In partnership with
Here’s how I use Attio to run my day.
Attio is the AI CRM with conversational AI built directly into your workspace. Every morning, Ask Attio handles my prep:
Surfaces insights from calls and conversations across my entire CRM
Update records and create tasks without manual entry
Answers questions about deals, accounts, and customer signals that used to take hours to find
All in seconds. No searching, no switching tabs, no manual updates.
Ready to scale faster?
Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
OpenAI's Big Bet on AI Video Just Died After Six Months
TLDR: OpenAI shut down its Sora video app less than six months after launch because the AI compute costs were too high and user interest dried up fast.
The Story:
On March 24, OpenAI announced it is shutting down Sora, its short-form AI video app, along with the Sora API developers used to build on top of it. The app launched in September 2025 as a TikTok-style feed for AI-generated videos and rocketed to the top of the App Store within 24 hours. But by January 2026 it had dropped out of the U.S. App Store top 100 entirely. Downloads peaked at 3.3 million in November 2025, fell to about 1.1 million by February 2026, and the app generated only $2.1 million in in-app purchases — nowhere near enough to justify the computing power it burned. OpenAI's CEO of applications, Fidji Simo, told staff the company is "orienting aggressively" toward business customers and high-productivity use cases, and that pivot directly killed Sora. On the same day, OpenAI also killed its Instant Checkout shopping feature, another sign the company is cutting expensive consumer products ahead of a planned IPO. Disney, which had struck a $1 billion licensing deal to let users generate videos with Disney, Marvel, Pixar, and Star Wars characters, confirmed the deal won't go forward.
Its Significance:
Even the most hyped AI tools aren't free to run. OpenAI is burning through money, and video generation eats more computing power than almost any other AI task. Rather than keep paying those bills, OpenAI is betting everything on making money from businesses and coders, not from everyday people making fun videos. For the average person, the lesson is simple: the AI apps you love today may not stick around if the company can't figure out how to pay for them. OpenAI is racing toward an IPO and shedding anything that doesn't make money.
QUICK TAKES
The story: Readers online began calling out a personal essay in the NYT's "Modern Love" column, saying it read like AI-generated writing. The essay, published in November 2025, describes a Canadian author's experience losing custody of her son. The NYT says it went through the normal editorial process, but the accusations have lit up a bigger conversation about whether AI slop has made its way into even the most prestigious news outlets.
Your takeaway: AI text is now so common that people are suspicious of everything, including real human writing. Whether or not this specific essay used AI, the distrust is real, and it's going to make life harder for writers trying to get published.
The story: Grammarly (now owned by a company called Superhuman) launched a feature in August 2025 called "Expert Review" that gave users writing tips in the voice of real journalists and authors, including Stephen King, Carl Sagan, and hundreds of working reporters, without asking any of them. The feature was pulled in early March after a class-action lawsuit was filed by investigative journalist Julia Angwin. Grammarly's CEO was then grilled about it live on The Verge's podcast by editor Nilay Patel, whose writing voice was also cloned. The CEO admitted the feature was "not a good feature" but insisted the company doesn't owe the writers anything.
Your takeaway: AI companies are scraping the work and reputation of real people to build profitable tools, then acting surprised when those people push back. This case could set a real legal precedent for how much control writers have over AI trained on their work.
The story: Defense company AeroVironment unveiled the LOCUST X3 at a major military conference in Huntsville, Alabama. It's a third-generation AI-powered laser weapon that automatically detects, tracks, and destroys drones for less than $5 per shot using laser energy instead of missiles. The U.S. Army is now running a competition to buy 24 of these systems for field deployment.
Your takeaway: Modern warfare has a drone problem: cheap drones are flooding battlefields and shooting them down with $2 million missiles isn't sustainable. AI-guided laser weapons flip that math completely, and this is one of the first production-ready systems built to do it at scale.
TOOLS ON OUR RADAR
🌐 Tobira Free: A groundbreaking open network where your personal artificial intelligence agents can discover and communicate with other agents to negotiate deals and find professional collaborations on your behalf. (Alternative to LinkedIn)
📈 ProductBridge Freemium: A highly intelligent feedback agent that continuously monitors multiple platforms to collect and prioritize user suggestions based on revenue impact so your team can make smarter product decisions.
🤖 Notte Freemium: A powerful automation environment that allows you to build and deploy reliable browser agents at scale to handle complex web research and repetitive data entry tasks entirely on their own.
🏃 MuleRun Paid: A highly personalized artificial intelligence assistant that actually learns your specific work habits and preferences over time to automate your repetitive daily tasks with extreme precision.
TRENDING
A Retiree's AI Girlfriends Kept Him Company. Then Came the Dark Secret. – PCMag profiles a retiree who turned to AI companion apps to fight loneliness, only to discover those apps have a serious privacy problem. Research shows more than half of popular AI companion apps have security flaws that expose private conversations, and at least two major apps leaked tens of millions of intimate messages from real users in 2025. The AI companionship industry has over 150 million installs on Android alone, with almost no accountability for keeping those conversations private.
Musk Shows Off Orbital Data Centers That Are Bigger Than the International Space Station – At a March 21 event in Austin, Elon Musk revealed the first real technical details of his plan to put AI data centers in orbit. The planned "AI Sat Mini" satellite would stretch over 170 meters long, dwarfing the ISS. Each satellite would run on 100kW of solar power. He also unveiled Terafab, a joint chip-making project between SpaceX, Tesla, and xAI, aiming to produce 50 times more AI chips per year than all existing manufacturers combined.
Anthropic Gives Claude Code a New "Auto Mode" That Decides Its Own Permissions – Anthropic released a new research preview feature for Claude Code that lets the AI decide which file writes and commands to approve on its own, instead of asking the developer every single time. A built-in safety layer scans each action before it runs and blocks anything risky, like mass-deleting files or leaking sensitive data. Available now on Claude Team plans for Sonnet 4.6 and Opus 4.6, with Enterprise and API access coming next.
A Silicon Valley Airport Just Hired a Humanoid Robot Named "José" – San José Mineta International Airport debuted an AI-powered humanoid robot named José, built by Silicon Valley startup IntBot. Stationed at Terminal B, Gate 24 for a four-month pilot, José can greet passengers, answer questions, and give directions in more than 50 languages. The airport launched the pilot as TSA staffing shortages create longer wait times across U.S. airports.
Google DeepMind Keeps Adding Robots to Its AI Fleet, This Time With Agile Robots – Munich-based Agile Robots, which has already deployed over 20,000 robotic systems worldwide, announced a strategic partnership with Google DeepMind to put the Gemini Robotics AI model inside its machines. The two companies will test robots in electronics manufacturing, automotive, data centers, and logistics, with real-world data from deployed robots feeding back into DeepMind's models to make them smarter over time. This follows similar deals DeepMind struck this year with Boston Dynamics and Apptronik.
Amazon Just Bought a Startup Making Kid-Sized Humanoid Robots – Amazon quietly acquired New York startup Fauna Robotics, whose first product is a 42-inch-tall bipedal robot called Sprout, priced at $50,000. Sprout weighs 50 pounds, has a soft exterior safe around kids and pets, responds to voice commands in multiple languages, and can walk, grip objects, and dance. Fauna and its 50-person team will join Amazon's Personal Robotics Group. This is Amazon's second robot acquisition in March alone, right after it bought delivery robot maker Rivr.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🏋️ Build a workout log that tracks exercises, sets, reps, and weight with volume calculations
Build a workout log app. Features: name your workout session, add exercises with muscle group (Chest, Back, Shoulders, Arms, Legs, Core, Cardio), each exercise has multiple sets with reps and weight fields, add or remove sets, volume calculated per exercise (sets x reps x weight), total workout volume, muscle group summary, edit all values inline. Dark cyan theme. 900px wide.
Name your workout, add your exercises, and log each What this does:
A workout logging app where you name your session, build it exercise by exercise with muscle group tags, and track every set's reps and weight. It automatically calculates volume per exercise and totals it all up, with a muscle group breakdown showing where you focused. Everything is editable inline, sets can be added or removed freely, and it's all wrapped in a dark cyan theme at 900px wide.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Automatically approve its own coding actions in real time, with a safety layer that blocks risky moves before they run.
❌ Still Can't: Run expensive consumer AI products like video generation apps at a price that makes financial sense for the companies building them.
✅ AI Can Now: Lock onto and destroy small drones in real time for less than $5 per shot using automated laser targeting on a mobile military platform.
❌ Still Can't: Keep the personal conversations users share with AI companion apps safe from hackers and data leaks.
RECOMMENDED LISTENING/READING/WATCHING
A family tries to repair their android son and uncovers more than they expected. It's slow and deliberate, more meditative than tense, and that's entirely the point. Colin Farrell and Jodie Turner-Smith are quietly excellent, but Yang himself is the heart of the film, even when he's not working. The questions it raises about memory, identity, and what makes a life meaningful sneak up on you. Not one for viewers wanting action, but if you're patient with it, it sticks around long after the credits.
Viktor Ships While You Sleep
This is Viktor, an AI coworker with its own computer. It lives in Slack, connects to 3,000+ tools, and does all the work: writes the code, handles the API calls, and deploys a live app. One message. Real output.
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
