Get the latest AI news, understand why it matters, and learn how to apply it in your work — all in just 5 minutes a day. Join over 2,000,000+ subscribers.
An exclusive Q&A with alibaba.com's Kuo Zhang
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. The launch of OpenClaw a few months ago marked a major shift in what agentic AI meant — and more importantly, what it could unlock for businesses worldwide.
Now, alibaba.com is offering an early glimpse of that future with Accio Work, a new agentic system that deploys a team of digital employees to handle complex business tasks 24/7.
To better understand what Accio Work delivers today, what comes next, and what it means for how businesses will be run, we sat down with Kuo Zhang, President of alibaba.com, for an exclusive Q&A.
In today’s AI rundown:
Going from AI assistants to operators
A dynamic workforce for your business
Agent-to-agent is the future of B2B
The new edge for professionals
Quick hits with Kuo
LATEST DEVELOPMENTS
ORGANIZATIONAL SHIFT
⚡️ From AI assistants to operators
The Rundown: To truly become an agent-centered organization, Zhang said teams will first have to focus on three key things — prioritizing the business value of agentic AI, building checkpoints for transparency, and rethinking how teams operate.
Cheung: As AI moves from assistant to operator, what has to change inside a company before handing core operations to an AI agent?
Zhang: The mindset shift needed to move from AI as a helper to an operator starts with focusing on value, not just activity. Early on, we were impressed by how fast AI could write or process data, but for an AI operator, those are just basic capabilities. Success is about the high-value decisions it can navigate and the actual profit it can bring for the business.
You also have to build checkpoints. Handing core operations to AI doesn’t mean letting it run wild. You need clear guardrails, breaking complex tasks into smaller steps where the AI proves its work at every stage, ensuring a small digital mistake doesn’t turn into a real-world disaster.
Finally, the team has to move from doing the “grunt work” to designing the logic. People become the architects and referees who set the standards and provide the final human sign-off — not replaced by AI, but upgraded to manage it.
Why it matters: The move to agent-driven work starts from the ground up. As new models and capabilities keep emerging, the advantage for businesses will come from identifying what actually moves the needle for them — and building workflows, guardrails, and policies around it.
ACCIO WORK
🤖 A dynamic workforce for your business
The Rundown: Accio Work sets up Qwen-powered agent teams for businesses (based on their goals), with different connectors, skills, and computer use capabilities enabling the “digital workforce” to handle multistep tasks across functions on its own.
Cheung: How does Accio Work build an agent team?
Zhang: Accio Work assembles a fleet of specialized agents based on your described goals — no code. But the system is fully customizable: you can manually configure agents, form your own team combinations, and most importantly, encapsulate your own expertise into reusable skills that shape how your agents behave.
Users manage the agents through an app. What changes is what management means. Instead of coordinating people and tasks, the founder's job becomes configuring and monitoring this digital workforce — setting the parameters, reviewing outputs at key checkpoints, and progressively refining how the agents operate.
Cheung: Can you walk us through a real workflow that it can handle today?
Zhang: Take launching an e-commerce business. The agent analyzes real-time market trends, selects products, and performs one-click store setup on Shopify, Amazon, or TikTok Shop. It then conducts multi-round supplier negotiations autonomously, securing the best price before surfacing the final deal for your approval.
Once approved, it handles logistics tracking, contract drafting, and VAT filing prep. The human role is final sign-off on payments and regulatory submissions, not managing the process itself.
Why it matters: This shows how work is starting to shift from managing each step to simply defining outcomes — set the goal, and a team of agents handles execution by chaining together tools and skills. Humans step in where it matters most, while the agents handle everything in between.
AGENTIC FUTURE
🦾 Agent-to-agent is the future of B2B
The Rundown: Zhang sees an agent-to-agent future for businesses, where autonomous agents communicate with each other and take actions across most workflows, leaving only business-critical decisions to their human managers.
Cheung: With Accio Work and similar agents, what will running a company look like five years from now?
Zhang: The future of B2B is A2A (agent to agent). Think most business tasks flowing between autonomous agents, with minimal human initiation. What makes this work in practice is that these agents will operate within sandboxed environments, staying within human-defined parameters for anything sensitive or high-stakes. So, autonomy and accountability will coexist, not trade off against each other.
Cheung: If an agent makes a costly mistake, who owns that outcome?
Zhang: Responsibility stays with the human, by design. Our system is built in such a way that any action with access to private files or real financial / legal consequences will require explicit human approval before it is executed.
In tasks like VAT filings, for example, Accio provides semi-automated assistance — it does the preparation (identifying local regulations, organizing the data, and reducing the manual burden), but the final submission goes through human channels. The agent handles execution, but accountability never leaves the person running the business.
Why it matters: Even as AI agents take over key business functions, responsibility for their actions will always stay with humans. This will shift the job from doing the work to overseeing it and owning the outcomes, making judgment and decision-making critical to running agent-driven businesses of the future.
HIRING
⚙️ The new edge for professionals
The Rundown: Zhang noted that agentic platforms like Accio Work will level the playing field between small startups and large enterprises, with the biggest skill for professionals becoming the ability to set success standards for AI and assess them.
Cheung: How will these agent platforms change hiring? Which skill do you think becomes more valuable?
Zhang: The concept of headcount as a proxy for capability breaks down entirely with agentic AI.
A solo founder with the right agent platform can handle sourcing, multi-market compliance, and negotiations that previously required a team — effectively leveling the playing field against much larger competitors.
Zhang added: The highest-value professionals amid this shift will be those with deep enough domain expertise to set meaningful success criteria and catch AI errors. That expertise itself becomes a new asset: packageable into reusable skills, monetised in a marketplace, portable across employers in a way it never was before.
Why it matters: As agent teams handle execution, the professional edge shifts from doing the work to knowing what good looks like. Zhang’s key insight: domain expertise becomes packageable and portable — turning into a monetizable asset in a way traditional job skills never were.
LIGHTNING ROUND
⚡️ Quick hits with Kuo
The biggest misconception about AI agents today is?
Zhang: That AI agents are meant to replace people — they’re not. They are meant to become a "dedicated team of experts." Many people believe that AI should be fully autonomous and complete complex tasks in a single step, but in reality, the true value lies in humans and AI co-constructing verifiable, iterative workflows.
One capability AI agents still fundamentally lack?
Zhang: While AI reasoning is scaling fast, they still lack 'physical grounding'— bridging the gap between digital intelligence and real-world execution.
What’s one thing you believe about AI agents that most tech leaders would disagree with?
Zhang: Many tech leaders are hyper-focused on general AI and consumer tools, but we firmly believe that consumer (C-end) apps are for killing time; Business (B-end) tools are for making money.
How far are we from the first one-person billion-dollar company?
Zhang: Probably just months.
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown
SpaceX slips the IPO script
Read Online | Sign Up | Advertise
Good morning, tech enthusiasts. Elon Musk may be gearing up for the strangest megadeal Wall Street has ever seen: a SpaceX IPO so massive it could put the usual rules of public markets under serious strain.
Behind the scenes, the plan reportedly pairs a sky-high valuation with an unusually large retail allocation, giving everyday traders — and more than a few Musk loyalists — a rare seat on the rocket.
In today’s tech rundown:
Musk wants to take SpaceX public — his way
Meta’s new Ray-Ban AI glasses leak via FCC
Defense startup Shield AI hits $12.7B
Meta pours $10B into Texas megadata center
Quick hits on other tech news
LATEST DEVELOPMENTS
SPACEX
🚀 Musk wants to take SpaceX public — his way
Image source: Ideogram / The Rundown
The Rundown: Elon Musk is engineering what could be the largest — and strangest — IPO in history, plotting a mid‑June SpaceX listing that may raise tens of billions of dollars, while carving out an unprecedented 30% of shares for retail investors.
The details:
SpaceX is targeting a mid‑June IPO window, with internal timelines synced to Musk’s birthday and fundraising estimates ranging from about $40B–$75B.
The company is weighing a valuation of around $1.5T trillion after folding xAI into SpaceX, a move designed to sell investors on orbital AI data centers.
Instead of the typical banker‑led roadshow, Musk wants investors flown into SpaceX’s campus to tour production lines and possibly watch launches.
Meanwhile, banks are reportedly confined to narrowly defined “lanes” and a retail tranche that could hit 30% of the float.
Why it matters: The IPO could make SpaceX one of the most valuable public companies on the planet while testing how far markets are willing to go on a Musk growth story. It also flips the IPO script by handing an unusually large slice to retail investors, giving Musk fans front‑row seats to one of the biggest deals of the decade.
META
👓 Meta’s new Ray-Ban AI glasses leak via FCC
Image source: Meta
The Rundown: Meta and EssilorLuxottica are gearing up to launch two new Ray-Ban smart glasses, after fresh FCC filings revealed production-ready hardware with upgraded Wi-Fi 6E connectivity and a new charging case design.
The details:
New FCC filings confirm the glasses, codenamed Scriber and Blazer, are production hardware, with reports suggesting a launch could come in weeks.
The Blazer will come in standard and large sizes. Both models keep a portable charging case, with what is predicted to be a significant hardware refresh.
The filings add support for Wi‑Fi 6/UNII‑4 at 5.9 GHz, a bandwidth upgrade that should boost livestreaming and on‑device Meta AI capabilities.
Meta is positioning Ray-Ban glasses as its primary AI hardware play, with the line already selling in the millions.
Why it matters: Meta is doubling down on Ray-Ban, pushing new Scriber and Blazer AI glasses toward launch as it scales production of smart specs already selling in the millions. Timing is complicated, with the devices arriving amid privacy and legal backlash over claims that Meta glasses funnel user footage to offshore contractors.
SHIELD AI
✈️ Defense startup Shield AI hits $12.7B
Image source: Shield AI
The Rundown: Shield AI, the San Diego startup building AI pilots for military aircraft, raised $1.5B and more than doubled its valuation to $12.7B — and it already has a contract to prove the technology works in the field.
The details:
The U.S. Air Force tapped Shield’s Hivemind software to power a program in which autonomous drone “wingmen” fly alongside human combat pilots.
Hivemind will also run on the Fury autonomous jet built by rival Anduril, which makes its own software stack and is eyeing an $8B raise at a $60B valuation.
Shield AI is already putting its new funding to work, acquiring Aechelon Technology, whose hyper-detailed simulators train U.S. pilots.
Why it matters: Shield AI’s new $1.5B funding round shows how quickly defense dollars are flowing into AI software capable of steering warplanes, and it sets up a direct rivalry with Anduril as both companies race to own the “brain” of autonomous combat aircraft.
META
🤑 Meta pours $10B into Texas megadata center
Image source: Meta
The Rundown: Meta bumped its planned investment in a new El Paso, Texas, data center from $1.5B to more than $10B — a sevenfold increase — as it races to build the compute backbone for its next generation of AI models.
The details:
Meta’s investment will grow the facility to 3.1M square feet, up from the originally planned 1.2M, with the site designed to scale to 1 gigawatt of capacity.
The gigawatt-scale facility is expected to come online in 2028, making it one of Meta’s largest data centers globally.
At peak construction, approximately 4K workers will be on site, with more than 300 permanent roles once the center is fully operational.
Meta has also committed to adding over 5K megawatts of clean power to the grid and will work with nonprofits to offset the facility’s water burden.
Why it matters: Meta expects its total capital expenditures in 2026 to land between $115B and $135B, with AI infrastructure at the core — a massive jump from the $72.2B spent last year. Meta is also spending on an astronomical scale to stay in the top tier of foundation-model players, instead of leaning on outside cloud providers.
QUICK HITS
📰 Everything else in tech today
NASA is ditching its planned lunar-orbit station and instead redirecting its hardware into a $20B moon base to be built on the surface over the next seven years.
A jury ordered Meta to pay $375M for misleading users about the safety of Facebook, Instagram, and WhatsApp and enabling child sexual exploitation on its platforms.
Apple discontinued the Mac Pro, with no plans for any future Mac Pro hardware, and Mac Studio instead being repositioned as the flagship pro desktop.
Trump plans to appoint Mark Zuckerberg, Larry Ellison, and Jensen Huang to a White House tech panel that will advise on AI policy and other technology issues.
A Los Angeles jury ordered Meta and YouTube to pay a woman $3M for harm caused by their addictive apps she used as a child.
WhatsApp’s upgraded Writing Help uses AI to draft suggested replies from your chats, plus it adds photo touchups, chat transfer, and multi-account support on iOS.
Epstein survivors are suing the Trump administration and Google for exposing their IDs in released case files and for keeping their private info visible in search results.
Netflix raised prices again, increasing its ad-supported plan to $8.99 a month, its standard ad-free plan to $19.99, and its premium tier to $26.99.
Meta is laying off several hundred employees across Reality Labs, Facebook, and other units as it restructures and shifts more investment toward AI.
Japanese lunar startup ispace pushed back its NASA-sponsored moon landing mission to 2030 and plans to shrink its global workforce.
Apple now forces UK users to prove they’re 18 via payment details or ID to keep full iPhone and iCloud access, else stricter safety filters kick in.
An 82-year-old Kentucky farmer turned down a $26M offer from an unnamed AI company to carve out part of her 1,200-acre farm for a data center.
Honda pulled the plug on the two Afeela-branded EVs it was co-developing with Sony, as the Sony Honda Mobility joint venture winds down its EV plans.
COMMUNITY
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: Meta’s new open-source brain AI
Read our last Tech newsletter: The deep-sea luxury race is back
Read our last Robotics newsletter: Amazon now has a kid-sized humanoid
Today’s AI tool guide: Use Perplexity Computer as a personal shopper
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — The Rundown’s editorial team
Meta's new open-source brain AI
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. AI companies keep trying to build systems that think like a brain. Meta went the other direction… Building one that reads them.
TRIBE v2 can predict how your neurons fire in response to anything you watch, hear, or read — a breakthrough for researchers studying the brain, and a pretty significant data point for the world's largest social media and advertising company.
In today’s AI rundown:
Meta's brain model beats real fMRI scans
Apple to unlock Siri for rival AI assistants
Use Perplexity Computer as a personal shopper
Wikipedia bans AI from writing its articles
4 new AI tools, community workflows, and more
LATEST DEVELOPMENTS
META
🧠 Meta's brain model beats real fMRI scans
Image source: Meta
The Rundown: Meta just open-sourced TRIBE v2, an AI model trained on brain scans from 700+ people that simulates neural activity across vision, hearing, and language — with its synthetic predictions actually outperforming real fMRI recordings.
The details:
Trained on 1,000+ hours of brain data, v2 leaps from 1,000 brain regions to 70,000, with 700+ subjects up from just 4 volunteers in the original.
TRIBE v2's predictions matched population-level brain activity better than most real scans, which often get clouded by heartbeats, movement, and noise.
The team replicated decades of neuroscience findings in software, correctly pinpointing brain regions for faces, speech, and text with zero scans.
Meta open-sourced the code, weights, and a live demo, letting any researcher start running virtual brain experiments without building from scratch.
Why it matters: Neuroscience has long required putting people inside expensive scanners for every new experiment, a bottleneck that's kept entire fields moving one study at a time. TRIBE v2 could do for brain research what AlphaFold did for protein structure: compress months of scanning into seconds of compute.
TOGETHER WITH SERPAPI
🌐 The web data API built for AI teams
The Rundown: Need real-time public web data without scraper headaches? SerpApi turns search results into structured, ready-to-use data with built-in scale and block resilience — powering AI apps, product research, price tracking, SEO insights, and more without the maintenance.
With SerpApi, you get:
Real-time, reliable results from any search engine
ZeroTrace Mode to keep your searches confidential
U.S. Legal Shield for scraper legal protection
Try SerpApi for free today — Rundown readers can get 50% off for 3 months via chat or email.
APPLE
📱 Apple to unlock Siri for rival AI assistants
Image source: Apple
The Rundown: Apple plans to open up the upcoming Siri revamp for other models starting with iOS 27, according to Bloomberg — ending ChatGPT's exclusive integration and letting users choose which AI handles their queries directly from the assistant.
The details:
Users will be able to pick their preferred AI in ‘extensions’ settings and route questions to models of their choice via Siri with the incoming iOS 27.
ChatGPT is currently the only model compatible with Siri commands via its 2024 deal, but use of that integration has reportedly been ‘minimal’.
Bloomberg said chatbots in the App Store could also be a revenue stream, with Apple taking a cut of AI subscriptions purchased across its devices.
Apple is expected to introduce the new Siri AI overhaul powered by Gemini at its WWDC developer event in early June.
Why it matters: Google is already rebuilding Siri's underlying tech with Gemini, and ChatGPT has had a spot since 2024. Now, Apple is letting the rest of the field in to provide more user choice. It’s a smart move — skip the model war entirely, layer the best AI on top of a billion iPhones, and let its hardware moat do the rest.
AI TRAINING
🛍️ Use Perplexity Computer as a personal shopper
The Rundown: In this guide, you will learn how to use Perplexity's Computer agent search feature (for Max subscribers or Pro users with credits) to shop for you, find items in stock in your size, and save a ton of time.
Step-by-step:
Go to Perplexity’s website or app (latest version), and look for the Perplexity Computer toggle
Prompt: “Find me a good deal on [item]. I prefer brands like [brands]. Start with the item type, then add brand preferences, cosmetic style, and sizing”
Computer will delegate sub-agents to browse product listings and compile a report for you. It is really good at formatting the report without you telling it to
If you want to watch prices or do daily market research, you can prompt: “Turn this into a daily/weekly automation,” and Perplexity will schedule it for you
Pro tip: To save a little bit on tokens, you can switch the orchestrator model to Claude Sonnet.
PRESENTED BY INNOVATING WITH AI
💼 How Dan built a 6-figure AI consultancy from scratch
The Rundown: Dan had no tech background and no business experience — just a love for AI and a hunch it could become something more. Through The AI Consultancy Project, he landed his first clients, found a niche, and built a six-figure consultancy that let him quit his 9-to-5 in under a year.
Dan’s case study breaks down:
How he went from zero clients to a six-figure AI consultancy in 12 months
The early stumbles and the system that finally worked
The exact niche and mindset shift that unlocked everything
WIKIPEDIA
❌ Wikipedia bans AI from writing its articles
Image source: Wikipedia
The Rundown: Wikipedia's volunteer editors banned the use of AI to write articles on the foundation’s English-language site, a move the policy's author called a "pushback against enshittification and forceful push of AI by so many companies”.
The details:
Prior attempts at broad AI rules failed to reach consensus, but mounting AI-generated errors pushed editors to a near-unanimous 40-2 vote.
The ban covers writing or rewriting articles with LLMs, with editors still allowed to use AI for grammar fixes and translations with human review.
The policy's author said the change could "spark a broader change" and "empower communities on other platforms" to set AI rules on their own terms.
StackOverflow and German Wikipedia have enacted similar bans, with Spanish Wikipedia going further to fully ban the use of AI, even for editing purposes.
Why it matters: AI text reportedly surpassed human output for the first time in 2025, and Wikipedia is trying to hold the human line, all while Elon pushes Grokipedia (an AI-created version of Wikipedia) in the exact opposite direction. The internet's most-used knowledge base bet against the current, but how long that holds is anyone’s guess.
QUICK HITS
🛠️ Trending AI Tools
🤖 Scrunch - See how AI interprets your site, run a free audit, and unlock the new way to reach customers*
🎶 Suno - New v.5.5 AI music generation model with upgraded personalization
🗣️ Gemini 3.1 Flash Live - Google's low-latency voice AI for real-time agents
💬 Voxtral TTS - Mistral's voice cloning AI for multilingual speech agents
*Sponsored Listing
📰 Everything else in AI today
Google rolled out Gemini 3.1 Flash Live, a new voice AI with upgrades in speed, task completion, and realism, to power convos across Search, Gemini Live, and its API.
Mistral released Voxtral TTS, a lightweight voice AI that clones any speaker from a 3-second clip and generates natural-sounding speech across 9 languages.
OpenAI has reportedly shelved its planned erotic chatbot mode indefinitely after pushback from staff and investors.
Novo Nordisk is deploying AI agents across clinical trial ops, with the pharma giant saying the tech is trimming approval timelines and reducing the need for contractors.
Suno launched v5.5 of its AI music generator, adding voice cloning, custom model tuning, and personalized style learning for Pro subscribers.
Cohere released Transcribe, a free open-source speech recognition model that tops HuggingFace's accuracy leaderboard across 14 languages — taking the No. 1 spot.
COMMUNITY
🤝 Community AI workflows
Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.
Today’s workflow comes from reader Hugh F. in Sandwich, MA:
"I play keyboard in an open mic group, but need to simplify song piano scores to two pages to avoid having to turn pages mid-song, and also to make the piece suitable for an accompaniment.
Feeding a PDF version of a full score to Claude and requesting such a simplification has resulted in quite acceptable pieces. So, less time adapting piano scores, which has resulted in more time practicing, which has itself resulted in more applause. Thanks, Claude!"
How do you use AI? Tell us here.
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: ARC-AGI-3 resets frontier AI scoreboard
Read our last Tech newsletter: The deep-sea luxury race is back
Read our last Robotics newsletter: Amazon now has a kid-sized humanoid
Today’s AI tool guide: Use Perplexity Computer as a personal shopper
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown
Amazon now has a kid-sized humanoid
Read Online | Sign Up | Advertise
Good morning, robotics enthusiasts. Amazon just acquired New York startup Fauna Robotics — and its pint-sized humanoid Sprout, a 3.5-foot bot that walks, crawls, jumps, and makes eyes at you with animated LED pupils.
It’s Amazon’s second robotics buy this month, after stair-climbing delivery bot Rivr. Where it goes from warehouse to living room, that part’s still wide open.
In today’s robotics rundown:
Amazon buys NYC-based humanoid startup
Zoox robotaxis roll into Austin and Miami
A wristband that can control robots
New 911 drone aims to replace police choppers
Quick hits on other robotic news
LATEST DEVELOPMENTS
AMAZON
🤖 Amazon buys NYC-based humanoid startup
Image source: Fauna Robotics / Reve
The Rundown: Amazon just snapped up New York-based Fauna Robotics this week, adding a pint-sized, developer-friendly humanoid to a robotics portfolio that’s now expanding well beyond the warehouse.
The details:
The deal, first reported by Bloomberg, brings Fauna’s roughly 50 employees, including its two founders, into Amazon in New York City.
Sprout stands 3.5 feet tall, weighs 50 pounds, and walks, crawls, jumps, dances, grips objects, and emotes with LED eyes and motorized eyebrows.
The $50K robot runs on a 64GB Nvidia Jetson AGX Orin and ships with a developer platform that lets researchers build apps via modular AI architecture.
It’s Amazon’s second robotics buy this month, following its acquisition of Zurich-based Rivr, the startup behind a stair-climbing delivery bot.
Why it matters: Amazon has tried to crack the home robot market twice now — Astro went nowhere, and iRobot got blocked by regulators. Sprout is a different approach: rather than starting with a mass-market product, Amazon is buying a full-stack humanoid dev platform to seed an ecosystem first and decide on the use case later.
TOGETHER WITH ROBOFLOW
🔥 Roboflow Vision for Robotics
The Rundown: Roboflow’s vision AI platform helps robotic teams build systems that handle unstructured tasks, navigate diverse terrain, and collaborate with humans, from data pipeline to production deployment faster.
With Roboflow, you can:
Navigate dynamic environments with full situational awareness
Handle irregular objects and mixed inventory without rebuilding pipelines
Deploy edge or cloud inference to your entire robot fleet
ZOOX
🚖 Zoox’s robotaxis roll into Austin and Miami
Image source: Zoox
The Rundown: Amazon-owned Zoox is deploying its steering‑wheel‑less robotaxis for testing on public roads in Austin and Miami this year, a key step toward launching commercial ride‑hailing beyond Las Vegas and San Francisco.
The details:
Initial rides in both cities will be limited to Zoox employees and their friends and families before opening to early public riders through an “Explorer” program.
Zoox is also massively expanding its service areas in San Francisco and Las Vegas, including dense neighborhoods and major entertainment venues.
The company says its robotaxis have already logged nearly 2M autonomous miles and carried more than 350K riders.
Zoox has inked a multiyear deal to put its robotaxis on the Uber app in Las Vegas starting this summer, with Los Angeles set to follow in 2027.
Why it matters: Zoox is pushing into Waymo’s territory, even though it still needs NHTSA safety exemptions before it can charge fares. Waymo is operating robotaxis in 10 cities nationwide, while Tesla’s long-promised robotaxi service remains nascent, narrowing the competitive field as Zoox pushes toward paid rides in 2026.
MIT
👉🏽 A wristband that can control robots
Image source: MIT / Melanie Gonick
The Rundown: MIT engineers just built a chunky wristband that transforms the tendons and muscles in your wrist into a real-time control interface for robots and virtual worlds — no cameras, no wired gloves, no physical controllers required.
The details:
The band fires grayscale ultrasound at tendons in your wrist, then feeds that data to an AI model that translates motion patterns into control signals.
Trained on synchronized ultrasound and motion-capture data, the system lets users puppeteer a robotic hand to manipulate virtual objects and play piano.
In tests with eight users performing 26 ASL letters and everyday grasps (like scissors and tennis balls), it reconstructed hand poses in real time.
Lead author Xuanhe Zhao envisions a miniaturized, general-purpose band that streams rich hand data to train robots and gives AR/VR more intuitive control.
Why it matters: This wristband points to a future where human dexterity becomes a universal interface: your hand movements streamed directly to robots, AR headsets, and whatever comes after the smartphone (Meta’s also doing something similar with its Ray-Bans). Simply put, a new, more intuitive kind of control is coming our way.
BRINC
🚁 New 911 drone aims to replace police choppers
Image source: BRINC
The Rundown: Backed by Sam Altman and the U.S.’s anti-DJI mood, former Thiel fellow Blake Resnick’s startup BRINC has unveiled Guardian, a Starlink-linked drone it claims can replace police helicopters with 60 mph flights and robotic battery swaps.
The details:
Guardian is designed to be deployed from rooftop “nests,” autonomously swapping batteries so it can stay in operation and launch quickly on 911 calls.
The drone links to Starlink for long-range, low-latency connectivity and streams video and data to command centers even in dead zones.
Its 60 mph top speed, heavy payload capacity, and sensor suite are pitched as delivering helicopter-like capabilities in search and rescue and suspect pursuit.
The launch comes as U.S. lawmakers seek to sideline Chinese-made DJI drones, creating an opening for domestic suppliers like BRINC.
Why it matters: Guardian arrives at a pivotal moment. U.S. lawmakers have spent the past year pushing to ground DJI drones over national security concerns, leaving a gaping hole in the domestic public safety market. BRINC is positioning itself to fill it — with a Starlink-tethered aircraft that promises to be cheaper and faster than helicopters.
QUICK HITS
📰 Everything else in robotics today
OpenAI is shutting down its Sora AI video app and API to redirect compute and research toward using Sora’s world‑simulation tech to train robots.
Melania Trump opened the White House’s Fostering the Future Together summit by striding into the East Room with a Figure 3 humanoid.
A self-driving food delivery robot crashed into a Chicago bus shelter, for the second time this week — this one being a Coco Robotics unit in Old Town.
Munich-based Agile Robots struck a long-term deal to integrate Google DeepMind’s Gemini Robotics models into its industrial robots.
Hyundai and Persona AI are building a humanoid welding robot for shipyards, with a prototype due in late 2026 and commercial deployment targeted for 2027.
MIT and Symbotic built an AI system that uses deep reinforcement learning plus fast planning algorithms to choreograph hundreds of warehouse robots in real time.
McDonald’s is piloting Keenon Robotics’ humanoids at a Shanghai restaurant to greet customers, deliver food, and test customer-facing automation’s impact on service.
China’s Westlake Robotics unveiled Titan 01, a humanoid powered by its in‑house General Action Expert AI model that can mirror human movements in real time.
Rimac spinout Verne is teaming up with Uber and Pony.ai to launch a robotaxi service in Zagreb, putting the little-known Croatian startup on the map.
San José International Airport launched a four‑month pilot of “José,” an AI-powered humanoid from local startup IntBot that greets travelers at Terminal B.
Lucid Bots, a Charlotte-based startup that builds industrial cleaning drones and ground robots for window washing, raised $20M in a Series B round of funding.
Researchers built a silkworm-moth–inspired robot that can track odor sources indoors and outdoors almost as well, even after one of its two smell sensors fails.
COMMUNITY
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: ARC-AGI-3 resets frontier AI scoreboard
Read our last Tech newsletter: The deep sea luxury race is back
Read our last Robotics newsletter: OpenClaw craze comes to robots
Today’s AI tool guide: Create branded reaction GIFs for Slack
RSVP to next workshop today @ 2PM EST: Intro to Vibe Coding pt. 3
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — The Rundown’s editorial team
ARC-AGI-3 resets frontier AI scoreboard
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. One of the AI industry's favorite talking points of being on the doorstep of AGI just ran into a test where the best models in the world can't even score above 1%.
ARC-AGI-3 is a harder version of the benchmark that's become the go-to reality check for AGI claims — and with Gemini Pro leading the pack at just 0.37%, frontier models just got a brand new challenge (to likely still crush in about six months).
Reminder: Our next live workshop is today at 2 PM EST — join part 3 of our Intro to Vibe Coding, where you’ll learn how to take your apps to production for real users and workflows instead of just prototypes. RSVP here.
In today’s AI rundown:
ARC’s new AGI test stumps every frontier AI
Reddit's AI bot crackdown skips the ID check
Create branded reaction GIFs for Slack
Google shrinks AI memory with zero accuracy loss
4 new AI tools, community workflows, and more
LATEST DEVELOPMENTS
AI BENCHMARKS
🧐 ARC’s new AGI test stumps every frontier AI
Image source: ARC Prize Foundation
The Rundown: François Chollet's ARC Prize Foundation just released ARC-AGI-3, the newest version of its interactive reasoning benchmark, where humans can solve 100% of tasks on the first try but AI models struggle, with top systems not even scoring 1%.
The details:
Labs spent millions training models on earlier versions of the test, pushing ARC-AGI-2 scores from 3% to around 50% in under a year.
Agents face game-like scenarios with zero instructions, and must discover rules, form goals, and plan strategies entirely from scratch.
Google’s Gemini Pro scored the highest among frontier models at just 0.37%, followed by GPT 5.4 High (0.26%), Opus 4.6 (0.25%), and Grok-4.20 (0%).
A $1M prize backs the challenge, and cofounder Mike Knoop says frontier labs are paying far more attention to V3 than they did to earlier versions.
Why it matters: It’s always jarring to see the top models get reset below 1% on a new ARC-AGI release, but if the older tests are any indicator, even more surprising will be how quickly frontier labs climb the ladder. Whether that reflects genuine reasoning or just more expensive brute-forcing is exactly what Chollet built V3 to find out.
TOGETHER WITH SLACK FROM SALESFORCE
🧑💻 Your AI teammates live in Slack
The Rundown: Agentforce brings powerful AI agents directly into Slack, with no new logins or context switching. DM an agent, @mention it in a channel, or let it take action by pulling Salesforce insights, updating records, and creating canvases on the fly.
In this guide, you'll learn how to:
Get started with agents right where your team already works
Take action faster by pulling insights, updating records, and more
Get started in minutes with ready-made templates or build custom agents for any team
Read the full guide to get started with Agentforce in Slack.
🤖 Reddit's AI bot crackdown skips the ID check
Image source: Reddit
The Rundown: Reddit CEO Steve Huffman outlined a plan to separate humans from bots across the site, including labeling automated accounts, flagging suspicious users for verification, and letting sub-communities self-police without mass ID checks.
The details:
Accounts running automation in approved ways on the social platform will carry an [App] label, with suspicious behavior leading to human verification.
To confirm proof of humanity, Reddit will offer passkeys or Sam Altman’s World ID scanner, with government IDs as a last resort, only where laws require it.
AI-written content isn’t being banned, with Huffman calling it 'annoying' but saying communities can set their own rules on AI-generated posts.
Rival platform Digg recently folded after being overrun with bots, and Cloudflare data shows automated traffic on pace to surpass humans by 2027.
Why it matters: The Dead Internet Theory was already here before the AI agent acceleration we’ve seen over the past six months. Now, it’s a reality every social media site is dealing with. While this feels a bit like a band-aid, it is a small step towards every platform needing a serious human-first solution if it wants to remain usable to them.
AI TRAINING
🤯 Create branded reaction GIFs for Slack
The Rundown: In this guide, you will learn how to make custom, branded reaction GIFs for your company’s Slack using Higgsfield (an image and video generator). The trick is to generate the starting frame before you animate it.
Step-by-step:
Go to Higgsfield image gen, decide the GIF’s look, and enter the reaction’s visual style and text, like “ESPN themed reaction gif with words ‘SLOW DOWN’”
If your brand is not recognizable, attach your logo or another brand reference image while generating the still
Generate a few stills and pick the best one, then click the camera’s Animate button on that still so that it becomes the start frame in Higgsfield video
Then, set the clip length to 3 seconds, turn off its audio, and prompt: "Reaction GIF". Finally, download the MP4 and turn it into a GIF with any MP4-to-GIF site
Pro tip: If you make a whole batch of MP4s, ask Claude Code to convert them to GIFs in bulk on your desktop so you do not have to use a converter site one file at a time.
PRESENTED BY TELY AI
💬 Market leaders get leads from ChatGPT and Google
The Rundown: Your buyers are asking AI questions — and AI is answering with your competitors, not you. Tely makes AI like ChatGPT, Google, and Claude recommend your business instead.
With Tely AI, you can:
Get recommended in ChatGPT, Google, Perplexity, and Claude in as little as 1 week
Fully hands-off: no writers, no agencies, no managing content
Costs less than hiring freelancers or maintaining a marketing team
Ideal for niche industries where expertise matters
💾 Google shrinks AI memory with zero accuracy loss
Image source: Google
The Rundown: Google Research introduced TurboQuant, an algorithm that compresses AI model memory over 6x without any retraining — while delivering up to 8x speed gains on Nvidia H100 chips and losing almost zero accuracy.
The details:
AI models keep a running log of each conversation, and as chats get longer, that storage balloons, which slows responses and drives up costs.
TurboQuant shrinks that storage by over 6x with zero accuracy loss, scoring perfectly on tests that bury a key detail in a large amount of text.
On Nvidia's top server chips, it also sped up response processing up to 8x compared to standard methods, without adding any extra cost to run.
The paper, set to be presented at ICLR 2026 in April, also topped rival methods in vector search — the tech search engines use to match similar results quickly.
Why it matters: Despite being first published in April 2025, top AI memory companies felt the heat of the official release, with stocks dropping 3-5%. One compression paper won't crater memory demand overnight, but the selloff shows Wall Street is pricing in a world where smarter software cuts into the premium AI memory commands.
QUICK HITS
🛠️ Trending AI Tools
🎶 Lyria 3 Pro - Google’s upgraded AI music model with longer track outputs
🌐 MolmoWeb - Ai2's open-source web browsing agent
🎨 Uni-1- Luma's unified model that reasons and generates across text, images
⚙️ Composer 2 - Cursor’s powerful, cost-effective coding model
📰 Everything else in AI today
Oracle Data Deep Dive NYC, April 10th: Hands-on AI labs and direct access to Oracle experts. Learn more and register for free.*
OpenAI is raising another $10B to push its record funding round past $120B, with Microsoft, a16z, and T. Rowe Price joining the round.
Google upgraded its music AI model to generate full 3-minute songs with intros, verses, and choruses, with Lyria 3 Pro rolling out in Gemini, Vertex AI, and Google Vids.
Bret Taylor’s Sierra introduced Ghostwriter, an AI agent that builds other AI agents — letting companies create customer service bots across voice, chat, and 30+ languages.
The U.S. Department of Labor launched "Make America AI-Ready," a free 7-day AI literacy course delivered entirely over text message to promote AI upskilling.
*Sponsored Listing
COMMUNITY
🤝 Community AI workflows
Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.
Today’s workflow comes from reader May F. in London, UK:
"I’m on maternity leave, but wanted to build up my AI knowledge, so I’ve used Claude Code to build a custom dashboard of the data I’m tracking - feed time, naps, etc. I now get an email each morning with a summary of the previous day, with coaching tailored to my baby’s current age and development.”
How do you use AI? Tell us here.
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: OpenAI’s Sora gets the axe
Read our last Tech newsletter: The deep sea luxury race is back
Read our last Robotics newsletter: OpenClaw craze comes to robots
Today’s AI tool guide: Create branded reaction GIFs for Slack
RSVP to next workshop today @ 2PM EST: Intro to Vibe Coding pt. 3
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown
OpenAI's Sora gets the axe
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. Weeks ago, OpenAI's Applications CEO Fidji Simo told staff to stop chasing "side quests." Today, one of the AI leader’s most high-profile ones got the axe.
Sora, the video generator that previously hit No. 1 on the App Store and landed a major Disney deal, is officially winding down — with focus turning to a new incoming ‘Spud’ model and the enterprise fight with Anthropic.
In today’s AI rundown:
OpenAI winds down Sora to prioritize ‘Spud’
Brett Adcock’s $100M stealth AI device startup
Let Claude use your computer with Dispatch
Apple’s standalone Siri app, chatbot for iOS 27
4 new AI tools, community workflows, and more
LATEST DEVELOPMENTS
OPENAI
❌ OpenAI winds down Sora to prioritize ‘Spud’
Image source: Ideogram / The Rundown
The Rundown: OpenAI just axed its Sora AI video generator to free up compute for Spud, the company's next major model that Sam Altman says will be ready in weeks, with the company continuing to shift priorities and roles in the race against Anthropic.
The details:
Altman reportedly told staff OAI would wind down all video products, including its mobile app and API, with employees calling Sora a “drag” on resources.
The freed-up compute will go toward ‘Spud’, OAI's next major model expected in the coming weeks, which Altman says “can really accelerate the economy”.
Sora head Bill Peebles said the team will now pursue "world simulation" for robotics, calling the prize "automating the physical economy".
The move also puts OAI’s December partnership and $1B investment by Disney on ice after the media giant initially agreed for its IP to be used in Sora.
Altman is also reportedly moving safety responsibilities under Mark Chen, with Fidji Simo’s division also being renamed to “AGI Deployment”.
Why it matters: There had been rumors of Sora being woven into a ‘super app’, but the video appears to be deemed one of the company’s “side quests,” distracting from its main core focus. Big changes are clearly underway, and OAI’s next release will reveal a lot more about where the organizational winds are blowing.
TOGETHER WITH AIRIA
💼 Redefine enterprise AI with Airia
The Rundown: Airia delivers unified AI security, orchestration, and governance built for enterprises accelerating AI adoption. Deploy agents fast while maintaining control, bridging the gap between innovation and compliance.
Dive deeper with Airia’s 2026 State of AI Report, which reveals:
Key insights on enterprise AI adoption trends
Critical challenges and emerging opportunities ahead
A comprehensive guide for navigating AI transformation
HARK
🤖 Brett Adcock’s $100M stealth AI device startup
Image source: Hark
The Rundown: Figure AI founder Brett Adcock just launched Hark, a new AI startup that aims to create a “new interface to AGI” via personalized AI paired with dedicated hardware.
The details:
Hark spent 8 months in stealth, with Adcock putting $100M of his own money in the startup, targeting what he calls the most advanced personal AI ever built.
The company is working on a “family of devices both for yourself and the home”, with systems that “begin to think like you, and sometimes ahead of you.”
The 45-person team pulls from Apple, Google, Meta, and Tesla, with hardware design led by Abidur Chowdhury, who previously designed the iPhone Air.
Hark signed a deal for thousands of NVIDIA B200 GPUs arriving in April, with the first AI models and software planned for this summer.
Why it matters: The AI device space has been tough sledding, but Adcock’s experience with Figure gives him a stronger pedigree than most. Paired with an ex-Apple design lead and strong team, Hark is one of the more interesting moves into the space yet — and joins OAI’s device on the list of hotly anticipated AI hardware releases.
AI TRAINING
📲 Let Claude use your computer with Dispatch
The Rundown: In this guide, you will learn how to use Dispatch to hand Claude a task from your phone and have it finish the work on your PC — a simple way to keep work organized when you step away and want Claude to keep moving in the background.
Step-by-step:
Update the Claude Desktop app on your computer by clicking Claude > Check for updates in the toolbar. Update the Claude mobile app, too
Open the Desktop app, click Dispatch > Get started, and turn on file access and keep-awake to ensure Claude works with files while keeping the PC active
When prompted, pair Dispatch with your phone app. Now, send Claude a simple task from your phone, like cleaning up emails or organizing files
Check the result on your desktop or mobile when Claude finishes. The same conversation stays synced, so you can keep going without restarting the thread
Pro tip: Set up a local project folder before you start. Use one clean folder for the files Claude should edit, save, or organize so the output is easy to find later.
PRESENTED BY LAMBDA
⚡Cut your AI training costs by 25% or more
The Rundown: Most large-scale AI training runs use less than half the computing power they're paying for. Lambda's team found the root causes and built a reproducible framework that boosted efficiency by over 25%, without changing the model itself.
Lambda’s whitepaper shows you how to address:
Memory inefficiencies silently inflating your costs
Training configurations that aren't making full use of your hardware
Bottlenecks that slow down GPU communication
APPLE
📱 Apple’s standalone Siri app, chatbot for iOS 27
Image source: Apple
The Rundown: Apple is reportedly testing a standalone Siri app alongside a new "Ask Siri" chatbot experience, according to Bloomberg insider Mark Gurman — with both slated to debut with its iOS 27 system at WWDC in June.
The details:
Gurman said Siri will get its own dedicated app for the first time, along with a redesigned interface that lets users type or speak requests.
The assistant will reportedly read across iMessages, emails, and notes to build context, also capable of executing actions inside third-party apps directly.
Apple is branding the experience "Ask Siri," positioning it as a chatbot-style interaction rather than the rigid voice commands of the previous iterations.
The full rollout is expected at WWDC on June 8 as part of iOS 27 and macOS 27, following a lackluster reception to Apple Intelligence last year.
Why it matters: After Apple Intelligence failed, this Gemini-powered revamp is probably Apple’s last chance to define Siri before users default to ChatGPT or Claude for everything. June 8 just became the most important keynote in years, though after the initial 2024 launch, we’ll have to see an actual rollout before believing the demos.
QUICK HITS
🛠️ Trending AI Tools
🗣️Unwrap Customer Intelligence - Turn unstructured customer feedback into data-backed insights that inform your product roadmap*
🧑💻 Figma - New MCP bridge for designing directly on canvas via agentic tools
💼 Dynamic Workers - Cloudflare's sandbox for running AI agent code at scale
💻 Claude Code - New auto mode for a hands-free coding permission system
*Sponsored Listing
📰 Everything else in AI today
Microsoft AI poached three top researchers from the Allen Institute for AI, including former CEO Ali Farhadi, with the trio joining Mustafa Suleyman’s Superintelligence team.
Figma opened its design canvas to coding agents, letting tools like Claude Code create/ edit designs directly using a team's existing components and brand standards.
Pharma giant Roche launched a new "AI factory" to speed up drug discovery and manufacturing, with the company now amassing over 3,500 Nvidia Blackwell GPUs.
Cloudflare launched Dynamic Workers, a new tool that lets AI agents quickly and safely run code they write on the fly at 100x the speed of other options.
The OAI Foundation committed $1B this year across disease, job displacement, and AI safety research, with co-founder Wojciech Zaremba joining as Head of AI Resilience.
COMMUNITY
🤝 Community AI workflows
Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.
Today’s workflow comes from reader Leslie in Delray Beach, FL:
"I give ChatGPT my product sales report and return reports and ask it to analyze my best-selling products and recommendations for improvement, and recommendations on what to discontinue. I sell 600 products, so doing this manually is impossible.
It also identified my very top opportunities, and several products were a shock to me as I had no idea they were doing that well. It also made me aware of products that are dragging my business down."
How do you use AI? Tell us here.
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: Anthropic’s Claude gets remote control
Read our last Tech newsletter: The deep sea luxury race is back
Read our last Robotics newsletter: OpenClaw craze comes to robots
Today’s AI tool guide: Let Claude use your computer remotely
RSVP to next workshop on Thursday @ 2PM EST: Intro to Vibe Coding pt. 3
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown
The deep-sea luxury race is back
Read Online | Sign Up | Advertise
Good morning, tech enthusiasts. China wants to take tourists to sunless depths: a state-backed deep-sea submersible is being built to carry paying passengers 1K meters (3,280 ft.) down into the ocean’s eerie “Midnight Zone.”
Less than two years after OceanGate’s Titan turned deep-sea tourism into a global cautionary tale, China is betting the deep still holds plenty of appeal, at least for anyone rich enough to buy a window seat.
In today’s tech rundown:
China eyes deep-sea tourism with new sub
The FCC bans all foreign-made routers
Alphabet brings drone delivery to the Bay Area
Sam Altman’s fusion future is getting serious
Quick hits on other tech news
LATEST DEVELOPMENTS
DEEP SEA TECH
🧜🏽♀️ China eyes deep-sea tourism with new sub
Image source: Ideogram / The Rundown
The Rundown: China is developing a deep-sea submersible that can carry paying passengers up to 1K meters (3,280 ft.) deep — a state-backed entry into the ultra-luxury “extreme travel” market that’s been reshaped by the 2023 OceanGate incident.
The details:
Engineers at the China Ship Scientific Research Centre in Wuxi have finalized the design of a 4-person vessel, with a prototype expected by year-end.
Commercial dives are targeted before 2030, with a panoramic viewing dome designed to offer sightseeing in what oceanographers call the Midnight Zone.
At 1K meters, pressure hits at 100x of the surface, demanding zero-tolerance hull engineering, shatter-resistant viewports, and fail-safe life support.
The OceanGate Titan disaster, which killed five, was attributed to flawed engineering, insufficient testing, and a rejection of industry safety standards.
Why it matters: The deep-sea tourism market didn’t end with OceanGate but rather regrouped. China’s sub joins an elite circuit that’s looking to send the wealthy to the furthest depths. This includes U.S.-based Triton Submarines, which is developing a Titanic-rated vessel, and U-Boat Worx, building underwater cars and “party subs.”
TRADE WARS
🚫 The FCC bans all foreign-made routers
Image source: Ideogram / The Rundown
The Rundown: The FCC just added all new foreign-made home routers to its national security “covered” list, effectively banning them from the U.S. market unless manufacturers can obtain a security exemption — a high bar few are expected to clear.
The details:
Routers already in homes can stay, and already approved foreign models can keep shipping for now, but the pipeline for new foreign hardware is largely shut.
The move particularly targets Chinese-linked manufacturers after U.S. intelligence tied compromised routers to cyberattacks on critical infrastructure.
TP-Link, the Chinese brand that dominates Amazon’s bestseller list, had already drawn scrutiny following a string of high-profile intrusions.
Why it matters: Coming on the heels of the drone ban, the FCC’s move hands Washington sweeping control over the hardware that carries most U.S. internet traffic. For consumer-router supply chains built around overseas (and especially Chinese) manufacturing, the pressure to reconfigure is now acute.
WING
🪽 Alphabet brings drone delivery to the Bay Area
Image source: Wing
The Rundown: Alphabet’s drone delivery unit Wing is bringing its fleet to the San Francisco Bay Area later this year, partnering with Walmart and DoorDash to offer residential deliveries across one of the country’s most tech-forward markets.
The details:
Wing says it has already logged more than 750K deliveries across Houston, Atlanta, Charlotte, Virginia, as well as Australia.
A national last-mile network is the goal, with Wing pitching its drones as faster and cleaner than delivery vans.
The Walmart partnership is set to scale to more than 270 stores by 2027, with DoorDash adding consumer reach on the platform side.
Urban airspace complexity and FAA regulatory constraints remain the biggest variables in how fast Wing and its rivals can actually expand.
Why it matters: After years of suburban pilots, Alphabet is making a direct play for commercial relevance for its moonshot drone program, just as the competitive pressure is heating up. Zipline just snapped up an extra $200M, and Amazon’s Prime Air is rolling out across the U.K. in 2026.
HELION
⚡️ Sam Altman’s fusion future is getting serious
Image source: Helion Energy
The Rundown: OpenAI CEO Sam Altman has resigned as chair of Helion Energy, the fusion startup he has backed for a decade, as the company enters early talks on a deal that could supply a significant share of OpenAI’s future electricity needs.
The details:
Helion and OpenAI are reportedly in early talks about a deal that could give OpenAI rights to 12.5% of Helion’s electricity output if the fusion tech pans out.
An early framework envisions Helion delivering up to about 5 gigawatts by 2030 and 50 gigawatts by 2035 from thousands of 50-megawatt reactors.
The company, which counts Altman among its backers, has raised $425M and already holds power agreements with Microsoft and steelmaker Nucor.
Helion’s system is designed to capture energy directly from expanding plasma via magnetic coils, bypassing steam turbines that most power plants rely on.
Why it matters: Altman is stepping away from the board to remove potential conflicts as Helion pursues a commercial relationship with OpenAI — the same move he made at nuclear startup Oklo. The pattern is deliberate: tie frontier AI’s future to a new generation of low-carbon energy sources that don’t fully exist yet.
QUICK HITS
📰 Everything else in tech today
Apple plans to introduce Google‑style search ads in Apple Maps as soon as this summer as part of a broader push to grow its services revenue, Bloomberg reports.
Samsung’s new Galaxy S26 phones can now use their Quick Share feature to send files directly to iPhones, iPads, and Macs via Apple’s AirDrop.
OnlyFans owner Leonid Radvinsky died of cancer at 43 while reportedly in talks to sell a majority stake in the company at a roughly $5.5B valuation.
Prediction platforms Kalshi and Polymarket are tightening trading rules by barring insiders from betting, as senators push to ban sports-style bets on prediction markets.
London fintech firm Revolut posted a record 2025 pretax profit of $2.3B on $6B in revenue, sharply higher than 2024, as it gears up for a big U.S. expansion.
Russia launched 16 Rassvet broadband satellites to low-Earth orbit, marking the start of a Starlink-style domestic internet constellation positioned as a rival to SpaceX.
Nintendo is cutting planned Switch 2 production by more than 30%, trimming this quarter’s output from 6M to 4M consoles after weaker-than-expected demand.
A French Navy officer jogging on the deck of the aircraft carrier used Strava to log his run, inadvertently exposing the warship’s precise location in the Mediterranean.
Sony is close to a $1B deal to sell a majority stake in its home entertainment business, including TVs, to Chinese rival TCL.
Singapore-based super app Grab agreed to buy Delivery Hero’s Foodpanda food-delivery operations in Taiwan for $600M in cash.
COMMUNITY
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: Anthropic's Claude gets remote control
Read our last Tech newsletter: Amazon’s secret phone project
Read our last Robotics newsletter: OpenClaw craze comes to robots
Today’s AI tool guide: Free up space on your computer with Claude
RSVP to next workshop on Thursday @ 2PM EST: Intro to Vibe Coding pt. 3
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — The Rundown’s editorial team
Anthropic's Claude gets remote control
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. Remote agents like OpenClaw are all the rage right now, and Anthropic's shipping spree is quickly giving Claude the building blocks to become one itself.
Claude can now control a Mac directly, clicking and handling tasks while you manage everything from your phone — the latest in a string of releases transforming the assistant from a chatbot into a full-time digital employee.
In today’s AI rundown:
Anthropic ships remote computer use
Luma AI’s new image model thinks as it generates
Free up space on your computer with Claude
Zuck ramps up Meta’s internal AI agent use
4 new AI tools, community workflows, and more
LATEST DEVELOPMENTS
ANTHROPIC
🚀 Anthropic ships remote computer use
Image source: Anthropic
The Rundown: Anthropic just released a research preview that hands Claude direct control of your desktop — letting it click, type, and navigate across any app on your Mac while you step away, with phone-based task assignment through Dispatch.
The details:
The newly released Dispatch turns the combo into a remote setup, allowing users to fire off a task from mobile and letting Claude handle it on the computer.
The system is built to avoid screen control when possible, checking for direct app integrations and browser access before resorting to clicking.
The feature is only available to macOS users on Pro or Max plans currently via Cowork and Claude Code, with a Windows version also in the pipeline.
Anthropic acquired computer use startup Vercept in February, with the new release marking the team’s first product launch after just four weeks.
Why it matters: Anthropic’s Alex Albert puts it well, saying, “the future where I never have to open my laptop to get work done is becoming real very fast”. While losing OpenClaw to OAI was considered by many to be a miss, the recent flurry of features has shown the building blocks forming to turn Claude into its own remote agent.
TOGETHER WITH ORACLE
😵💫 Why your AI has amnesia and how to fix it
The Rundown: The race to create an amazing Agentic AI experience is hitting a wall because of AI’s memory gap between sessions. Oracle shares how proper memory infrastructure in a ‘convergent database’ is the key to a competitive advantage creating an authentic Agentic AI digital partner.
In this blog, you’ll find:
Why context windows fall short for agent memory
Technical practices for agent memory competitive advantage
Code snippets and the notebook to get started
LUMA LABS
🎨 Luma AI’s new image model thinks as it generates
Image source: Luma AI
The Rundown: Luma AI rolled out Uni-1, an image model that processes text and visuals through the same pipeline — thinking through what it's asked to do before and while it creates, with the company calling this approach "path to general intelligence."
The details:
Uni-1 runs on the same type of architecture as GPT Image 1.5 and Nano Banana Pro, processing text and images in a single pipeline instead of diffusion.
The model also features real-world understanding, enabling creative decisions and use cases such as infographics, manga, and specific aesthetics.
In testing, Uni-1 topped human preference rankings for style, editing, and reference-based work, trailing only Nano Banana Pro in text-to-image ELO.
Uni-1's API price of ~$0.09 / image at 2K resolution undercuts Nano Banana Pro's $0.134 rate by roughly a third, though the API is waitlist-only for now.
Why it matters: Luma made its name in video, so an image model is a new direction. If the same system can extend into video, voice, and interactive worlds as Luma is teasing, Uni-1 could set the foundation for one model that can do it all creatively — moving into the creative agent territory that users are starting to expect.
AI TRAINING
💾 Free up space on your computer with Claude
The Rundown: In this guide, you will learn a simple way to use Claude as a storage-cleanup copilot. If you use AI to code, you’re likely eating up a ton of storage space without even knowing it.
Step-by-step:
Start Claude/Claude Code and prompt: “I need to free up Mac storage, but I don’t want to delete anything important. Ask me about my tools, find storage culprits, and help me investigate before suggesting any cleanup commands”
Answer the questions and ask Claude to rank the culprits by size and risk. It should give you terminal commands to check the size of top storage-eaters
Work through each category. Have Claude explain what the files are, why they get large, what is safe to remove, and what should be reviewed manually
Once identified, prompt: “Start with the lowest-risk cleanup wins first. For each one, explain what would be removed, how much space it might save, and anything I should double-check before deleting it”
Pro tip: Ask Claude to turn this into a reusable skill file that you can feed into any AI.
PRESENTED BY ORKES
🏗️ The 4-layer stack every AI agent needs
The Rundown: Orkes is the agentic orchestration platform that powers scalable, reliable, and agentic workflows across modern applications — backed by Conductor OSS, originally built at Netflix. Join our upcoming webinar to see how modern engineering teams are building production-ready agents.
The April 16th session will cover:
How to expose APIs and services as agent tools through an MCP gateway
A framework for composing, iterating, and executing agent workflows on a fault-tolerant orchestration layer
Best practices for tracking, auditing, and governing agent behavior in production
Register now for the April 16th webinar.
META
🤖 Zuck ramps up Meta’s internal AI agent use
Image source: Lovart / The Rundown
The Rundown: Mark Zuckerberg is creating a personal "CEO agent" to shortcut the chain of command when he needs quick answers, according to the WSJ, coming as part of a company-wide mandate that now factors AI usage into performance reviews.
The details:
Zuck's agent is still in development, but already handles tasks like pulling answers that typically require going through multiple layers of Meta's org chart.
Staffers have spun up custom agent tools, including one called "My Claw" that reads their work files and negotiates with coworkers' bots directly.
Another Claude-powered internal tool called "Second Brain" acts as an AI chief of staff, pulling answers from any internal document on demand.
Zuckerberg had previously courted OpenClaw creator Peter Steinberger, and also acquired Chinese agentic platform Manus in December.
Why it matters: Meta may have tens of thousands of employees, but that isn’t stopping the newer parts of the org from trying to move as fast and lean as some of its more AI-native rivals. With Zuck seemingly very invested in the AI agent boom, Meta’s integration of Manus will be one of the more interesting implementations to watch for.
QUICK HITS
🛠️ Trending AI Tools
🔒 Incogni - remove your personal data from the web so scammers and identity thieves can’t access it. Limited time offer: use code RUNDOWN to get 58% off*
🎨 Uni-1- Luma's unified model that reasons and generates across text, images
🚀 Stitch - Google’s newly updated UI creation tool for “vibe design”
🦞 NemoClaw - Nvidia's open-source security layer for OpenClaw agents
*Sponsored Listing
📰 Everything else in AI today
Nvidia CEO Jensen Huang appeared on the Lex Fridman Podcast, saying, “I think it’s now. I think we’ve achieved AGI” when asked about his intelligence timelines.
Apple announced its WWDC 2026 event will run June 8-12, teasing ‘AI advancements’ that are speculated to include its Siri overhaul powered by Google Gemini.
OpenAI is reportedly guaranteeing a 17.5% minimum return to lure private equity firms into its enterprise joint venture — outbidding Anthropic as both prep for IPOs.
Agentic personal software builder Dreamer announced it is licensing its tech to Meta, with its full team joining Meta Superintelligence Labs in an undisclosed deal.
OpenAI hired former Meta VP of global clients Dave Dugan to run its ad sales, coming as the company continues its initial advertising push into ChatGPT.
COMMUNITY
🤝 Community AI workflows
Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.
Today’s workflow comes from reader Mark C. in Toronto, Canada:
"I was in a vehicle accident and was assessed to be at fault. I uploaded my dash cam video to Gemini and prompted it to use the persona of a skilled traffic lawyer in my jurisdiction and to defend me.
It produced a detailed 3-page letter to my insurance company, quoting numerous sections of legislation and associating each with time stamps on the video. I was reassessed to be not at fault."
How do you use AI? Tell us here.
🎓 Highlights: News, Guides & Events
Read our last AI newsletter: Elon Musk’s ‘Terafab’ AI chip factory
Read our last Tech newsletter: Amazon’s secret phone project
Read our last Robotics newsletter: OpenClaw craze comes to robots
Today’s AI tool guide: Free up space on your computer with Claude
RSVP to next workshop on Thursday @ 2PM EST: Intro to Vibe Coding pt. 3
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown
No matching search results
Try using different keywords, double-check your spelling, or explore related categories.
Stay Ahead on AI.
Join 2,000,000+ readers getting bite-size AI news updates straight to their inbox every morning with The Rundown AI newsletter. It's 100% free.