Forbes's Recent LinkedIn Posts

AI NEWS: Grok 3 drops tomorrow night—xAI’s billion-dollar bet on scaling. Reminder: xAI built Colossus, the world’s most powerful AI training cluster (100,000+ NVIDIA H100s in just 122 days) to train Grok 3. This comes after DeepSeek-R1 tanked the stock market by delivering a strong model on a *reported* $5M of compute, raising doubts about whether massive training runs are still necessary. Grok 3 is the counter-test: Does sheer scale still dominate, or are we entering the era of smarter, cheaper AI? If it delivers, it’s a big day for NVIDIA. If it underwhelms, DeepSeek’s efficiency argument just got stronger.

Amazon just unveiled the new Alexa+ powered by Amazon's AI and Claude. It's like ChatGPT Voice taken to the next level with personalization, memory of past conversations, and of course, agentic action features. Pricing starts at $19.99/month but is free for all Amazon Prime members. Most importantly, this means ~100M+ people are about to have their first experience with real AI-powered voice agents soon. This could be another 'ChatGPT moment' for consumers outside of the tech community. (Let's just hope it runs more smoothly than the launch of Apple Intelligence) Video taken live by Max Weinbach

Big news: I'm officially announcing two new newsletters: Robotics and Tech. Right after our AI newsletter hit 1,000,000 subscribers, we quietly launched both to our readers: > The Rundown Tech: now at 250k+ subscribers > The Rundown Robotics: now at 140k+ subscribers This brings The Rundown’s total active subscriber count to over 1.5 million while maintaining some of the highest engagement rates in the industry. We were nervous to launch more newsletters, unsure if it might pull focus from our core AI focus. That’s why we wanted to get feedback from existing readers first. But the response has been overwhelmingly positive, with an avg. 95% 5-star feedback rating on thousands of votes (please keep sharing your feedback 🙏). So today we're officially announcing it to the public! If you want a bi-weekly, 5-minute curated email on Robotics and/or Tech -- with the same need-to-know, no-fluff approach, sign up below. We’ve brought on expert writers and journalists to lead each vertical, and we’re just getting started.

We're hiring someone to get paid to play with AI tools all day. (Not kidding!) The Rundown is growing fast (2M+ readers), and we're looking for another writer who's obsessed with AI to join us. Your job: > Test every new AI tool that drops > Write guides that actually help people use them > Explain why some AI is useful and some is garbage Basically, you'll be the friend everyone texts when they hear about a new AI thing. Here's what we're looking for: > Experience with copywriting, journalism, or content creation (around AI is a plus) > Exceptional research skills and you know what's happening in AI before Twitter does > Deep AI Intuition - knowing when (and when not) to use AI The role involves publishing 5 times per week. Your writing will reach millions of people a day. If this sounds fun rather than stressful, you're probably who we're looking for. Link to apply here with more details: https://lnkd.in/gG6W6_r5

Honored to be recognized on the Forbes 30u30 2026 list for The Rundown! Crazy to think I dropped out just a couple years ago. Even crazier to think I’m now educating millions of people as that same kid who dropped out. Grateful for the opportunity… So much more to build!

My favorite AI workflow lately is my thought-to-content pipeline. I just go on walks, have a good content idea, ramble it, and have an optimized post in my writing style without typing. It's super simple: 1. Download an AI-powered voice dictation app to your phone (I use Wispr Flow) 2. Go on long walks and let ideas flow - when you get a good one, open Wispr Flow and ramble your thoughts (doesn't need to be perfect) 3. Notes auto-save. These become the core ideas for posts later 4. Open Claude and create a new Project called "Post generator" 5. Use this prompt: "I’m going to provide you with my own written material, and your task will be to understand and mimic its style. You'll start this exercise by saying "BEGIN.” After, I'll present an example text, to which you'll respond, "CONTINUE". The process will continue similarly with another piece of writing and then with further examples. I'll give you unlimited examples. Your response will only be "CONTINUE.” You're only permitted to change your response when I tell you "FINISHED". After this, you'll explore and understand the tone, style, and characteristics of my writing based on the samples I've given. Finally, I'll prompt you to craft a new piece of writing on a specified topic, emulating my distinctive writing style" 6. Now's the fun part: Go to Analytics and copy your top posts 7. Paste your best-performing posts into Claude repeatedly until it says "FINISHED" 8. Take your voice notes, paste them into your trained Claude Project, prompt "make a post in my writing style" 9. Post is ready to go. Polish and edit slightly *if* needed. The AI is trained on how you actually write, not generic content. Your voice notes capture your real, raw thoughts without the friction of typing. I have my best ideas while walking. If I try to write them in my notes app mid-walk, I forget halfway through. Voice dictation captures everything as I ramble. Game changer for turning scattered thoughts into polished posts!

Using Perplexity Comet to chat with YouTube videos is game-changing. It's an AI tutor for every educational video. Whenever I don't fully understand something, I pause the video and do a mini rabbit hole session. I've been doing this while reading with ChatGPT Voice too (eg asking it to explain certain phrases), and now using Comet for videos is just as useful. It's incredibly useful for understanding complex topics more deeply. With memory and personalization, it'll get even more useful, but we're *already* in the early innings of having AI tutors for everything. Imagine how much smarter 10 year olds are going to be in just a few years!

The new ChatGPT connectors are really useful! Chat can now access Gmail, Google Cal, and Drive, so it can: -Skim unread emails & give a summary -Summarize threads + draft replies -Pull key info from old convos -Do meeting prep + agendas Before I get into some prompts you can copy, quick reminder that enabling connectors gives ChatGPT view access to your private data. I'm an AI power user and it's part of my job to test everything, so personally, I make the sacrifice for the extra features. But please be careful if your job is data-sensitive! 1. Email Intelligence Get ChatGPT to skim your recent unread emails, give you a summary, and rank them in importance: "Summarize all my unread emails in Gmail that I received over the past 48 hours. Rank it in order of importance." 2. Search & Data Extraction Get ChatGPT to find threads and draft in your tone "Find all my emails with [Jennifer], then compare it to my most recent email from [Jennifer]. Give me the takeaways in bullet points. Then, draft a short reply in my tone based on that context." 3. Inbox Memory Get ChatGPT to pull key info from old convos and compare to recent emails: "When is OpenAI DevDay this year and when was it during the previous years (2023, 2024)? What makes this year different. Use a mix of internet information and my email inbox context" 4. Meeting Briefs / Agendas Get ChatGPT to look at your Google Cal and your email to prep for upcoming meetings "Look at my next Google Calendar event. Give me a rundown of the person and the event with context from my Gmail so I come prepared. Also suggest a meeting agenda." I'm just the surface here on how you can use these connectors, but they've been very useful so far as someone who spends hours in my inbox every day If you enable them, I would love to hear your use cases! Will be featuring the top responses in The Rundown (1M+ readers!)

It's only been 1 day of CES 2025, and the announcements have already been incredible. The 10 most impressive reveals of CES 2025 so far: 1. A 360° AI-powered body scanning health mirror that can scan your heart, weight, and metabolic health https://lnkd.in/gwPCNXRc 2. Roborock's Saros Z70: A robotic vacuum that has a mechanical arm for picking up objects in the way of cleaning the floor https://lnkd.in/gFxiM9mA 3. Halliday Glasses: Smart glasses with a 3.5-inch internal monochrome display, equipped with an AI agent that can listen to conversations, answer questions during meetings, and do live translation https://lnkd.in/gMU8HUyC 4. Project DIGITS by NVIDIA: A $3000 personal supercomputer that's 1,000x the power of an average laptop https://lnkd.in/ghxkGTrZ 5. BYD's supercar, Yangwang U9, jumping 6 meters forward (over potholes) https://lnkd.in/g3Q2Cb2e 6. A pen with three cameras at the tip that turns any surface of the world into a canvas with notes synced and recorded to your phone https://lnkd.in/gNPAKWUM 7. NVIDIA Cosmos: An open source, open weight Video World Model designed for the upcoming age of robotics source https://lnkd.in/gUZhrx2h 8. Portalgraph: A 3D projector that projects VR space into the real world https://lnkd.in/gcxRkm9G 9. Samsung Vision AI TVs that come with real-time translation, the ability to adapt to user preferences, AI upscaling, and instant content summaries https://lnkd.in/gNSr_9Ev 10. NVIDIA GB200 NVL2: A datacenter superchip with 72 Blackwell GPUs, 1.4 exaFLOPS, and 130 transistors https://lnkd.in/gUaSqs5k That's it for day 1. It's going to be a wild week! I'll be covering everything else that comes out at #CES2025, follow me Rowan Cheung to stay updated or join my free AI newsletter, and I'll send you a quick email rundown with everything you need to know: https://lnkd.in/g7jyJiPR

Google DeepMind just announced Genie 3, the FIRST real-time interactive world model that creates worlds from a text prompt. We’re entering the era of infinite AI training environments. The video below shows a controllable environment generated by Genie 3 in real time. Insane. We're entering an era where world models like Genie 3 enable AI agents and robots to learn from their own experiences in simulated real-world environments. For example: > Anyone can create entirely imagined virtual worlds from a prompt > AI agents learning and improving from their own experiences in these worlds > Robots training and improving over time in simulated environments before touching the real world This is the convergence of world simulation, AI training, and creative expression all happening in real time. Genie 3 technical breakthroughs: > Real-time interaction at 24 frames per second > Visual memory extending back one full minute > Environmental consistency for several minutes > "Promptable world events" - you can change weather, add objects, introduce characters mid-session DeepMind also tested generating worlds to train SIMA (their AI agent for 3D environments announced in 2024). They instructed the agent to achieve distinct goals while Genie 3 simulated the future based purely on the agent's actions -- no awareness of goals needed. AI progress 📈

Google just released Gemini 3, its most intelligent AI model yet! I caught up with Demis Hassabis, CEO of Google DeepMind, to ask about: -Gemini 3 and Google’s AI strategy -Google's new Antigravity tool -Medical-grade AI Here's what he said: 1/ Gemini 3 significantly gaps up on reasoning. But beyond intelligence, Demis says they're "going deep into personalization, memory and context". Personalization and connections across Gmail, Calendar, and Google's ecosystem will make AI assistants "truly indispensable", pre-AGI Clip: https://lnkd.in/gAfGsMp8 2/ With Gemini 3, Google also is releasing Antigravity -- an AI coding platform that operates across the editor, terminal, and browser. The differentiation from other AI coding tools is they're "reimagining the IDE from an agent-first perspective", not just another plugin Clip: https://lnkd.in/gci2nESP 3/ Demis revealed they're building medical-grade AI for "poor parts of the world" without primary healthcare. Android devices are already "lifelines" there -- Google plans use that reach to deliver care. Gemini 3 is the "foundation stone" for eventual doctor-level AI assistants Clip: https://lnkd.in/gqQK8njP 4/ There's been questions around how AI will be tied into existing Google products like Search Demis says that you can now "think of Google DeepMind as being the engine room of Google" They're now powering up EVERY big Google product from Maps to YouTube to Search to Workspace Clip: https://lnkd.in/gBvn7T2G 5/ Demis' dream digital coworker vision: an AI assistant "useful every moment of your life" for both work and leisure. It follows you across all devices - computer, phone, even smart glasses - understanding the physical world around you through multimodal capabilities. Clip: https://lnkd.in/g2EMNn5U It's always an honor speaking with Demis Hassabis! In the full interview, we chat more about: -Building proactive AI healthcare -The skill of "learning how to learn" -Secret internal models and research Listen in on Apple Podcasts: https://lnkd.in/gh8PYqcp Listen in on Spotify: https://lnkd.in/gsHq2Wnf

I asked Sam Altman if he's worried about watermark removers on Sora. Everyone's seen the unfiltered Jake Paul videos with no watermark (generated by Sora) going viral on socials... So what's the pull for creators to enable face cameos with that risk? Sam's take: open source deepfake models are coming whether we like it or not, so releasing Sora early with guardrails helps society adapt before the flood hits. "One of the reasons we release technology like this is we see something coming and know that in some number of months or years it's gonna be widely available through open source models." Anybody will be able to make a video of you doing whatever they want with publicly available footage of you on the internet. Society will adapt... but one of the ways we've found to help society with these transitions is to release it early with guardrails so that society and technology have time to co-evolve. And that worked with text, I think video's going to be harder because video hits in a way that text doesn't always. But I think we'll learn to adapt, and we'll learn very quickly that there will be a lot of fake video of you on the internet, generated with no watermark by some open source model, and not that easy to trace. That's just gonna happen, so getting society inoculated to that probably has some value."

Today, we're launching the world’s first personalized AI education platform with tailored training, use cases, and live workshops for your career: We've all heard “use AI or be replaced.” But beyond ChatGPT prompts and expensive AI consultants, there’s no real roadmap to learn AI for your job, skill level, or daily workflow. It's all broad. So we built a personalized AI learning roadmap -- by crowdsourcing the best workflows across the 1M+ AI early adopters in our community and turning them into actionable playbooks. Here's what's inside: 1. Industry-specific AI certificate courses across 16 tech verticals 2. 300+ real-world AI use cases filtered by what's relevant to you 3. Live, expert-led workshops with our in-house AI research team + AI tool partners 4. A personalized community and network of AI-minded professionals in your industry 5. $1,000+ in exclusive discounts on top AI tools That means whether you're in marketing, dev, ops, or any of the 16 industries we cover, you’ll only see what’s actually working in your field. No more noise. We'll show you the vetted AI tools and workflows actually being used in the real world by the early adopters in your industry. We believe the future of work belongs to those who combine AI with domain expertise. And now, there’s no excuse not to start. The first 14 days are 100% free, enough to earn your first certificate 👇

What a wild couple days at Microsoft Build and Google I/O. In between the announcements, I had the incredible opportunity to sit down for interviews with Satya Nadella (CEO @ Microsoft) and Demis Hassabis (CEO @ Google DeepMind). Grateful for how far The Rundown has come - can’t wait to share these conversations!

OpenAI just released GPT-4.5, the startup's largest AI model to date! Available now to Pro users ($200/mo tier) and developers on paid tiers via API. GPT-4.5 is a leap forward in scaling unsupervised learning. Reasoning models (like o1 + o3-mini) are great for STEM/logic... But unsupervised learning increases world model accuracy and intuition — leading to better EQ and a broader knowledge base (and less hallucinations). OpenAI says early testers reported GPT 4.5 interactions feeling more natural and intuitive. In human preference testing: — 70.8% preferred GPT-4.5 for professional tasks — 58.4% chose it for creative work — 55.9% picked it for everyday queries With the focus on unsupervised learning over reasoning, 4.5 isn't a step up from previous models on math or science. However, it does surpass o3-mini and o1 on SWE-Lancer, OpenAI's recently released freelance coding task benchmark. GPT-4.5 currently supports search, file/image uploads, and canvas for writing/code — no multimodal features like Voice Mode, video, or screensharing yet. OpenAI expects to roll out the new model more broadly to Plus and Team plans next week.

Microsoft just unveiled Muse — an AI model that can generate minutes of cohesive gameplay from a single second of frames and controller actions. The implications for gaming are absolutely massive. Muse is the first World and Human Action Model (WHAM), and the scale of training is mind-blowing: — Trained on 1B+ gameplay images — Used 7+ YEARS of continuous gameplay data — Learned from real Xbox multiplayer matches From a single second of gameplay + controller inputs, Muse can create multiple unique, playable sequences that follow actual game physics, mechanics, and rules. The version shown in research was trained on just a single game (Bleeding Edge). We're watching the beginning of a major shift in how games are created. Game development traditionally takes months/years of character design, animation, and gameplay testing. Scaled up models like Muse could eventually compress this cycle from months to minutes. Microsoft's research is published in Nature today, and researchers will also be releasing Muse's model weights, training data samples, and sharing the testing interface. Another win for open-source! Hot take: We'll see the first fully AI-assisted indie game hit top 10 on Steam within 18 months. Not because AI makes better games, but because it lets creative people iterate 100x faster.

He dropped out of university because lectures didn't work. Now, The The Rundown AI founder Rowan Cheung is proving traditional education is overrated. Cheung learned to code on YouTube, then launched a media company that now boasts: - 2 million newsletter subscribers - Interviews with Mark Zuckerberg and Satya Nadella - $7 million projected revenue His goal: Simplify complex AI for the everyday person. Read more: https://lnkd.in/gufHxz2P #ForbesUnder30 (Photo: Sebastian Nevols for Forbes)

My AI avatar just hit 100,000 followers. The videos below isn't actually me, it's an AI avatar of my face/voice. Some raw thoughts on this experiment: 1) People don't care it's AI-assisted. Good content is good content, end of story. 2) Critics are generalizing. People calling AI content "slop" are judging the majority of AI-generated content (which, candidly, often is slop) without even realizing the good content *made with the help of AI*. 3) This strategy has 10x'd my content output. My team and I could never produce daily videos without my AI avatar since I'm way too busy as a CEO or writing. 4) In addition to increasing my content output, it's increased quality too, by allowing me to focus on research and writing (what I'm way better at). 5) My AI avatar is *way* better than me on camera and creates better content than I would. Back to point #1 - that's all that matters. 6) “Good” AI-created content needs human fine-tuning and taste. It's lazy to just try to automate everything. Spend the extra hours adding the extra 5% and you'll win. We still research, write scripts, and edit everything. 7) The whole point was finding a way to distribute my writing/tweets through video without adding more to my workload—it worked. I'm reaching new audiences that I've never reached before. 8) Industry hypocrisy is nuts. The same creators publicly calling AI-assisted content "slop" are privately DMing me asking how to do it themselves (you know who you are 😂). Top creators are already leveraging the tools and multiplying output and quality -- whether through an avatar or thumbnails, scripting, and editing. And the content explosion is just getting started. My best advice to take advantage of this opportunity would be to start experimenting every day with AI and finding workflows that work for you. I'd start here: - Claude 3.7 Sonnet for editing and polishing scriptwriting and title generation (paste your scripts + prompt "make it better" or "make it more viral") -GPT4o for thumbnails and image b-roll -KLING for video b-roll -Heygen or Synthesia for AI avatar cloning -ElevenLabs for voice cloning -Descript (if you're new to editing) In the end, I think if you’re not using all the available tools to create better and more content as a creator, you're doing your audience a massive disservice. It's time to start embracing the tools and serving your audience better!

Mind blown. I was an early tester of Claude 3.7 Sonnet (which just dropped), and it created this Minecraft clone in one prompt: And the best part? It became instantly playable with Claude's Artifacts feature. This is truly prompt-to-game, with not a single line of code written. What's different about Claude 3.7 Sonnet is that it uses a hybrid reasoning approach. Basically, it can produce instant responses when needed or think step-by-step like OpenAI o3 or DeepSeek R1. The "Extended Thinking Mode" has huge improvements in math, physics, instruction-following, coding, and more. It also is now officially the best coding AI model in the world.

I've been fascinated by the potential of AI tutors/agents. And it feels like we’re finally seeing that potential become reality (!): — AI-powered private school student scores just soared to the top 2% in the country — In Nigeria, students using GPT-4 as a tutor learned two years of material in just six weeks — A recent Harvard study found students with AI tutors learned twice as much in less time than those without If you've successfully implemented an AI tutor in your curriculum, education plan, or program, I want to hear from you for some research I'm doing. Whether it's a school, a startup, or personal tutoring, I'm looking to ask a few questions via a short call or email Q&A. I'll compensate you for your time and give you credit if I end up sharing your story. Please DM me with any stats, case studies, or experiences!

That's a wrap for day 2 of the world's largest consumer tech event, CES 2025. The top 10 tech and gadget reveals from day 2: 1. A stretchable Micro LED display that turns 2D into 3D by Samsung https://lnkd.in/givm49PR 2. A multitasking household robot that does everything from vacuuming, organization, air purification, monitoring pets, and even delivering you food while you sit on the couch by SwitchBot https://lnkd.in/gC4PX94b 3. An immerse location-based entertainment concept that allows players to use flashlights and guns in an LED environment by Sony https://lnkd.in/gwg83pBv 4. The world's first 'rollable' PC that allows you to expand your screen upwards by Lenovo https://lnkd.in/gND-fM66 5. A neural wristband that turns your hand into a mouse or keyboard by Mudra Link https://lnkd.in/gyCc2F59 6. BMW’s Panoramic iDrive, a 3D heads-up display that projects information across your entire windshield https://lnkd.in/g76Sbc-d 7. A massive, 11-inch handheld gaming PC by Acer https://lnkd.in/gJ3QXap3 8. Honda's newly debuted 0 SALOON electric car https://lnkd.in/gsx6KbcH 9. Unitree's G1 humanoid robot in-person demo (that was also spotted shaking attendees hands) https://lnkd.in/gWmw3XdF 10. A budget-friendly ($100) fitness watch with voice control, health metric tracking, and offline maps by Amazfit https://lnkd.in/giDBxAC2 That's it for day 2. It's going to be a wild week! I'll be covering everything else that comes #CES2025, follow me Rowan Cheung to stay updated.

BREAKING: OpenAI just launched ChatGPT Agent I had early access, and the agent built me a complete early retirement plan: > Found local tax laws (Vancouver) > Analyzed average monthly spend rates > Calculated savings needed to retire at 30 > Researched optimal investment allocations > Found tax optimization strategies I'd never heard of > Built multiple FIRE scenarios > Created a downloadable presentation with results This would've cost me $5,000+ from a financial advisor and taken weeks. The new feature essentially allows ChatGPT to think, plan, and execute complex tasks on its own virtual computer while you do other things. It does this by bringing these three ChatGPT capabilities into one unified system: - Operator's web interaction skills - Deep Research's information synthesis - ChatGPT's conversational intelligence I think with ChatGPT Agent now, and especially as it gains access to more tools, we're finally going to see the rise of a new AI skill category in *Agent Management* Agents are finally becoming capable of doing real work autonomously, so anyone who learns how to effectively orchestrate agents will have a huge advantage.

BREAKING: OpenAI just launched ChatGPT Operator, a new AI agent that can autonomously take action across the web on your behalf. I got early access, and here are 9 impressive use cases I found: 1. Ordering dinner ingredients to my house based on a picture and a recipe (video attached) 2. Planning a weekend trip based on hidden gems, my budget and interests. At 0:06, ChatGPT Operator was blocked from Reddit but then decided to just do a Bing search with "Reddit" at the end: https://lnkd.in/gqqDCGV3 3. Crypto investment research based on tokens that are actually worth looking into. ChatGPT Operator got hit with a "Are you human" CAPTCHA, then pinged me to take control to confirm: https://lnkd.in/gtfpZV99 4. Booking a one-way flight from Zurich to Vienna using the Booking integration: https://lnkd.in/gZypshPg 5. Scheduling an appointment with my barber after looking at my Google Calendar schedule/availability: https://lnkd.in/gTVNnmgz 6. Researching a good birthday gift for my mom based on what she likes and finding the best deals for those items across the internet: https://lnkd.in/gTJXZvHg 7. Booking a one-time house cleaner for my home through the Thumbtack integration based on my budget: https://lnkd.in/gQAMx6WX 8. Finding the best/cheapest health insurance coverage in Switzerland (an interesting one since most prices are not publicly available and gated behind a meeting): https://lnkd.in/gh3WDq26 9. Finding a top-rated dog walker in Vancouver BC and presenting me with 3 top options: https://lnkd.in/gQNpgKgY Overall, I was very impressed by the research preview of Operator. I loved that it can do tasks for me as I do other work, and simply ping me when it needs me to "take over". I also really enjoyed the saved tasks tab, and adding Custom Instructions for specific websites. But it's important to note that Operator is still a research preview and is improving. I found that: -Quite a few sites were blocked after they detected the AI -There's a limited set of partner integrations -It's true purpose is to take actions across the web Operator *operates* within ChatGPT, but it's a completely different tool. Its output lengths are small, and its true purpose is to take actions across the web (typing, clicking, scrolling). Meaning it's not like ChatGPT, which can produce essays and write long code. The future of tech work is here. And personally, I'm incredibly excited about it. Agents can do the boring work, so I can spend more time doing what I love. I'll be publicly sharing all the ways I automate my work with agents, so follow me Rowan Cheung and join my newsletter The Rundown AI for more.

OpenAI just launched GPT-5 in ChatGPT, and made it FREE for all users. I had early access, and it blew my mind: -GPT-5 is going to be available to ALL free users of ChatGPT -Out of 700M weekly users, I'd bet 95%+ of them have only used GPT-4o -GPT-5 is faster than GPT-4o, and smarter than o3 -The upgrade will be a major intelligence boost to most of the world Test 1: The first thing that is immediately noticeable about GPT-5 is the ability to code good front-end/UI GPT-5 generated a fully functioning budgeting app in one shot with ~1000 lines of code, and made it Tetris-themed and even added sound effects Demo video: https://lnkd.in/gEwhcAQs Test 2: To test reasoning, I got GPT-5 to create a complete launch plan for an AI app from a single idea. It did competitor research, product specs, logo, pricing, GTM strategy, roadmap, and more for me. If I were starting with zero business knowledge, this is an insane resource. Demo: https://lnkd.in/gHSPugPk Test 3: To give you a sense of GPT-5's vibes, I exported my Tweet data over the last year and got it to write like my top posts. Then took my newsletter and made it create 3 separate long-form tweets. It's not 100% there, but it beats Claude, which was previously my go-to for editing. Overall, the general vibes of GPT-5 feel much more human-like. It's hard to measure *vibes*, but the combination of speed, lower hallucination rate, and intelligence is very noticeable. As a power user, I've always enjoyed o3, but the speed makes it impossible for daily queries. To round things up, GPT-5 is a huge step for democratizing intelligence to the world 700M weekly users now have state-of-the-art AI in their pockets and can do/learn/build wildly more things than ever before We're living in incredible times--go build something this weekend :^)

Something huge from my convo with Zuck: AI content on Instagram. Soon, Meta will not only recommend content, but also generate content... and you’ll be able to talk to it. Not obvious slop, but hyper-personalized, addictive content tailored to each user's preferences. He thinks this could happen within 2-3 years, not the longer 5 years+ timeline most people expect. The way we consume media is about to change drastically. This got me thinking... right now, we're in a content creator gold rush. It's never been easier to create content, go viral, make money and distribute your message. I think we'll look back at these days of content creation similarly to how we look back at early Facebook ads. In my opinion, there are two paths forward as a media brand/content creator: 1. Go all-in on AI slop - multiple accounts, automation farms, mass-produced content at scale 2. Double down on authentic content - community building, live streaming, personal stories, takes from your lived experience Everything in the middle dies.

OpenAI's DevDay 2025 announcements will change how businesses operate. I sat down with CEO Sam Altman to discuss: → Zero-person billion dollar companies (companies with NO employees) → AI is starting to make small, novel discoveries → Why ChatGPT's 800M weekly users make it the next platform → The Stanford "work slop" study costing companies $186/employee → What happens when AI agents work autonomously for weeks Key moment: Sam said AI is starting to make small, novel scientific discoveries — the beginning of AGI-like breakthroughs. Watch the full conversation: https://lnkd.in/gJ_naAW3

This story is insane. The founders of Fireflies AI (now worth $1B+) pretended to be an AI notetaker before the tech existed... Sam Udotong would: > Join meetings on mute as "Fred from Fireflies" > Take notes by hand while sitting silently > Send the "AI-generated" notes 10 minutes later They did this for 100+ meetings to pay rent, before actually building the product that went on to become a unicorn. Talk about taking the quote "do things that do not scale" in your early days to another level.

I'm HIRING 7 exceptional people and paying a $2,000 referral bonus for each. We’ve bootstrapped The Rundown to $10M+ annual revenue and need to scale faster. We’re tackling one of the biggest problems of the decade: helping 1B workers turn AI into a superpower. Open roles: -Mobile Developer -GM, AI University -Head of Growth -Strategic Partnerships Lead -Product Marketing -Platform Designer -Thumbnail Designer We move faster than any company you've worked at. Competitive comp with performance-based scaling as we grow Every role requires deep taste and AI fluency. You'll be expected to command an army of AI agents working underneath you. Fully remote. Apply at jobs[.]therundown[.]ai or with the link in comments.

Meta just unveiled AI glasses with a built-in display, controlled by a band that reads muscle signals. I sat down with Mark Zuckerberg to unpack: -The timeline for glasses replacing smartphones -How the Neural Band will personalize to you over time -Building "personal superintelligence" while raising kids -Why Zuck rebuilt Meta's AI lab within 15 feet of his desk Watch the full video with the first link in the comments!

NEWS: DeepSeek just dropped ANOTHER open-source AI model, Janus-Pro-7B. It's multimodal (can generate images) and beats OpenAI's DALL-E 3 and Stable Diffusion across GenEval and DPG-Bench benchmarks. This comes on top of all the R1 hype. The 🐋 is not stopping.

Perplexity Comet is not like other agents. It's actually useful. An honest review with 5 use cases (and caveats): First, let's start with the caveats: - It's still early, and struggles with longer workflows - Most data is retained by Perplexity (privacy concern) - Only available to Max users ($200/mo) or free with an invitation code (we just dropped 500 codes in The Rundown University) With that out of the way, here are 5 things I've found it remarkably useful for: 1: Email and Calendar management With the Google Cal and Gmail integrations, it can summarize emails, schedule meetings, and manager calendar events directly from the browser. It's honestly strange that Google hasn't done this yet. 2: Cross-tab product comparison If you're looking for a new product or looking for flights, Comet can compare tabs in real time. It's surprisingly fast and analyzes the reviews of the tabs, too. 3: Summarize any YT video with a click You can summarize + chat with any long YT video and get key moments. This is also possible in Gemini, but having it in the browser means you can watch the video AND chat/learn with Comet in the side tab at the same time. 4: Grab highlights from social feeds If you don't want to ever open up LinkedIn, it can scroll your feed for you and tell you what's happening today. Getting an AI agent to doom scroll for you is a next gen life hack 😂 5: LinkedIn connection request automation If you get a lot of LinkedIn connection requests, you can go through them and figure out cool people you may have missed. I get hundreds a day, and it's impossible to manage, so this one is quite useful to me personally. Overall, I'm impressed with Comet -- it's the first agent (slash AI-native browser) that is actually sticking with me. It's very good for a beta product. I expect it to become even more useful with extended memory and longer, complex workflows. Comet is unfortunately only available to Perplexity Max users ($200/mo) or free via an invitation code. This post was NOT sponsored in any way (hence the honesty), but Perplexity did give me 500 free invites, which are available to Rundown University members today. The AI browsing wars are here!

OpenAI just launched 'Tasks', allowing users to schedule actions and reminders within ChatGPT. Tasks can be one-time reminders or recurring actions (like a daily news rundown), with up to 10 active tasks able to be scheduled at a time. A new '4o with scheduled tasks' model will be available in the dropdown menu, and ChatGPT will also be able to suggest frequent tasks from a user's chat history. The beta feature is rolling out to Plus, Team, and Pro ChatGPT subscribers over the next few days. Is this the first (baby) step in ChatGPT's agentic era?

BREAKING: OpenAI just released two open-weight models: gpt-oss-120b and gpt-oss-20b. The 120B model is on par with o4-mini on reasoning benchmarks and can run on a single 80GB GPU. The 20B model achieves similar results to o3-mini and can run on edge devices with 16GB of memory. More details: -First open-weight models from OpenAI since GPT-2 -Trained using techniques from o3 -Support three reasoning efforts (low, medium, high) like the o-series -Full chain-of-thought reasoning avail to developers Apache 2.0 license For non-technical folks, this means you can now run state-of-the-art OpenAI models locally on your computer, with NO internet. Also, the timing of the launch is super interesting. This comes right after China's open models have been consistently catching up and going viral (Kimi K2, MiniMax, Qwen, DeepSeek). OpenAI really wants to stay king of the hill. After years of keeping their best models closed, this is OpenAI's biggest gift to developers yet.

I think China's second DeepSeek moment is here. This AI agent called 'Manus' is going crazy viral in China right now. Probably only a matter of time until it hits the US. It's like Deep Research + Operator + Claude Computer combined, and it's REALLY good On the GAIA benchmark (an AI benchmark designed to test agents), Manus achieved state-of-the-art performance and beats OpenAI's recently launched Deep Research We noticed Manus gaining some traction at The Rundown AI and wrote about it in the newsletter this morning Shortly after publishing, one of the cofounders reached out with an invitation code! So I dropped my work for the morning (emails can wait) and tested it out. Here's what I found: 1. For my first test, I asked Manus to create a biography on Rowan Cheung and deploy a website based on that biography. It was insanely impressive watching it go through my social channels, browse articles, and deploy the site. And it was 100% accurate, info up to date as of today. 2. Next, I decided to test something a bit more practical: Top rental spots in SF with - Low crime - Lots of AI activity - High density of ambitious young entrepreneurs I got a full detailed report with 4 options that were spot on. 3. Next I wanted to test how Manus would do very long research tasks. So, I asked it to create an entire course on AI for Content Creation It took nearly 2 hours to complete, but what I got was an impressive 8-chapter course with tools, use cases, and even prompt examples. Manus is only available via invite code right now. This post was NOT sponsored in any way, but the cofounder @ Manus gave The Rundown 500 invite codes. We just added it to The Rundown University's perks (which now has $1000+ worth of AI tool savings!)

Huge milestone: The Rundown just crossed 1,000,000 active subscribers! Two years ago, I started this tiny newsletter as a side project—just a way to share my curiosity about AI. I never imagined what it would become. What started as a passion project has turned into something so much bigger: -Thousands of readers message me every day with feedback, ideas, and support. -Interviews with Mark Zuckerberg, Demis Hassabis, and Mustafa Suleyman—conversations I once only dreamed of having. -Our team has grown to 16 incredible people who make this possible every day. -100% of our sponsors come inbound from readers, and 80%+ sponsor again—a testament to the impact of our community. And today, The Rundown officially reaches 1 MILLION daily readers. None of this would have happened without the readers, the supporters, the sponsors, and my amazing team. One milestone down, and many more to go. We’re just getting started. 🚀

Today, we're launching the world’s first personalized AI education platform with tailored training, use cases, and live workshops for your career: We've all heard “use AI or be replaced.” But beyond ChatGPT prompts and expensive AI consultants, there’s no real roadmap to learn AI for your job, skill level, or daily workflow. It's all broad. So we built a personalized AI learning roadmap -- by crowdsourcing the best workflows across the 1M+ AI early adopters in our community and turning them into actionable playbooks. Here's what's inside: 1. Industry-specific AI certificate courses across 16 tech verticals 2. 300+ real-world AI use cases filtered by what's relevant to you 3. Live, expert-led workshops with our in-house AI research team + AI tool partners 4. A personalized community and network of AI-minded professionals in your industry 5. $1,000+ in exclusive discounts on top AI tools That means whether you're in marketing, dev, ops, or any of the 16 industries we cover, you’ll only see what’s actually working in your field. No more noise. We'll show you the vetted AI tools and workflows actually being used in the real world by the early adopters in your industry. We believe the future of work belongs to those who combine AI with domain expertise. And now, there’s no excuse not to start. The first 14 days are 100% free, enough to earn your first certificate 👇

Microsoft just revealed its next big AI bets at Build 2025. I sat down with Microsoft CEO Satya Nadella to unpack: -Microsoft’s vision for the “agentic web” -Why your next job might be AI agent manager -What happens when 95% of code is AI-generated -Where AI agents will create the most value first Watch the full conversation with the first link in the comments!

Google just dropped Gemini 2.5 Pro, the company's most intelligent AI model yet! The "thinking" model debuts at No. 1 on LMArena, and brings new SOTA capabilities across a variety of benchmarks, including: — 18.8% on Humanity's Last Exam — 63.8% on SWE-Bench Verified (agentic coding) — GPQA Diamond and AIME 2025 (STEM) — Long-context and visual reasoning 2.5 Pro also comes with a 1M token context window (2M coming soon!) — and is able to process and understand massive datasets from multiple sources (text, audio, images, and video). Available now via Google AI Studio, the Gemini app for Advanced users, and is coming soon to Vertex AI. Google is cooking, and the acceleration continues 📈

What if AI could cure every major disease in our lifetime? That’s not science fiction, according to Google DeepMind CEO Demis Hassabis. In our recent conversation, Demis shared how he believes AI could cure all diseases by coming up with hundreds of new drugs. He clarified that FDA approvals and regulations could slow things down, but that doesn’t mean the potential cures won’t exist. We also discussed other important topics like: • The future of human-AI relationships • How “vibe coding” is enhancing creatives • Startup moats in an AI-native world • Whether humans risk becoming dumber • What schools and educators must start teaching now It was one of the most insightful conversations I’ve ever had. Watch the full interview 👇 Apple: https://lnkd.in/g6QC4Hqz Spotify: https://lnkd.in/gdepPPYu

Inspired by Wade Foster, we created a role-by-role AI fluency chart at The Rundown AI. Even as an AI-first startup, frameworks like this are very helpful to *set the standard* for new hires. Highly recommend you do the same at your company! Note: Our team uses AI far beyond what’s listed here, but this chart is now our high-level guide for new hires to see what’s unacceptable vs. exceptional.

Over the past few months, we’ve been quietly scaling The Rundown's AI University platform... — $1,000+ in free perks on AI tools like Manus, Bolt, Freepik, Genspark, Synthesia & more — 16 industry-specific AI certificate course tracks — Live workshops with Canva, Zapier, Windsurf, Meta & more And soon, we’re launching the biggest update in our company’s history. I’m biased, but I think it will reshape the AI upskilling/implementation landscape - a problem that urgently needs solving right now for the future of work. Don’t think I’ve been this excited in a long time :^)

I asked Microsoft’s Head of Windows & Devices about the future of PCs in the age of agentic AI. Pavan Davuluri revealed a ton of fascinating insights: 1. On redefining the PC experience: > “We believe technology should adapt to you—not the other way around.” > Microsoft is building toward a future where your PC anticipates, personalizes, and acts on your behalf. 2. On the rise of agentic AI: > “Agentic experiences will become front and center in people’s daily lives.” > AI will soon be capable of running long, autonomous reasoning loops on-device, completing complex tasks in the background while you work. 3. On redefining the user experience: > “Just like we don’t use our full brain at all times, AI will run in the background—anticipating needs and scaling up when needed.” > Microsoft’s vision is shifting from “point and click”PCs to “command and delegate.” Read the full exclusive Q&A on rundown .ai or with the first link in the comments below 👇

Hidden gold mine: AI Documentation If you're a business owner planning on selling your company one day, start keeping detailed documentation of your company's AI processes (!) It could *significantly* boost the valuation of your company in the future. Document everything—tools you’re using, tasks and time automated, money saved or earned, and the impact. These processes might even be worth more than your actual company one day. E.g. your simple AI sales agent that prospects, validates, and does initial outreach to leads could be worth millions *every month* to a bigger company The majority of enterprises I talk to right now don't even know what Claude is. But it's not completely their fault. It's actually incredibly hard to constantly teach AI skills across an entire organization, especially with the pace of AI progress. This leaves a huge opportunity for the smaller players. Startups that can move fast and deploy the right processes and AI agents will be *so far ahead* soon that bigger companies might just start buying these startups for the systems. We're in the golden era for anyone who recognizes the opportunity.

I've been ignoring Apple Shortcut Automations... BIG mistake. 15 Apple Automations that improved the lives of my community members: 1) Record voice memos, email them to yourself, then use an AI agent platform (Gumloop, n8n, or Lindy) to turn them into long-form articles 2) Show recently added contacts in reverse chronological order so you don't forget the new people you've just met 3) A share extension that automatically removes trackers from URLs 4) Scan outside music, get the song through Shazam, then auto-add to a playlist on Apple Music 5) When YouTube is open, automatically set orientation lock 6) Press action key to take a screenshot of any text, translate it, and show it in a window 7) Double tap that switches phone from gray scale to colour for 1 minute to make it less addictive 8) Making the text on your phone gigantic during workouts so you can read phone at gym without needing glasses 9) Using Siri to activate Smart home appliances on command (gates, deck fountains, jets, etc) 10) Copy a URL, push to n8n, scrape the page, send it to an LLM for a summary, and save to notes (Obsidian) 11) Enter car, trigger Google Maps to get a time estimate, and auto-texts wife ETA 12) Location based trigger that detects when near the gym, opens workout app, and chooses the next workout based on previous day's workout 13) Setting 100% volume every morning so you don't miss alarms 14) Taking a URL from YouTube, combining it with a propmt to extract frameworks or ideas from the video, putting the prompt in the clipboard, and opening Gemini 15) Automatically setting alarms an hour before meetings There were 200+ answers, so it was TOUGH to pick the top 15. Thanks to everyone for contributing! If this gets enough traction on LinkedIn, I'll do some more here as well.

NEW: Google just released Gemini 2.0 Pro experimental — the company's most advanced AI model yet. 2.0 Pro Experimental features include: — a massive 2M token context window — the ability to use tools like Google Search and code execution — Upgrades in reasoning, coding, and complex tasks over 1.5 Pro On the reasoning front, 2.0 Flash Thinking Experimental also goes live in the Gemini app — which currently sits atop the LM Arena's model leaderboard. Google is also launching a separate version of Flash Thinking that can interact directly with apps like YouTube, Search, and Maps. Additionally, some smaller, faster 2.0 options are now available: Flash and Flash-Lite. The major AI labs continue to ship, and the time between releases is only getting shorter.