worst part is that when I use my own api key it keeps saying the same.

Spent all day vibe coding, fixing bugs in AI Studio, and polishing the experience :) So much fun!

You’re looking at one of the most brilliant minds on Earth. Don’t miss a single second of this animation.🤯

bon c’est un truc assez fou mais un mec en Australie vient de concevoir un vaccin ARNm personnalisé contre le cancer pour sauver son chien en utilisant ChatGPT & AlphaFold sachez déjà qu’il a 0 formation en biologie et malgré tout ça ça a marché…la tumeur a réduit de moitié, les chercheurs en génomique sont sous le choc & je pense que cette histoire est infiniment + importante qu’elle en a l’air voilà ce qu’il a fait concrètement il a payé $3000 pour séquencer l’ADN de la tumeur de son chien, il a donné les données à ChatGPT pour identifier les mutations d’intérêt, ensuite il a utilisé AlphaFold pour trouver la structure 3D des protéines mutées & identifier des cibles thérapeutiques, puis il a conçu un vaccin ARNm sur mesure ciblant spécifiquement les néoantigènes de la tumeur…le tout depuis son ordinateur portable mdr le professeur de génomique qui a reçu la demande au départ pensait que c’était une blague, quelques mois + tard il regarde les résultats et dit si on peut faire ça pour un chien pourquoi on le fait pas pour tous les humains atteints de cancer et je pense que c’est là qu’il faut comprendre VÉRITABLEMENT ce qu’AlphaFold représente vraiment (j’ai fait de nombreux threads sur ce programme dans le passé) parce que la plupart des gens en ont entendu parler sans saisir la profondeur de ce qui se passe je vous explique pour ceux qui découvrent le sujet pendant des décennies pour comprendre la structure 3D d’une seule protéine il fallait des mois voire des années de cristallographie aux rayons X ou de cryomicroscopie électronique + des labos entiers mobilisés sur une seule molécule MAIS AlphaFold 2 a résolu ce problème en prédisant la structure de pratiquement toutes les protéines connues soit + de 200 millions de structures, ce qui lui a valu le prix Nobel de chimie 2024 par ailleurs, AlphaFold 3 sorti en 2024 est allé encore +++ loin, là où AlphaFold2 prédisait la structure d’une protéine isolée, AlphaFold 3 prédit les interactions entre protéines, ADN, ARN, petites molécules & ligands dans un système unifié, en gros il modélise comment une molécule médicamenteuse va se fixer sur une protéine cible avec une précision améliorée …et il le fait en quelques heures au lieu de quelques années et c’est exactement ce que ce mec a exploité pour son chien, il a utilisé AlphaFold pour voir la forme 3D des protéines mutées de la tumeur & comprendre comment un vaccin ARNm pourrait apprendre au système immunitaire à les rreconnaître & les détruire d’ailleurs ce qui me fascine c’est ce que ça annonce pour la suite dans la mesure ou isomorphic labs la filiale de Deepmind dédiée à la découverte de médicaments a déj signé des partenariats de plusieurs milliards avec Eli lilly et Novartis, et les premiers médicaments entièrement conçus par IA grâce à AlphaFold 3 devraient entrer en essais cliniques humains prochzinement on parle de candidats en oncologie et en immunologie qui ont été conçus par design rationnel, c’est à dire que l’IA a littéralement dessiné la molécule pour qu’elle se fixe parfaitement sur la cible au lieu de tester des millions de composés au hasard comme on le fait depuis 50 ans!!!!! et le mouvement s’accélère car Deepmind a rendu AlphaFold3 open source fin2024, la communauté scientifique a immédiatement construit dessus, des modèles comme OpenFold3 soutenu par Amazon et Novonordisk, des startups comme Recursion qui développent des versions spécialisées, on entre dans l’ère du «labo autonome » (sujet dont j’ai parlé l’année dernière) où en gros pour vulgariser l’IA conçoit une molécule, des robots la synthétisent et des plateformes de screening la testent sans intervention humaine et je crois que le prochain horizon c’est la modélisation temporelle, aujourd’hui Alphafold prédit la forme statique d’une molécule mais demain on prédira comment elle bouge & vibre au fil du tps à l’intérieur d’une cellule vivante et c’est extrêmement fascinant pour l’avenir! voilà, voilà ;)

Now you can create beautifully designed and Google Slides that match your deck’s style instantly. Turn sketches into editable charts and messy notes into polished layouts in a click. goo.gle/3OW1BJu

🚨BREAKING: Google DeepMind dropped a paper that changes how AI agents will work forever. It's called Intelligent AI Delegation and it's a full framework for AI agents to safely hand off tasks to other AI agents and humans. No brittle heuristics. No ad-hoc handoffs. No accountability black holes. Here's how it works: The framework runs across 5 core pillars: → Dynamic Assessment: agents evaluate each other's real-time capabilities before delegating → Adaptive Execution: mid-task re-delegation when things go wrong → Structural Transparency: every step is auditable, not a black box → Scalable Market Coordination: decentralized bidding systems where agents compete for tasks → Systemic Resilience: hard stops before cascading failures hit Here's the wildest part: They borrowed concepts from aviation and medicine to explain why AI agents blindly follow bad instructions the same way junior doctors don't question senior surgeons. And they built a fix for it. This isn't just theory. They mapped it onto real protocols like MCP, A2A, and AP2 and showed exactly what's missing from each one. This is the infrastructure layer the agentic web has been waiting for. Full paper link in first comment 👇

🚨 BREAKING: Someone built a swarm of thousands of AI agents with real memories and personalities and used it to predict the future. MiroFish is a universal swarm intelligence engine. And the live demos are scarily accurate. Here is what it actually does: → Spins up thousands of autonomous agents simultaneously → Each agent has its own memory, personality, and behavior → Feeds on real-world data powered by GraphRAG → Predicts markets, public opinion, and narrative outcomes → Simulates how crowds think before it happens The live demos are what got people. Scarily accurate is the phrase everyone keeps using. 17,300 stars. +2,907 in a single day. It's 100% free and open source.

🔥🚨BREAKING: You can now turn any screenshot into a fully editable diagram in seconds. It's called Edit Banana. → Upload a screenshot of any flowchart, architecture diagram, or technical schematic → SAM 3 segments every shape, arrow, and icon with pixel-perfect accuracy → Multimodal LLMs scan the image in 4 passes so nothing gets missed → OCR extracts all text and converts formulas to LaTeX → Outputs a native .drawio XML file every element draggable and editable No more redrawing diagrams from scratch. No more screenshotting → losing all editability. Upload image. Get editable DrawIO in seconds. 566 stars. Apache 2.0 license. (Link in the comments)

🚨BREAKING: Someone just open-sourced a headless browser that runs 11x faster than Chrome and uses 9x less memory. It's called Lightpanda and it's built from scratch specifically for AI agents, scraping, and automation. Not a Chromium fork. Not a hack. A completely new browser written in Zig. Here's why this changes everything for AI builders: ↓

Lip sync can be hard. We fixed that 💥 Introducing Speak on Freepik Upload a visual, add your own audio or a script, and get a lip-synced talking video in seconds → Custom voices in 30+ languages → Up to 5 min talking video → A lip sync tool, made simple Available now

Introducing the new Gemini powered Docs, Sheets, Slides, and Drive experience featuring AI Overviews, fulled editable AI made slides, and new grounding sources to make writing docs context aware 📃 Available today to G1 Pro and Ultra users : )

🚨 Someone built a fully open-source Figma alternative that designers and developers actually love and it's been quietly getting better for years. It's called Penpot and it's not a watered-down clone. It's a full design and code collaboration platform built on open standards SVG, CSS, HTML with no vendor lock-in and no $45/month seat fees. Here's what it actually does: → Full vector design tool with real-time multiplayer collaboration → Interactive prototyping built in no plugin needed → Inspect mode spits out ready-to-use SVG, CSS, and HTML code → Native design tokens first design tool to do this properly → CSS Grid Layout support that Figma still doesn't have natively → Component systems, variants, and design systems at scale → Plugin API to extend and integrate with anything → Self-host on Docker, Kubernetes, or use their cloud for free → Webhooks + API for full dev toolchain integration Figma charges $45/seat/month for teams. This is free. Your server. Your designs. Your code. Zero handoff drama. The Figma monopoly just met its open-source match. 100% Open Source. (Link in the comments)

13,500 GitHub ⭐ in just a few days for a newly open-sourced project. 🤯 And for good reason. Paperclip by @dotta isn't just another AI wrapper, it's a powerful Node.js backend and React dashboard designed to tame the wild west of AI agents. You need Paperclip if you: 🧠 Have 20 Claude tabs open and can't remember who is doing what. 🤖 Rely on different types of agents (Codex, OpenClaw) and need them to collaborate. 📈 Want agents working for you 24/7 with fully auditable work and costs. 🚀 Want to build an autonomous business, not spend your days managing pull requests. Here is why the entire dev timeline is obsessed with it right now: → It forces structure: Budgets, org charts, and goal alignments are built right in. → Deep integration: It coordinates tasks atomically with models like OpenClaw. → Removes the mental load: You manage the vision, the agents handle the execution. Any agent, any runtime, with seamless coordination and zero lock-in. 🔥 Also an MIT-licensed, fully open-source project that's moving incredibly fast with a great Discord community. Link to the repo in 🧵↓

🚨 BREAKING: SOMEONE JUST BUILT A FULL AI AGENCY YOU CAN RUN FROM YOUR LAPTOP. ENGINEERS, DESIGNERS, MARKETERS, ALL FREE. Most people are still prompting one AI to do everything. This person structured it like an actual company. And the internet lost its mind. Here is the setup that should concern every founder still hiring freelancers. ↳ A developer just dropped Agency Agents on GitHub. 55 specialized AI agents across 9 divisions, each with their own personality, workflows, and deliverables. ↳ Not generic prompts. Not chatbot wrappers. Actual role specific agents that coordinate with each other to ship real work. It crossed 7600 stars and 1200 forks. Born from a Reddit thread. Built in months of real world iteration. The insight that changes how you think about AI forever. Every AI power user hits the same wall. One giant agent trying to do everything produces mediocre everything. This repo fixes that by doing what every successful company already figured out. Specialize the roles. Define the responsibilities. Build the handoffs between them. Your frontend agent does not write Reddit copy. Your growth hacker does not review API tests. Each agent stays in its lane and goes deep. What is actually inside 📈 ↳7 engineering agents covering frontend, backend, mobile, ai, devops, prototyping, and senior development ↳7 design agents, including a whimsy injector focused on delight and micro-interactions ↳8 marketing agents with dedicated specialists for twitter, tiktok, instagram, reddit, and the app store ↳3 product agents for sprint planning, trend research, and feedback synthesis ↳6 spatial computing agents built specifically for vision pro and webxr ↳50+ requests in the first 12 hours after posting on reddit One solo developer shipped what agencies charge $20k a month to provide. And here is the part that should keep agency founders up at night. The repo includes real world use cases. A startup MVP built by 5 coordinated agents. A full marketing campaign launched across 5 platforms simultaneously. An enterprise feature shipped with quality gates, design review, and legal compliance checks. All of it documented. All of it free. All of it running inside Claude Code in under 5 minutes. The question is no longer whether AI can replace agency work. Someone already proved it can.

🚨 ByteDance (the company behind TikTok) just open sourced an AI SuperAgent that can research, code, build websites, create slide decks, and generate videos. All by itself. It's called DeerFlow. Give it a task that would take you hours. It breaks it down, spawns sub-agents, and delivers the finished result. Not a chatbot. Not a copilot. An AI employee with its own computer, filesystem, and memory. Here's why this is different from every other AI agent: It has its own sandbox. A real isolated Docker container with a full filesystem. It reads files, writes files, executes code, runs bash commands. It doesn't just suggest things. It actually does them. No other agent framework gives the AI its own actual computer. Here's what it can do out of the box: → Deep research across the entire web with cited sources → Generate full reports with charts and analysis → Build complete websites and web apps → Create slide decks from scratch → Generate images and videos → Run Python scripts in its sandbox → Spawn sub-agents that work in parallel on different parts of a task → Remember your preferences, writing style, and workflows across sessions Here's how it handles complex tasks: You say "Research the top 10 AI startups in 2026 and build me a presentation." DeerFlow's lead agent breaks that into sub-tasks. One sub-agent researches each company. Another collects funding data. Another finds competitor analysis. They all run in parallel. Results converge. A final agent builds the slide deck with generated visuals. One prompt. Multiple agents. Complete deliverable. Here's the wildest part: It started as a simple deep research tool. Then the community started using it to build data pipelines, automate content workflows, spin up dashboards, and create full applications. ByteDance realized it wasn't a research tool anymore. It was a SuperAgent harness. So they rewrote it from scratch. DeerFlow 2.0 hit #1 on GitHub Trending on launch day. Works with GPT-4, Claude, Gemini, DeepSeek, Ollama, or any OpenAI-compatible API. Skills load progressively. Only what the task needs, when it needs it. No context window bloat. 22.7K GitHub stars. 2.7K forks. 1,531 commits. Hit #1 on GitHub Trending. 100% Open Source. MIT License.

Converts static images into editable diagrams and slides github.com/BIT-DataLab/Ed…

🚨 BREAKING: Someone just open-sourced the operating system for zero-human companies. It's called Paperclip. Think of it as the company layer on top of your AI agents. If OpenClaw is an employee, Paperclip is the entire company. What's inside: → Bring any agent (Claude Code, Codex, Cursor, OpenClaw) with real reporting lines → Give them org charts, titles, budgets, and goals → Monthly budgets per agent when they hit the limit, they stop. No runaway costs → Full ticket system with tool-call tracing and immutable audit logs → Agents run 24/7 on heartbeats while you monitor from your phone Instead of having 20 Claude Code tabs open with no idea what's happening… One deployment. One dashboard. Your agents run the company while you sleep. 1.4K stars. MIT License. 100% Opensource.

A Pakistani animation team produced an animated film recounting the tragedy of the schoolgirls martyred at Minab School by an American-Israeli attack.

I accidentally discovered how to compress a semester of learning into 48 hours. A grad student at MIT showed me his NotebookLM setup. I thought he was just organized. Then I watched him pass a qualifying exam on a subject he'd never studied before. Here's exactly what he did: First: he didn't upload a textbook. He uploaded 6 textbooks, 15 research papers, and every lecture transcript he could find on the subject. Then he asked NotebookLM one question: "What are the 5 core mental models that every expert in this field shares?" Not "summarize this." Not "explain this topic." Mental models. The stuff that takes professors years to develop. But the next part is what broke my brain. He followed up with: "Now show me the 3 places where experts in this field fundamentally disagree, and what each side's strongest argument is." In 20 minutes he had a map of the entire intellectual landscape of the field: the debates, the consensus, the open questions. Most students spend a full semester just figuring out what those debates even are. Then he did something I've never seen before. He asked: "Generate 10 questions that would expose whether someone deeply understands this subject versus someone who just memorized facts." He spent the next 6 hours answering those questions using the source material. Every wrong answer triggered a follow-up: "Explain why this is wrong and what I'm missing." By hour 48, he could hold a conversation with his thesis advisor without getting destroyed. The tool didn't change. The questions did. Most people treat NotebookLM like a fancy highlighter. These students are using it like a private tutor who has read everything ever written on the subject. The difference between a semester and 48 hours isn't the amount of content. It's knowing which questions to ask.