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.
Nvidia's GR00T N1 AI for humanoids
Read Online | Sign Up | Advertise
Good morning, robotics enthusiasts. At this week’s jam-packed Nvidia GTC conference, CEO Jensen Huang unveiled Isaac GR00T N1, the world’s first open-source and fully customizable foundation model for humanoids.
Huang says that “the age of generalist robotics is here,” and that GR00T N1 is what developers need to train smarter, more capable humanoids—faster than we ever imagined. But will Nvidia’s advances live up to the hype?
In today’s robotics rundown:
Nvidia’s open-source AI for humanoids
Disney’s Star Wars droid comes to life
A robot that leaps like a squirrel
Humanoid with touch-sensitive ‘skin’
Quick hits on other robotics news
LATEST DEVELOPMENTS
NVIDIA
🧠 Nvidia’s open-source AI for humanoids
Image source: Nvidia
The Rundown: At its GTC 2025 conference in San Jose, Nvidia unveiled Isaac GR00T N1, its open-source foundational model designed to accelerate the development and capabilities of humanoids and “open the next frontier in the age of AI.”
The details:
GR00T N1 features a dual system architecture designed for “fast” and “slow” thinking systems, inspired by principles of human cognition.
The slow thinking system enables robots to perceive, reason, and plan actions, while the fast thinking system translates these plans into real-world action.
Nvidia is also releasing simulation frameworks and blueprints for generating synthetic training data, making GR00T N1 highly adaptable for various uses.
Pre-trained on both synthetic and real data, this model has demonstrated tasks such as grasping, object manipulation, and executing multi-step instructions.
Why it matters: While GR00T N1 is a big jump in humanoid robotics, its ability to fully deliver on its promises will become clearer as it's more widely integrated and tested in various real-world applications. Companies that are already implementing the tech in their humanoids include 1X, Boston Dynamics, Agility, and Figure.
DISNEY
🎢 Disney’s Star Wars droid comes to life
Image source: Nvidia/YouTube
The Rundown: During his GTC keynote, Jensen Huang revealed Nvidia has partnered with Disney and DeepMind to develop Newton, an open physics engine aimed at bringing a new wave of robots—starting with a Star Wars-inspired droid named “Blue.”
The details:
Huang introduced the “Blue” prototype on stage, showcasing how it uses Newton for complete real-time simulation and action.
Built on Nvidia's Warp framework, Newton will provide a customizable and scalable simulation environment, optimized for AI-driven robotic characters.
Nvidia says Newton will be compatible with DeepMind's MuJoCo and its own Isaac Lab, accelerating robotics workloads by over 70x.
It plans to release an early, open-source version of Newton later in 2025, allowing developers worldwide to experiment with the technology.
Why it matters: BDX droids like Blue are expected to roam around Disney parks this year, with Nvidia and DeepMind playing an integral role in powering the company’s future in robotics. However, it’s worth noting that the usefulness of this tech will go far beyond entertainment to more capable, expressive robots across various industries.
UC BERKELEY
🐿️ A robot that leaps like a squirrel
Image source: UC Berkeley
The Rundown: UC Berkeley researchers developed Salto, a squirrel-like robot that can take a flying leap and land on a narrow pipe—marking what the team says is the first time a bot has been able to stick the landing on such a small target.
The details:
Salto is the result of extensive research into squirrel biomechanics, including how a squirrel’s front legs absorb kinetic energy from jumping.
The robot features a motorized flywheel for balance, adjustable leg forces, and a passive gripper designed to minimize torque applied to the landing spot.
In tests, it successfully leaped from one PVC pipe to another 25 times out of 30, with two trials resulting in a perfect upright balance on the pipe.
Originally developed in 2016, Salto has been a work in progress; it started sticking landings on flat surfaces in 2020.
Why it matters: The team hopes that Salto’s small size and high, reactive jumps could allow the bot to quickly move through uneven terrain, which could mean life and death in search and rescue missions. Salto could also lead the way to agile bots working on construction sites, hopping from pipes while carrying cameras for inspection.
NEURA ROBOTICS
🇪🇺 Humanoid with touch-sensitive ‘skin’
Image source: Neura Robotics
The Rundown: German robotics company Neura Robotics is launching its third-gen 4NE-1 humanoid in June, with CEO David Rege saying that Europe’s new take on the generalist robot will be the best in the world.
The details:
The 4NE-1 stands nearly 6 feet tall and features 3D vision, touch sensors, and AI-driven cognition designed after human thought processes.
However, Neura says the difference is in its advanced sensor “skin,” enabling it to sense touch, predict contact, and measure pressure with precision.
The company is developing the cognitive humanoid using Nvidia’s GR00T N1 AI — joining players like Boston Dynamics and Agility Robotics with early access.
It has doubled its workforce to over 300 employees in a year and recently secured $130M in Series B funding for its Neuroverse platform.
Why it matters: In an era of tariff wars, Neura sees the development of its all-purpose 4NE-1 as crucial for European competitiveness. This push from the company comes as the UK’s Humanoid is still in the process of developing its robot and Norway’s 1X is gearing up to manufacture tens of thousands of units of its NEO humanoid by 2026.
QUICK HITS
📰 Everything else in robotics today
Apptronik, the company behind Apollo humanoid, announced it added an extra $53M to its Series A funding round, bringing the total funding from the round to $403M.
Boston Dynamics released a clip of its Atlas humanoid demonstrating some impressively fluid moves, including crawling, breakdancing, and cartwheeling.
Nvidia is partnering with GM to advance the autonomous driving systems of the carmaker as well as its manufacturing processes.
Unitree released a video showing its G1 humanoid performing a standing side flip—a world-first—using 23 degrees of freedom and cutting-edge technology.
Chinese developer Dobot unveiled a new humanoid called Atom, which has 28 degrees of freedom, ±0.05 mm precision, and a price tag of $27K.
PETA India donated a robotic elephant to a temple in India to be used in place of live animals in ceremonies, in a commitment to protecting endangered Asian elephants.
Chinese researchers developed a tiny robot to explore ocean depths of 10,600 meters (34,776 feet), now being tested in the Mariana Trench.
Researchers from Princeton and the University of Edinburgh found that an AI-powered robot can take over coffee-making duties in busy kitchens and cafes.
U.S. researchers created a reinforcement learning-based framework that allows legged robots to successfully ride a skateboard.
Chinese scientists developed a six-legged robot inspired by insect movements, for future mining operations on the Moon and missions to deep space asteroids.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Friday at 4 PM EST to learn about how to use AI to transform and enhance images using AI for content creation and creative projects, with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
See you soon,
Rowan, Jennifer, and Joey—The Rundown’s editorial team
AI's 'Moore's Law' emerges
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. The AI capability curve just found its "Moore's Law" moment — with new research showing task completion abilities doubling every 7 months since 2019.
With systems tackling hour-long human tasks today and potentially month-long projects by 2030, is the world prepared for the automation tsunami quickly approaching?
In today’s AI rundown:
AI capabilities following ‘Moore's Law’
Hollywood against AI copyright proposals
Improving non-reasoning AI responses
Nvidia’s open-source reasoning models
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
AI RESEARCH
📈 Study: AI capabilities following ‘Moore's Law’
Image source: Metr
The Rundown: Researchers at METR just published new data showing that the length of tasks AI agents can complete autonomously has been doubling approximately every 7 months since 2019, revealing a "Moore's Law" for AI capabilities.
The details:
The study tracked human and AI performance across 170 software tasks ranging from 2-second decisions to 8-hour engineering challenges.
Top models like 3.7 Sonnet have a "time horizon" of 59 minutes — completing tasks that take skilled humans this long with at least 50% reliability.
Older models like GPT-4 can handle tasks requiring about 8-15 minutes of human time, while 2019 systems struggle with anything beyond a few seconds.
If the exponential trend continues, AI systems will be capable of completing month-long human-equivalent projects with reasonable reliability by 2030.
Moore's Law predicts that computing power doubles roughly every two years — explaining why devices get faster and cheaper over time.
Why it matters: The discovery of a predictable growth pattern in AI capabilities provides an important forecasting tool for the industry. Systems that can handle much longer (months-long tasks for humans) and more complex tasks independently will completely reshape how businesses across the world approach AI and automation.
TOGETHER WITH MONGODB
📘 The hub for building better GenAI
The Rundown: Step into MongoDB’s Generative AI Use Cases Repository to discover how MongoDB powers GenAI applications— from text and vector search to robust AI Retrieval methods.
Inside, you’ll find:
Notebooks on real-world scenarios like advanced RAG techniques and AI agents
Integration with top LLM providers and frameworks
Pre-built datasets and embedding tools to accelerate your GenAI development
A central resource and community for developers using MongoDB
CELEBRITIES VS. AI
⭐️ Hollywood against AI copyright proposals
Image source: Ideogram / The Rundown
The Rundown: More than 400 Hollywood creatives signed an open letter urging the Trump administration to reject OpenAI and Google’s proposals to expand AI training on copyrighted works—arguing that it would let them "freely exploit" creative industries.
The details:
The letter is a direct response to OpenAI and Google's AI Action Plan submissions, which argued for expanded fair use protections for AI training.
OpenAI framed AI copyright exemptions as a "matter of national security," while Google said the current fair use framework already supports AI innovation.
Ben Stiller, Mark Ruffalo, Cate Blanchett, Paul McCartney, Taika Waititi, and Aubrey Plaza are among the high-profile creatives who have signed the letter.
They have emphasized that AI companies could simply "negotiate appropriate licenses with copyright holders — just as every other industry does."
Why it matters: Hollywood vs. AI represents a values collision — the tech industry’s "move fast and iterate" mindset vs. Hollywood's centuries-old IP frameworks. But with AI giants across the globe already ingesting the world’s data even without copyright protections, this fight, in reality, may be more symbolic than action-oriented.
AI TRAINING
🧠 Improving non-reasoning AI responses
The Rundown: In this tutorial, you will learn how to dramatically improve the intelligence of non-reasoning AI models by implementing a structured reasoning approach with XML tags—forcing the model to think step-by-step before answering.
Step-by-step:
Structure your prompt with XML tags like <thinking> and <answer> to separate the reasoning process from the final output.
Provide specific context and task details, including examples.
Force step-by-step reasoning by explicitly instructing the model to “think” first, then answer.
Compare results with and without your reasoning framework to see the dramatic improvements in quality.
Pro tip: You can use this technique especially when asking AI to match writing styles or analyze complex information before generating content.
PRESENTED BY DAGSTER
🚀 Streamline your AI stack with Dagster
The Rundown: Dagster consolidates your AI capabilities into one powerful orchestrator that developers love — helping reduce costs, eliminate complexity, and ensure reliable pipelines from prototype to production.
With Dagster, you can:
Consolidate all AI capabilities under one intuitive interface
Save 40%+ on infrastructure costs by optimizing AI workloads
Ship AI features 3x faster with standardized development practices
Schedule a demo to learn more about how Dagster can simplify your AI platform.
NVIDIA
🧠 Nvidia’s open-source reasoning models
Image source: Nvidia
The Rundown: Nvidia released its Llama Nemotron family of open-source reasoning models, designed to accelerate enterprise adoption of agentic AI capable of complex problem-solving and decision-making.
The details:
The new model family comes in three sizes: Nano (8B), Super (49B), and Ultra (249B) — each optimized for different deployment scenarios.
Early benchmarks show impressive performance, with the Super version outperforming both Llama 3.3 and DeepSeek V1 across STEM and tool testing.
The models feature a toggle that allows AI systems to switch between intensive reasoning and direct responses based on the task.
Post-training resulted in 20% better accuracy than base Llama models and 5x faster speed than rival open reasoners.
Nvidia is also releasing an "AI-Q Blueprint" framework in April to help businesses connect AI agents with their existing systems and data sources.
Why it matters: Nvidia’s reasoning models may be overshadowed by the insane amount of releases over the past 48 hours, but the chipmaking giant has seemingly built every block necessary to be a force across the entire AI stack — from the most advanced hardware to high-quality reasoning models ready for the agentic era.
QUICK HITS
🛠️ Trending AI Tools
💎 Diamond - Graphite’s agentic AI-powered code review companion
📋 Canvas - Gemini’s new collaborative space for document editing and coding
🎥 Stable Virtual Camera - Images into 3D videos with dynamic camera paths
🧊 Hunyuan 3D 2.0 MV - Open model for high-quality 3D shape generations
📰 Everything else in AI today
Google AI and UC Berkeley researchers proposed "inference-time search" as a new AI scaling method, producing several answers in parallel and selecting the best option.
LG released EXAONE Deep, a reasoning AI that achieves comparable performance to models like DeepSeek V1 in math, science, and coding with just 32B parameters.
Muse released Muse S Athena, a headband wearable combining EEG and sensors to measure both brain activity and oxygen levels for AI-powered cognitive fitness training.
Nvidia and xAI are joining Microsoft, BlackRock, and MGX in the AI Infrastructure Partnership, aiming to raise $30B initially and potentially $100B for AI data centers.
xAI debuted its first image generation API featuring the ‘grok-2-image-1212’ model, allowing developers to create multiple JPG images per request at $0.07 each.
Microsoft is partnering with neuroscience AI startup Inait to develop brain-inspired AI that learns from real-world experiences rather than data patterns.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Friday at 4 PM EST to learn how to use AI to transform and enhance images for content creation and creative projects with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Nvidia's 'AI Super Bowl' kicks off
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. One of AI’s greatest showmen just kicked off his own ‘AI Super Bowl’, with Nvidia CEO Jensen Huang delivering a wide-ranging keynote on the chip leader’s latest and greatest.
With Huang's dramatic reveals of new AI chips, personal supercomputers, Star Wars-style robots, and a fleet of autonomous vehicles, the company’s explosive growth may be just getting started.
P.S. Our next workshop is TODAY at 4 PM EST — come learn how to apply AI to enhance advertising and creative production workflows with Kling AI! RSVP here.
In today’s AI rundown:
Nvidia’s AI and robotics advances
Adobe’s army of enterprise AI agents
Expand Claude's abilities using MCPs
Anthropic testing new voice features
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
NVIDIA
🏟️ Nvidia’s AI and robotics advances
Image source: NVIDIA
The Rundown: “AI Jesus” Jensen Huang kicked off Nvidia’s GTC 2025 conference with a two-hour-long keynote, calling the event “AI’s Super Bowl” and revealing exciting new updates on upcoming chip releases, robotics, autonomous vehicles, and more.
The details:
Nvidia’s new GPU lineup includes Blackwell Ultra in late 2025, Vera Rubin in 2026, and Feynman in 2028, each promising major performance gains.
Huang said that scaling is not slowing down, and the computation needed for AI is “easily 100x more than we thought we needed at this time last year.”
He also revealed Isaac GR00T N1, the first open humanoid robot foundation model, alongside a comprehensive physical AI dataset for training robots.
The new DGX Spark and DGX Station bring data center-grade AI computing to personal workstations, with Huang calling it “the computer for the age of AI.”
A robotics physics engine, Newton, also debuted in collaboration with Google DeepMind and Disney — demoed with ‘Blue’, a Star Wars-style robot on stage.
Nvidia also announced a new partnership with automaker GM, with plans to build the company’s first fleet of self-driving cars.
Why it matters: Huang said that AI is currently at an ‘inflection point’ — and this sweeping set of announcements shows the sheer scale of Nvidia’s web of AI-powered infrastructure for nearly every industry and use case. If the chipmaking leader’s roadmap and releases are any indication, AI is not slowing down any time soon.
TOGETHER WITH VANTA
🦺 Simplify compliance with automation and AI
The Rundown: Navigating new compliance requirements can be daunting, but with the right automation tools, it doesn't have to be. Whether you’re a fast-growing startup or a big security team, Vanta can help you achieve continuous compliance (and more).
Join the live demo on April 3 to learn how Vanta can help you:
Automate compliance for frameworks like SOC 2, ISO 27001, HIPAA, HITRUST, and ISO 42001
Continuously monitor controls to build and reinforce your security foundation
Scale your compliance program and manage vendor risk with automated tools and AI
ADOBE
🤖 Adobe’s army of enterprise AI agents
Image source: Adobe
The Rundown: Adobe just released a comprehensive AI agent strategy centered around its new Experience Platform Agent Orchestrator, introducing ten specialized agents for enterprise tasks like customer experiences, marketing workflows, and more.
The details:
The platform includes agents for audience targeting, content production, site optimization, and B2B account management within Adobe's enterprise apps.
A Brand Concierge helps businesses create personalized chat experiences, with traffic from AI platforms to retail sites jumping 1,200% in February.
Adobe Marketing Agent also integrates with Microsoft 365 Copilot, allowing teams to access Adobe’s agentic capabilities directly within Microsoft apps.
The company has also partnered with tech firms like AWS, Microsoft, SAP, and ServiceNow, allowing its agents to work across third-party enterprise systems.
Why it matters: Adobe's latest play shows how AI agents could reshape enterprise workflows by orchestrating AI to work together on handling complex processes instead of just individual tasks. But with enterprise AI becoming increasingly crowded, Adobe will need to prove its agents can deliver more than just promises.
AI TRAINING
🔍 Expand Claude's abilities using MCPs
The Rundown: In this tutorial, you will learn how to use Claude's powerful MCP (Model Context Protocol) feature and connect the AI assistant to the Internet, giving it access to real-time information and up-to-date responses.
Step-by-step:
Download and install the latest Claude desktop application.
Visit the Brave Search API portal and register for a free API key (2,000 monthly requests free).
Locate or create your Claude configuration file and add the Brave Search MCP server settings with your API key.
Restart Claude, verify Brave Search appears in the developer settings, and test with questions about current events.
Pro tip: You can ask more specific questions to make the most of Claude's newly enhanced knowledge capabilities and get more reliable results.
PRESENTED BY UNSTRUCTURED
🎯 Improve RAG accuracy by 84%
The Rundown: Contextual chunking is the next level in document retrieval, and Unstructured is going to show you how it beats Gemini 2.0’s long context windows every time, helping you say goodbye to high error rates and hello to precise RAG apps.
Join Unstructured’s live session to learn:
How contextual chunking amplifies retrieval precision
Comparisons with Gemini 2.0’s long context window approach
Real-world strategies to reduce error rates and build stronger RAG pipelines
ANTHROPIC
🗣️ Anthropic testing new voice features
Image source: Getty Images
The Rundown: Anthropic is reportedly planning to launch a voice mode for its Claude platform, with a new report from The Financial Times also detailing a dedicated push towards enterprise and business users over consumer markets.
The details:
CPO Mike Krieger revealed the company is targeting users who "spend all day in meetings or in Excel or Google Docs" with features to streamline workflows.
A new launch in the coming months will analyze calendars and create detailed client reports from internal and external data, designed for meeting preparation.
Krieger also revealed that Anthropic already has prototypes of voice experiences ready for Claude, calling it a “useful modality to have.”
The company has reportedly explored partnerships with “a bunch of partners,” including Amazon and ElevenLabs, to accelerate the launch of voice for Claude.
Why it matters: Anthropic’s models have consistently been at the top of the industry ranks, but the startup has now set eyes on carving out a niche in the lucrative enterprise market instead of mass adoption — targeting a segment where companies are most willing to pay premium prices for productivity gains.
QUICK HITS
🛠️ Trending AI Tools
🧊 Cube 3D - Roblox’s new open-source text-to-3D object generator
🤝 Zoom AI Companion - Agentic AI for meeting productivity, and other tasks
🤖 Mistral Small 3.1 - Fast, open-source model with 128k token context
🎥 ReCamMaster - Edit camera angles and movement in existing videos
💼 AI Job Opportunities
🏗️ Mistral AI - Partner Solution Architect
💰 Coreweave - Finance Manager
🛠️ Databricks - Solutions Engineer
🗣️ Dataiku - User Communications Intern
📰 Everything else in AI today
Google introduced Canvas, a collaborative space in Gemini, for document editing and code creation, along with the addition of Audio Overviews into the Gemini platform.
Meta announced that its Llama open-source model officially reached 1B downloads this month—a significant rise from its 650M downloads in December 2024.
OpenAI’s VP of Research Liam Fedus is departing to launch an AI materials science startup, with OpenAI planning to invest in and partner with the new venture.
Google revealed TxGemma, a new collection of Gemma-based open AI models, to accelerate drug discovery—set for release later this month.
Tencent’s Hunyuan released 3D 2.0 MV and 3D 2.0Mini, two new 3D generation models for high-quality multiview shape generations.
Stability AI unveiled Stable Virtual Camera, a new diffusion model that transforms single images into 3D videos with 14 dynamic camera paths
Anthropic-backed Graphite launched Diamond, an AI code review tool that provides codebase-aware feedback and fixes, and announced a new $52M Series B round.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop today at 4 PM EST to learn how to apply AI to enhance advertising and creative production workflows — whether you’re just starting out or looking to level up content skills — with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Roblox's open-source 3D generator
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. One of the biggest gaming platforms in the world just made 3D content creation as simple as a text command — and its 85M daily users are about to build and monetize like never before.
Between Roblox’s new open-source Cube 3D model and other recent AI and coding upgrades, building the game of your dreams is suddenly just a few prompts away.
In today’s AI rundown:
Roblox’s open-source 3D generation AI
Zoom’s AI Companion goes agentic
Turn logos into ready-to-sell product mockups
Google’s AI satellite to spot wildfires early
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
ROBLOX
🧊 Roblox releases open-source 3D generation AI
Image source: Roblox
The Rundown: Roblox just announced Cube 3D, a new open-source AI system for generating 3D objects and scenes from text prompts — alongside a slew of other tools and updates for AI-assisted game development.
The details:
Cube 3D generates complete, functional 3D objects from text prompts, training on native 3D data instead of traditional image-based reconstruction.
Developers can generate assets through simple commands like "/generate motorcycle," with image input capabilities also coming in the future.
Cube uses ‘3D tokenization’ to predict and generate shapes the same way language models predict text, enabling future 4D scene generation capabilities.
Roblox also released updates to its Studio content creation suite including improved performance, real-time collaboration features, and monetization tools.
Why it matters: Between ‘vibe-coding’, Gemini’s new native multimodal image capabilities, and open-source tools like Cube 3D, it has never been easier to take a game from idea to reality. With 85M+ daily active users, these AI tools will supercharge both Roblox’s growth and the ability for users to build and monetize on the platform.
TOGETHER WITH DEEPGRAM
🗣️ Speech-to-text API designed for healthcare
The Rundown: Deepgram's Nova-3 Medical delivers unmatched accuracy for clinical environments — helping transform your healthcare applications with transcriptions that get medical terminology right the first time, every time.
With Nova-3 Medical, you can:
Capture medical terminology with 63.7% higher accuracy than alternatives
Customize and fine-tune Nova-3 for medical specialties with Keyterm Prompting
Stay secure and compliant with HIPAA-approved, flexible deployment options
Every word matters in healthcare. Start building with Nova-3 Medical today.
ZOOM
🤖 Zoom’s AI Companion goes agentic
Image source: Zoom
The Rundown: Zoom announced a major upgrade to its AI Companion, introducing new agentic capabilities and skills that allow the assistant to identify and complete tasks across the platform’s ecosystem.
The details:
AI Companion is getting new memory and reasoning capabilities to problem-solve and leverage the correct agentic tools for each task.
Zoom Tasks automatically detects and executes action items from meetings, scheduling follow-ups, or generating documents without user intervention.
Other new features include calendar management, clip generation, writing assistance, voice recording transcriptions, and live notes for meetings.
A new $12/month Custom AI Companion add-on is also launching in April, offering personal AI coaches, AI avatars for video messages, and more.
Why it matters: Zoom pivoted in 2024 to be entirely ‘AI-first’, so it’s no surprise to see it now enter the agentic game with automated workflows and tasks on the platform. CEO Eric Yuan previously shared a vision of AI digital twins that can attend meetings in a user’s place, and April’s launch may be the first step towards that goal.
AI TRAINING
👕 Turn logos into ready-to-sell product mockups
The Rundown: In this tutorial, you will learn how to use Recraft's mockup tool to place your brand logos directly onto product images for instant e-commerce and presentation visuals.
Step-by-step:
Sign in to Recraft (free account includes 50 daily credits) and click anywhere on the blank canvas to begin.
Select "Product Photo" and choose a preferred product template (like a t-shirt).
Click "Import Image" to upload your logo to the canvas and click "Remove Background" to make it transparent.
Drag your logo onto the product photo and watch as it automatically adapts to the fabric's contours.
Pro tip: You can try different placements on your product to find the most appealing position before finalizing your mockup.
PRESENTED BY INNOVATING WITH AI
🤝 Turn AI passion into a consulting career
The Rundown: Innovating with AI's new program, AI Consultancy Project, transforms AI enthusiasts into professional consultants — tapping into a market projected to reach $54.7B by 2032.
The 6-month program delivers:
Proven frameworks for client acquisition and service delivery
A step-by-step path to six-figure consulting income
Students who land their first AI client in as little as 3 days
Click here to request early access to The AI Consultancy Project.
🔥 Google’s AI satellite to spot wildfires early
Image source: Google
The Rundown: Google Research and Muon Space just launched the first AI-powered FireSat satellite, designed to provide powerful tools for early wildfire detection — with the ability to spot fires as small as a classroom within minutes of ignition.
The details:
Current wildfire detection relies on infrequently updated, low-resolution imagery that often can't spot fires until they've grown to several acres in size.
Using infrared sensors and AI analysis from space, FireSat can detect fires as small as 5x5 meters — dramatically smaller than current satellite systems.
The satellite is the first of more than 50 planned for the entire FireSat constellation, which will scan nearly all of Earth's surface every 20 minutes.
When fully deployed, the system will create a global historical record of fire behavior to help scientists better understand and model wildfire patterns.
Why it matters: Wildfire seasons continue to intensify worldwide, and early detection can be the difference between a small containment and a catastrophic event. In addition to the wealth of real-time data being fed to emergency responders, FireSat’s data will also advance scientific understanding for more efficient firefighting.
QUICK HITS
🛠️ Trending AI Tools
🤖 Ernie X1 & 4.5 - Baidu’s new deep-thinking and native multimodal models
🧠 Gemini Personalization - Responses based on chats and search history
🗣️ 1B CSM - Sesame’s newly open-sourced text and audio-to-speech model
🧪 Prompt Engineering Studio - Deploy AI prompts across 1600+ models
💼 AI Job Opportunities
💼 UiPath - Commercial Account Executive
🧮 Perplexity AI - Backend Software Engineer, Billing
⚖️ Grammarly - Legal Analyst
🛩️ Shield AI - Production Supervisor, V-BAT
📰 Everything else in AI today
Mistral AI released Small 3.1, a fast, open-source multimodal model with a 128k token context window that outperforms Gemma 3 and GPT-4o Mini across key benchmarks.
xAI acquired generative video startup Hotshot, with the company sunsetting new video creation to integrate and scale up training its models on xAI’s Colossus cluster.
Chinese researchers introduced ReCamMaster, an AI system that can edit the camera angle and movement in a video while preserving details from the original scene.
MagicLab showcased its Magicbot humanoid running continuously outdoors for four minutes, with the robot preparing for an upcoming half marathon in Beijing.
Google is partnering with MediaTek for its next generation of Tensor Processing Units, aiming to reduce reliance on Broadcom while bringing more chip development in-house.
Perplexity shared a new commercial featuring Emmy-winning actor Lee Jung-jae, directly pitting the product against Google Search in a Squid Game-style scenario.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop on Wednesday at 4 PM EST to learn how to apply AI to enhance advertising and creative production workflows — whether you’re just starting out or looking to level up content skills — with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Google’s $30B cybersecurity bet
Read Online | Sign Up | Advertise
Good morning, tech enthusiasts. Google is acquiring cloud security startup Wiz for $32B—its biggest deal in history.
By bringing Wiz into the fold, Google hopes to provide cloud customers with AI-driven cybersecurity solutions to combat critical risks—and better compete with Microsoft and Amazon. Will this bold bet pay off?
In today’s tech rundown:
Google acquiring Wiz for $32B
Apple designing a foldable iPhone
BYD’s super-fast charging system
Alphabet’s Starlink competitor ‘Taara’
Quick hits on other major news
LATEST DEVELOPMENTS
🔐 Google acquiring Wiz for $32B
Image source: Ideogram/The Rundown
The Rundown: Google is acquiring Israeli cybersecurity startup Wiz for $32B in an all-cash transaction—aiming to bolster its cloud division’s cybersecurity products and take a bigger piece of the cloud security market.
The details:
Wiz specializes in AI-powered cloud security solutions that help businesses identify and mitigate critical vulnerabilities across major cloud platforms.
With this deal, Google will help strengthen the security platform offered by Wiz, although the startup will continue to operate independently across all clouds.
This also marks Google’s largest acquisition to date, surpassing its $12.5B, consumer tech-focused purchase of Motorola Mobility in 2011.
However, it has come through on the second attempt; Google’s first bid for Wiz, reportedly at $23B, collapsed last year due to regulatory concerns.
Why it matters: Google’s acquisition of Wiz underscores its urgency to bolster its cloud security footprint amid the AI boom and compete more aggressively with Microsoft and Amazon. However, the deal’s scale and Alphabet’s market dominance are likely to draw regulatory scrutiny in the coming months.
APPLE
🍎 Apple designing a foldable iPhone
Image source: Ideogram/The Rundown
The Rundown: Apple is reportedly prototyping foldable devices, with plans to release its first foldable iPhone and potentially a foldable iPad by 2026 to compete with Samsung, Google, OnePlus, and Huawei.
The details:
The foldable iPhone is expected to feature a "book-style" design with a 7.8-inch internal display and a 5.5-inch external screen.
Apple will likely use materials like titanium and stainless steel for the hinge to address common issues with foldables, such as creasing and fragility.
The device is also rumored to include high-end hardware—such as ProMotion technology—and come at a price tag well above $2,000.
And that’s not all: The company is also said to be working on a larger foldable, potentially an 18.8-inch MacBook-iPad hybrid that could run macOS.
Why it matters: The foldable rumors come as Apple faces lukewarm iPhone sales due to delays in AI-powered Siri. Analysts believe the launch of foldable devices could rejuvenate the company’s market presence as it would refine a design that has already been tried and tested by competitors.
BYD
🚘 BYD’s super-fast charging system
Image source: Ideogram/The Rundown
The Rundown: Chinese electric vehicle giant BYD unveiled a groundbreaking fast-charging system that it says charges electric vehicles in five minutes—about the same as it takes to fill up a gas tank.
The details:
The new "Super e-Platform" boasts peak charging speeds of 1,000 kilowatts (kW), allowing vehicles to gain 400 km (249 miles) of range in just five minutes.
This charging rate is twice as fast as Tesla's latest Superchargers, addressing one of the biggest barriers to EV adoption — charging time.
The ultra-fast charging tech includes batteries with 10C charging capability, high-power motors, and silicon carbide power chips.
The company’s Han L sedan and Tang L SUV will be the first models to feature this new technology, with sales starting in April 2025.
Why it matters: BYD’s breakthrough matches the convenience of refueling gasoline vehicles and outpaces competitors like Tesla, effectively eliminating range anxiety and enhancing the practicality of EVs for long-distance travel. The company also plans to build 4,000 ultra-fast charging stations across China, just like Tesla’s Superchargers.
TAARA
⚡️Alphabet’s Starlink competitor ‘Taara’
Image source: Taara/X
The Rundown: Alphabet is spinning off its laser-based internet project, Taara, from its X moonshot incubator — positioning it as a competitor to SpaceX's Starlink by delivering high-speed internet via laser beams instead of satellites.
The details:
Taara uses laser beams to transmit data between fixed points at speeds of up to 20 gigabits per second over distances of 20 kilometers (about 12.5 miles).
The approach aims to provide high-speed internet access to hard-to-reach areas where fiber optic infrastructure is impractical or too expensive to install.
The tech differs from Starlink, which relies on a satellite constellation in low Earth orbit, along with radio frequencies, to connect terminals to satellites.
Taara is now developing a next-gen chip, expected to launch in 2026, with thousands of miniature light emitters to significantly boost bandwidth.
Why it matters: By spinning off from Alphabet, Taara is expected to attract investors to help scale its operations, allowing it to compete with Starlink for internet coverage in rural or remote areas. So far, Taara has provided internet services in more than a dozen countries, including Kenya and India, and has been used at events like Coachella.
QUICK HITS
📰 Everything else in tech today
Waymo received a 30-day permit to begin mapping roadways at San Francisco International Airport, moving toward its goal of offering robotaxis in the Bay Area.
Intel's new CEO Lip-Bu Tan plans to make broad changes to its chip manufacturing methods and AI strategies in a move to revive the struggling tech giant.
Elon Musk’s xAI acquired AI video startup Hotshot to further train their advanced video generation models and integrate them into Grok.
MIT researchers found that climate change will make space junk—mostly old satellites—pile up to interfere with internet services and national defense.
Food e-commerce startup GrubMarket raised $50M in new equity at a $3.5B valuation as part of its goal to build AI to transform the U.S. food supply chain industry.
Amazon Echo users will soon lose the option to process their Alexa voice recordings locally—meaning those recordings will be sent to the company’s cloud.
Dating app Bumble is adding new safety features, including an ID verification tool that lets users submit a government-issued ID to the system to prove their credibility.
More than 400 Hollywood creatives signed an open letter urging the Trump administration to not relax copyright restrictions as requested by OpenAI and Google.
France's leading publishers and writers filed a lawsuit against Meta for allegedly using “copyright-protected content on a massive scale” to train its AI systems.
Klarna, a Swedish fintech company, filed for an IPO in the U.S., aiming to raise at least $1B at a valuation above $15B.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop this Wednesday at 4 PM EST to learn how you can apply AI to enhance your advertising and creative production workflows—whether you’re just starting out or looking to level up your content skills—with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
See you soon,
Rowan, Jennifer, and Joey—The Rundown’s editorial team
Roblox's open-source 3D generator
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. One of the biggest gaming platforms in the world just made 3D content creation as simple as a text command — and its 85M daily users are about to build and monetize like never before.
Between Roblox’s new open-source Cube 3D model and other recent AI and coding upgrades, building the game of your dreams is suddenly just a few prompts away.
In today’s AI rundown:
Roblox’s open-source 3D generation AI
Zoom’s AI Companion goes agentic
Turn logos into ready-to-sell product mockups
Google’s AI satellite to spot wildfires early
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
ROBLOX
🧊 Roblox releases open-source 3D generation AI
Image source: Roblox
The Rundown: Roblox just announced Cube 3D, a new open-source AI system for generating 3D objects and scenes from text prompts — alongside a slew of other tools and updates for AI-assisted game development.
The details:
Cube 3D generates complete, functional 3D objects from text prompts, training on native 3D data instead of traditional image-based reconstruction.
Developers can generate assets through simple commands like "/generate motorcycle," with image input capabilities also coming in the future.
Cube uses ‘3D tokenization’ to predict and generate shapes the same way language models predict text, enabling future 4D scene generation capabilities.
Roblox also released updates to its Studio content creation suite including improved performance, real-time collaboration features, and monetization tools.
Why it matters: Between ‘vibe-coding’, Gemini’s new native multimodal image capabilities, and open-source tools like Cube 3D, it has never been easier to take a game from idea to reality. With 85M+ daily active users, these AI tools will supercharge both Roblox’s growth and the ability for users to build and monetize on the platform.
TOGETHER WITH DEEPGRAM
🗣️ Speech-to-text API designed for healthcare
The Rundown: Deepgram's Nova-3 Medical delivers unmatched accuracy for clinical environments — helping transform your healthcare applications with transcriptions that get medical terminology right the first time, every time.
With Nova-3 Medical, you can:
Capture medical terminology with 63.7% higher accuracy than alternatives
Customize and fine-tune Nova-3 for medical specialties with Keyterm Prompting
Stay secure and compliant with HIPAA-approved, flexible deployment options
Every word matters in healthcare. Start building with Nova-3 Medical today.
ZOOM
🤖 Zoom’s AI Companion goes agentic
Image source: Zoom
The Rundown: Zoom announced a major upgrade to its AI Companion, introducing new agentic capabilities and skills that allow the assistant to identify and complete tasks across the platform’s ecosystem.
The details:
AI Companion is getting new memory and reasoning capabilities to problem-solve and leverage the correct agentic tools for each task.
Zoom Tasks automatically detects and executes action items from meetings, scheduling follow-ups, or generating documents without user intervention.
Other new features include calendar management, clip generation, writing assistance, voice recording transcriptions, and live notes for meetings.
A new $12/month Custom AI Companion add-on is also launching in April, offering personal AI coaches, AI avatars for video messages, and more.
Why it matters: Zoom pivoted in 2024 to be entirely ‘AI-first’, so it’s no surprise to see it now enter the agentic game with automated workflows and tasks on the platform. CEO Eric Yuan previously shared a vision of AI digital twins that can attend meetings in a user’s place, and April’s launch may be the first step towards that goal.
AI TRAINING
👕 Turn logos into ready-to-sell product mockups
The Rundown: In this tutorial, you will learn how to use Recraft's mockup tool to place your brand logos directly onto product images for instant e-commerce and presentation visuals.
Step-by-step:
Sign in to Recraft (free account includes 50 daily credits) and click anywhere on the blank canvas to begin.
Select "Product Photo" and choose a preferred product template (like a t-shirt).
Click "Import Image" to upload your logo to the canvas and click "Remove Background" to make it transparent.
Drag your logo onto the product photo and watch as it automatically adapts to the fabric's contours.
Pro tip: You can try different placements on your product to find the most appealing position before finalizing your mockup.
PRESENTED BY INNOVATING WITH AI
🤝 Turn AI passion into a consulting career
The Rundown: Innovating with AI's new program, AI Consultancy Project, transforms AI enthusiasts into professional consultants — tapping into a market projected to reach $54.7B by 2032.
The 6-month program delivers:
Proven frameworks for client acquisition and service delivery
A step-by-step path to six-figure consulting income
Students who land their first AI client in as little as 3 days
Click here to request early access to The AI Consultancy Project.
🔥 Google’s AI satellite to spot wildfires early
Image source: Google
The Rundown: Google Research and Muon Space just launched the first AI-powered FireSat satellite, designed to provide powerful tools for early wildfire detection — with the ability to spot fires as small as a classroom within minutes of ignition.
The details:
Current wildfire detection relies on infrequently updated, low-resolution imagery that often can't spot fires until they've grown to several acres in size.
Using infrared sensors and AI analysis from space, FireSat can detect fires as small as 5x5 meters — dramatically smaller than current satellite systems.
The satellite is the first of more than 50 planned for the entire FireSat constellation, which will scan nearly all of Earth's surface every 20 minutes.
When fully deployed, the system will create a global historical record of fire behavior to help scientists better understand and model wildfire patterns.
Why it matters: Wildfire seasons continue to intensify worldwide, and early detection can be the difference between a small containment and a catastrophic event. In addition to the wealth of real-time data being fed to emergency responders, FireSat’s data will also advance scientific understanding for more efficient firefighting.
QUICK HITS
🛠️ Trending AI Tools
🤖 Ernie X1 & 4.5 - Baidu’s new deep-thinking and native multimodal models
🧠 Gemini Personalization - Responses based on chats and search history
🗣️ 1B CSM - Sesame’s newly open-sourced text and audio-to-speech model
🧪 Prompt Engineering Studio - Deploy AI prompts across 1600+ models
💼 AI Job Opportunities
💼 UiPath - Commercial Account Executive
🧮 Perplexity AI - Backend Software Engineer, Billing
⚖️ Grammarly - Legal Analyst
🛩️ Shield AI - Production Supervisor, V-BAT
📰 Everything else in AI today
Mistral AI released Small 3.1, a fast, open-source multimodal model with a 128k token context window that outperforms Gemma 3 and GPT-4o Mini across key benchmarks.
xAI acquired generative video startup Hotshot, with the company sunsetting new video creation to integrate and scale up training its models on xAI’s Colossus cluster.
Chinese researchers introduced ReCamMaster, an AI system that can edit the camera angle and movement in a video while preserving details from the original scene.
MagicLab showcased its Magicbot humanoid running continuously outdoors for four minutes, with the robot preparing for an upcoming half marathon in Beijing.
Google is partnering with MediaTek for its next generation of Tensor Processing Units, aiming to reduce reliance on Broadcom while bringing more chip development in-house.
Perplexity shared a new commercial featuring Emmy-winning actor Lee Jung-jae, directly pitting the product against Google Search in a Squid Game-style scenario.
COMMUNITY
🎥 Join our next live workshop
Join our next workshop on Wednesday at 4 PM EST to learn how to apply AI to enhance advertising and creative production workflows — whether you’re just starting out or looking to level up content skills — with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Tesla Optimus is heading to Mars
Read Online | Sign Up | Advertise
Good morning, robotics enthusiasts. Elon Musk just announced that SpaceX will launch Tesla’s Optimus robot to Mars aboard Starship in 2026—paving the way for human landings a few years later.
As SpaceX reportedly develops domes and spacesuits for a future Martian colony, Optimus might get a head start on setting up camp before humans arrive. Wildly ambitious—or is life on Mars closer than we think?
In today’s robotics rundown:
Tesla Optimus heading to Mars
Figure’s high-volume humanoid factory
The world’s strongest robotic arm
China’s grand humanoid marathon
Quick hits on other robotics news
LATEST DEVELOPMENTS
SPACEX
🚀 Tesla Optimus heading to Mars
Image source: Tesla
The Rundown: Elon Musk just announced on X an ambitious plan to send Tesla's Optimus humanoid to Mars by the end of 2026—with the mission slated to be carried out by SpaceX’s massive Starship rocket.
The details:
Optimus will be key to Musk’s Mars colonization vision, as it could capture insights from the planet while serving as a test-bed for future missions.
After sending Optimus, Musk anticipates that human landings on Mars could commence as early as 2029, although he considers 2031 a more likely timeline.
The robot will head to Mars on a full-assembled Starship, which stands nearly 124 meters tall as the largest launch vehicle ever built.
It is designed to be fully reusable, which is essential for long-term space missions aimed at transporting humans and cargo to Mars and back.
Why it matters: Musk has said that his goal is to make it so that anyone can travel to Mars, if they would like, and eventually set up a self-sustaining civilization on the Red Planet. But the road is bumpy: SpaceX must first perfect its launch vehicle to ensure it doesn’t explode, as seen in test flights, before carrying Optimus or human passengers.
FIGURE
🔥 Figure’s high-volume humanoid factory
Image source: Figure AI
The Rundown: California-based robotics major Figure AI unveiled BotQ, its new manufacturing facility dedicated to producing 12,000 humanoids annually, with plans to scale up to 100,000 in the next four years.
The details:
Figure’s engineering teams spent the last eight months designing BotQ with automation and high-volume manufacturing techniques.
The company said it chose to bring the manufacturing of its humanoids in-house to ensure full control over the build process, quality, and efficiency.
It switched from CNC machining to injection molding and diecasting—ensuring parts that once took a week to make now get prepped in 20 seconds.
With this facility, Figure plans to speed up automation by using Figure 02 humanoids to build the next-gen Figure 03 robots.
Why it matters: Figure is employing several strategies—from software infrastructure to partnerships—to create a scalable manufacturing process that could give it a competitive edge. However, rivals are also moving fast—Agility Robotics produces 10,000 Digit humanoids annually, with several already deployed in Amazon facilities.
RISE ROBOTICS
💪🏽 The world’s strongest robotic arm
Image source: Rise Robotics
The Rundown: Rise Robotics’ Superjammer industrial robotic arm is set to claim a spot in the Guinness Book of World Records as the world’s strongest non-hydraulic robot arm—lifting an astonishing 6,460 lb. (2,930 kg) some 5 meters off the ground.
The details:
Unlike traditional hydraulic systems, the Superjammer uses Rise's proprietary Beltdraulic technology—a belt-and-pulley system powered by electric motors.
This innovative approach eliminates the need for energy-intensive hydraulic fluids, reducing emissions and energy consumption by 65% to 90%.
The system is said to be 3x faster than hydraulics, offering precise control, minimal maintenance, and unmatched durability even in extreme conditions.
The current world record is held by the Fanuc M-2000iA/2300, which hoisted approx. 2,300 kg using electronic gears, belts, and servos.
Why it matters: The Superjammer will showcase its power on March 20 at the Guinness World Records ceremony, marking a breakthrough in heavy machinery and robotics. The game-changer? It’s emissions-free and operates without petroleum-based hydraulic fluids—yet still outperforms conventional hydraulic systems.
BEIJING ROBOT HALF MARATHON
🏃🏻 China’s grand humanoid marathon
Image source: Unitree
The Rundown: China is gearing up to host the world's first-ever robot half-marathon in Beijing on April 13, where humanoids from 20 international companies will compete alongside 12,000 human runners.
The details:
The event is designed to showcase robots’ speed, agility, and performance with the 21 km distance serving as a stress test for robotic hardware.
Race rules are that humanoids must be capable of bipedal walking or running and cannot use wheels or other non-human mobility aids.
The robots can be either semi-autonomous or fully autonomous and will compete on separate lanes secured with barriers to ensure safety.
They can replace batteries or even switch out entirely in a relay format during the race; however, each robot replacement will incur a 10-minute penalty.
Why it matters: While humanoids aren’t typically designed for long-distance running, robots can nab prizes for best gait, most innovative design, and creativity. This event provides a way for companies like Unitree, Star 1, and Tiangong to push the limits of what their humanoids can do on a global stage, with future investors likely looking on.
QUICK HITS
📰 Everything else in robotics today
eVTOL company Joby Aviation is partnering with Virgin Atlantic to launch electric air taxis in the UK, offering short-haul flights for Virgin customers to and from the airport.
Shenzhen-based UBTECH launched its Tiangong Walker humanoid for research and education at RMB 299,000 ($42,310), with deliveries starting in the coming months.
Japanese researchers are investigating transporting AI-equipped swarm robots to the Moon to explore lava tubes by 2028.
French automaker Renault is testing a new driverless mini-bus in Barcelona after teaming up with autonomous vehicle company WeRide to make the prototype.
U.S. researchers are exploring how AI and a robot called Ruyi can improve caregiving and support for older adults with early-stage Alzheimer’s disease and dementia.
Japanese researchers developed a self-contained, wireless microbot dubbed HB-3 that operates with “incredible precision” in tight, dangerous spaces.
Micropolis, a Dubai-based robotics company, raised $15.5M through its NYSE IPO for building autonomous mobile robots, including delivery and police patrol bots.
Polish researchers published a study exploring how humanoid robot supervisors are perceived as authority figures by human staff in the workforce.
Researchers in Munich created a new encyclopedia called the “Tree of Robots” designed to make learning about robotic systems and comparing them easier.
COMMUNITY
🎥 Join our next live workshop
Join our next private workshop this Wednesday at 4 PM EST to learn about how you can apply AI to enhance your advertising and creative production workflows, with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
See you soon,
Rowan, Jennifer, and Joey—The Rundown’s editorial team
China releases GPT-4.5 rival at 1% the cost
Read Online | Sign Up | Advertise
Good morning, AI enthusiasts. China’s AI acceleration is back, with tech giant Baidu launching two powerful AI models at just 1% the price of OpenAI’s GPT-4.5, and half of the price of DeepSeek’s R1.
With impressive performance claims and radically lower costs, we’re likely witnessing the beginning of a global AI price war.
In today’s AI rundown:
China’s Baidu launches ultra-cheap AI models
Judge rejects Musk's bid to block OpenAI's evolution
How to code with AI, directly in your editor
Harvard team creates AI agent for personalized medicine
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
BAIDU
🇨🇳 China’s Baidu launches ultra-cheap AI models
Image source: Baidu
The Rundown: Chinese tech giant Baidu just released two aggressively priced multimodal AI models — ERNIE 4.5, a significant upgrade to their foundational model, and a new deep-think-capable model called Ernie X1.
The details:
ERNIE 4.5 stands out with enhanced “EQ” and language skills as well as improved hallucination prevention, logical reasoning, and coding abilities.
According to Baidu, ERNIE 4.5 outperforms GPT-4.5 across multiple benchmarks while costing just 1% of its price — approximately $0.55 and $2.20 per million input and output tokens.
ERNIE X1, Baidu’s first reasoning model, matches the capabilities of top Chinese competitor DeepSeek's R1 at half the price.
Like DeepSeek’s R1, ERNIE X1 uses a step-by-step "thinking" approach, excelling in complex calculations and tasks like document understanding.
Why it matters: China continues to build intelligence too cheap to meter. With ERNIE 4.5 at 1% the cost of GPT 4.5, and ERNIE X1 matching DeepSeek's R1 at half the price, we're likely witnessing the start of a global AI price war. This strategy could force Western competitors to slash rates, democratizing access to advanced AI worldwide.
TOGETHER WITH HUBSPOT
⛏️ The AI gold rush: Are you in or out?
The Rundown: HubSpot’s ‘AI For Business Builders’ free guide shows you how to join the savvy entrepreneurs quietly leveraging artificial intelligence for real business growth — with no P.H.D. needed.
This no-BS guide cuts through the hype, giving you:
A jargon-free breakdown of large language models (the engine behind AI's magic)
Concrete steps to leverage AI for immediate business growth
Real-world examples that turn "prompt engineering" from gibberish into a secret weapon
OPENAI
🧑⚖️ Judge rejects Musk's bid to block OpenAI's evolution
Image source: Getty Images / Reuters
The Rundown: A federal judge has denied Elon Musk’s request for a preliminary injunction against OpenAI's structural changes, while fast-tracking the trial and dismissing several of Musk's claims entirely, according to a March 4, 2025 ruling.
The details:
The judge said Musk didn’t have "the high burden” for a preliminary injunction but allowed other aspects of the lawsuit with an expedited trial this fall.
Internal emails cited by OpenAI allegedly show Musk once wanted to merge OpenAI into Tesla as a for-profit entity—contradicting his current legal position.
Musk sued OpenAI and its CEO, Sam Altman, last year over straying from their original mission of developing AI for the good of humanity, not corporate profits.
OpenAI and Altman have denied the accusations, saying the non-profit is not going—and any restructuring of for-profit subsidiaries will only better support it.
Why it matters: With OpenAI's rumored ~$40B SoftBank investment hanging on its pivot to a for-profit model, this lawsuit could reshape the company's trajectory and its AI models' evolution. Having last raised $6.6B at a $157B valuation, the stakes for both the company and the broader AI landscape could not be higher.
AI TRAINING
💻 How to code with AI, directly in your editor
The Rundown: In this tutorial, you will learn to use ChatGPT's updated macOS app and its "Work with Apps" feature to edit code directly in your preferred development environment.
Step-by-step:
Install the proper extension in your code editor—search for "openai.chatgpt" in the Extensions marketplace (VS Code/Cursor) or sideload the VSIX file for other editors.
Open the ChatGPT app, find the "Work With" panel in the sidebar, and select your code editor from the list.
Open any code file in your editor, and ChatGPT will automatically connect to it.
Ask ChatGPT to explain sections or modify your code by typing natural language requests, review it, and click "Apply" to instantly update your file.
Pro tip: Different ChatGPT models offer varying levels of code expertise - choose based on whether you need quick edits or complex refactoring.
PRESENTED BY FIDDLER
🤖 Making AI agents transparent and trustworthy
The Rundown: Fiddler AI and AWS’s upcoming webinar breaks down how to increase visibility and security for AI agent workflows — helping you drive innovation without sacrificing compliance or trust.
In this live webinar, you'll discover:
How AI observability boosts transparency in agent decision-making processes
Best practices to protect agents from risks using tools like Guardrails
Essential tips for balancing innovation, compliance, and operational security
Register today to attend live or receive the recording afterward.
AI RESEARCH
🧪 Harvard team creates AI agent for personalized medicine
Image source: TxAgent
The Rundown: Researchers from Harvard and MIT just introduced TxAgent, an AI agent that leverages multi-step reasoning and real-time biomedical knowledge retrieval to generate trusted, personalized treatment recommendations for patients.
The details:
TxAgent utilizes 211 specialized tools to analyze drug interactions and contraindications and suggest patient-specific treatments in real time.
It evaluates drugs at molecular, pharmacokinetic, and clinical levels, identifying risks based on comorbidities, ongoing medications, age, and genetic factors.
The system synthesizes evidence from biomedical sources, iteratively reasoning and refining recommendations through structured function calls.
TxAgent's toolkit is available as "ToolUniverse" featuring trusted data sources like openFDA and Open Targets to ensure validated medical insights.
Why it matters: TxAgent signals a major step in personalized medicine, offering doctors an AI-powered assistant for delivering safer, more tailored treatments. With optimized suggestions, this system can also fast-track medical consultations, especially in developing regions where healthcare systems are already strained.
QUICK HITS
🛠️ Trending AI Tools
🚀 OLMo 2 32B - First fully open model that beats GPT-3.5 and GPT-4o mini
🎥 Freepik AI Video Upscaler - Upscale low-quality videos to 4K in one click
🧠 Command A - Cohere’s ultra-fast enterprise AI with 256K context window
💻 BrowserAgent - Create and run AI workflows in browser with no API costs
💼 AI Job Opportunities
✍️ The Rundown - Course Copywriter/Designer
📈 Findem - PLG Senior Product Marketing Manager
🎨 Sanctuary AI - Technical Artist, Simulation
⚡ C3 AI - AI Engagement Director (Energy)
📰 Everything else in AI today
Google’s new Gemini 2.0 Flash model is reportedly being used by people to remove watermarks from images, including those from platforms like Getty.
Humanoid robot maker Figure announced the launch of BotQ, a new manufacturing facility capable of producing 12,000 humanoids a year—with plans to scale up to 100,000.
Patronus AI launched the industry’s first multimodal LLM-as-a-judge, a tool designed to help developers detect and mitigate reliability issues in multimodal AI apps.
Pika Labs released 16 new effects for its AI video platform, enabling users to morph images into different character videos.
Sesame, which recently went viral for its realistic AI voice tech demo, open-sourced its Conversation Speech Model (CSM-1B) for text-to-speech tasks.
Vogent AI launched voice agents that design and improve themselves, learning from real failure cases and requiring no prompt engineering.
Y Combinator CEO reported that nearly a quarter of their current startups now have 95% of their code written by AI, enabling small teams to reach $10M in revenue with dramatically fewer engineers.
OpenAI's CPO Kevin Weil just predicted 2025 is the year AI will permanently surpass humans at programming, calling it a "democratizing effect" that will allow anyone to create software.
COMMUNITY
🎥 Join our next live workshop
Join our next private workshop this Wednesday at 4 PM EST to learn about how you can apply AI to enhance your advertising and creative production workflows, with Tony Pu, Product Marketing Lead at Kling AI.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
🤝 Share The Rundown, get rewards
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
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.