Good morning, AI enthusiasts. OpenAI's market dominance is set to receive a $40B stamp of approval — a record-breaking funding round that would value the ChatGPT maker at a whopping $300B.

With the AI leader currently burning billions but projecting a massive $125B in revenues by 2029, are investors making the safest bet in tech history…Or the riskiest?

P.S. Our next workshop is today at 4 PM EST — join and learn how to use the latest AI tools to take your content creation to the next level!

In today’s AI rundown:

• OpenAI nears $40B funding round

• Anthropic reveals how Claude ‘thinks’

• Add Deep Research to your AI code editors

• Qwen’s QVQ-Max visual reasoning model

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

OPENAI

💰 OpenAI nears $40B funding round

Image source: Grok / The Rundown

The Rundown: OpenAI is reportedly finalizing a massive $40B funding round led by SoftBank, which would make it the largest private funding in history — and nearly double the ChatGPT maker's valuation to $300B.

The details:

• SoftBank will invest an initial $7.5B, followed by another $22.5B later this year with other investors including Magnetar Capital, Coatue, and Founders Fund.

• OpenAI expects its revenue to triple to $12.7B in 2025 and become cash-flow positive by 2029 with over $125B in projected revenue.

• The company reportedly lost as much as $5B on $3.7B of revenue in 2024, attributed to AI infrastructure and training costs.

• The funding will also partially support OpenAI's commitment to Stargate, the $300B AI infrastructure JV announced with SoftBank and Oracle in January.

Why it matters: OpenAI’s for-profit turn is looking to be a record-breaking one, and both company projections and investor wallets are signaling that the AI boom is not slowing down any time soon. The competition may be getting tougher — but as we’ve seen this week, no company dominates mindshare in the AI world quite like OpenAI.

TOGETHER WITH TURING

🤝 Network with AI’s best at Google Cloud Next

The Rundown: Heading to Google Cloud Next in Las Vegas? Turing and Fireworks AI invite you to an exclusive evening reception on April 10  — where you can connect with fellow leaders in the AI community and unwind after a day packed with insights.

Register today to:

• Network with fellow Google Cloud Next attendees

• Connect with the top leaders in AI / LLMs

• Enjoy fantastic food and creative cocktails

Click here to secure your spot.

ANTHROPIC

🔬 Anthropic reveals how Claude ‘thinks’

Image source: Anthropic

The Rundown: Anthropic released two research papers that reveal how its AI assistant Claude processes information, helping to better understand internal mechanisms that explain capabilities like multilingual reasoning and advanced planning.

The details:

• The researchers developed an "AI microscope" that reveals internal “circuits” in the model, showing how Claude transforms input to output in key tasks.

• Claude uses a universal "language of thought" across different languages, with shared conceptual processing for English, French, and Chinese.

• When writing poetry, Claude plans ahead several words, identifying rhyming options before constructing lines to reach those planned words.

• The team also discovered a default that prevents speculation unless overridden by strong confidence, helping explain how hallucination prevention works.

Why it matters: The closer we get to superintelligent AI, the more important understanding how models process internally becomes. With research already detailing AI’s deceptive qualities and more powerful systems being integrated into life across the globe, cracking the inner workings becomes more crucial by the day.

AI TRAINING

🔍 Add Deep Research to your AI code editors

The Rundown: In this tutorial, you will learn how to connect Firecrawl's Deep Research to your AI code editors, like Cursor and Windsurf, to access real-time web information directly in your coding environment.

Step-by-step:

1. Visit Firecrawl, create an account, and generate your free API key from the Dashboard.

2. For Windsurf: Click "Configure MCP," paste the provided JSON code, replace the API key placeholder with your key, and refresh the server list.

3. For Cursor: Go to Settings > MCP > Add New MCP Server, name it "firecrawl-mcp," select "command", and enter it with your API key.

4. Start using Deep Research by typing queries like "Deep Research the latest advancements in React state management" in your editor's chat.

Pro tip: You can find the JSON code (for Windsurf) and the command (for Cursor) here. Also, be specific with your research queries for more focused and relevant results.

PRESENTED BY SAMBANOVA

🧠 Deep Research, powered by the fastest Deepseek R1 671B

The Rundown: SambaNova just released a new Deep Research AI Agent that produces detailed reports and analysis in seconds, not minutes or hours — allowing you to run complex research at a fraction of the time and cost.

The details:

• Research tasks take just 5-30 seconds to complete, so you can quickly iterate on the final report

• Fully open source – connect your own data sources

• Integrates with SambaNova Cloud, which delivers lightning-fast inference on the best open source models, including DeepSeek R1 671B at over 255 tokens/second

Sign up to try Fast Deep Research for yourself!

ALIBABA

👁️ Qwen’s QVQ-Max visual reasoning model

Image source: Alibaba

The Rundown: Alibaba's Qwen team released QVQ-Max, a new visual reasoning model that goes beyond basic image recognition to analyze and reason about visual information across images and videos.

The details:

• The model is an evolution of QVQ-72B-Preview, expanding capabilities across mathematical problem-solving, code generation, and creative tasks.

• QVQ-Max features a "thinking” mechanism that can be adjusted in length to improve accuracy, showing scalable gains as thinking time increases.

• Other complex visual capabilities shown include analyzing blueprints, solving geometry problems, and providing feedback on user-submitted sketches.

• Qwen said that future plans include creating a complete visual agent capable of operating devices and playing games.

Why it matters: This is Qwen’s third model release this week! Between Omni, Qwen2.5-VL, and now QVQ-Max, the Chinese powerhouse continues to crank out capable models across the AI spectrum. With China flooding the market with advanced systems, the gap between the U.S. and China has never been smaller.

QUICK HITS

🛠️ Trending AI Tools

• ⚡Kilo Code for VS Code - AI agent extension that generates code, automates tasks, and provides suggestions. Get $15 Claude 3.7 tokens now.*

• 🎆 Ideogram 3.0 - New text-to-image AI with text & graphic design capabilities

• 🤖 Qwen2.5-Omni-7B - Alibaba’s multimodal AI for consumer hardware

• 📦 Amazon Interests - Shop and discover new products with natural language

* Sponsored Listing

💼 AI Job Opportunities

• 🧠 Mistral AI - AI Solution Architect, Deployment

• 📅 Databricks - Executive Assistant

• ⚙️ OpenAI - ML Infrastructure Engineer

• 📱 Meta - Social Media Manager, Technology

📰 Everything else in AI today

OpenAI released an updated version of GPT-4o to paid users, featuring better prompt adherence, improved coding and creativity, and more “freedom.”

Butterfly Effect, the Chinese startup behind the Manus AI agent, is seeking new funding at a $500M valuation as it faces massive cash burn from Claude API costs.

OpenAI is delaying the rollout of its 4o image generation to free users and imposing rate limits, with CEO Sam Altman saying the demand is “melting” the company’s GPUs.

AI infrastructure giant CoreWeave reduced its IPO target from $4B to $1.5B ahead of its Nasdaq debut on Friday, with Nvidia stepping in as an anchor investor.

Archetype AI introduced “Lenses” — a new category of physical AI applications for its Newton model that transform sensor data into actionable insights.

PwC unveiled agent OS, a platform to integrate multi-platform AI agents into enterprise workflows up to 10x faster than traditional methods.

Lockheed Martin is partnering with Google Public Sector to integrate genAI into its AI Factory ecosystem, aiming to enhance national security applications.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop today at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. You'll learn how to use the latest AI tools to take your content creation to the next level.

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

Good morning, robotics enthusiasts. Chinese robotics startup Agibot is ramping up its humanoid production to 5,000 units this year—on par with Tesla’s Optimus targets.

As superpowers China and the U.S. compete in the humanoid race, Chinese firms like Agibot are seen as leaders in agile robots and hardware, while the U.S. dominates in AI—the 'brain' of humanoids. But is that gap closing fast?

In today’s robotics rundown:

• Agibot’s army of robots rivaling Tesla

• Tiny liquid robots to target cancer cells

• UC students 3D-print a $20 walking bot

• Tesla to launch first robotaxis in Austin

• Quick hits on other robotics news

LATEST DEVELOPMENTS

AGIBOT

⚡️Agibot’s army of robots rivaling Tesla

Image source: Agibot

The Rundown: Shanghai-based robotics startup Agibot just announced a plan to scale its production capabilities and deliver up to 5,000 humanoids this year—matching the targets set by Elon Musk for Tesla’s Optimus humanoid.

The details:

• Agibot is investing in cutting-edge factories equipped to handle large-scale production of Lingxi X2, its latest humanoid.

• The company aims to deliver between 3,000 and 5,000 robots in 2025, up from fewer than 1,000 units in 2024.

• This month, it also launched an AI model—Genie Operator-1—to enhance X2’s ability to interpret human actions and perform real-world tasks.

• The general-purpose humanoid debuted in March and comes with the ability to walk, run, ride a bike, and interpret emotions.

Why it matters: Agibot, co-founded by Peng Zhihui—a former participant in Huawei's 'Genius Youth' program—is a key player in China's push to become a robotics powerhouse, alongside XPeng, UBTech, and Unitree. It remains to be seen just how quickly they can match the AI standards set by U.S. rivals like Tesla and Figure.

BIOTECH ROBOTICS

🔬 Tiny liquid robots target cancer cells

Image source: Science Advances, Vol. 11, No. 12, Jeon et al.

The Rundown: Inspired by the dynamic behavior of biological cells, South Korean researchers have designed a tiny, gummy-like liquid robot that can help target cancer cells, among many other use cases, in the future.

The details:

• The robot, PB, is made of water droplets coated with densely packed Teflon particles, creating a hydrophobic shell that maintains the liquid's shape.

• Each bot is controlled using sound and can shape-shift, pick up debris, move across water, and merge with another PB—without losing structural integrity.

• They behave the way pretty much like cells, imitating biological forms and functions, with several potential use cases, starting with biomedicine.

• For instance, in the future, the rice grain-sized robots can help with drug delivery, invasive procedures, or even target and destroy cancerous cells.

Why it matters: Though still in the research phase, these robust liquid bots hold immense potential. Beyond drug delivery and tumor targeting, their ability to navigate narrow spaces could prove invaluable for machinery maintenance, disaster response, and even environmental cleanup.

UC SAN DIEGO

🤖 UC students 3D-print a $20 walking bot

Image source: UC San Diego Department of Mechanical and Aerospace Engineering

The Rundown: Researchers at the University of California, San Diego have developed a 3D-printed, six-legged robot that can “walk” without any electronics, using only compressed gas for power.

The details:

• The robot was created using a desktop 3D printer and off-the-shelf printing material, costing only $20 to manufacture.

• The entire bot, including its artificial muscles and control systems, was printed in one go using a single soft material.

• A pneumatic oscillating circuit controls its movement, coordinating the behavior of the six legs using air pressure.

• The best part? It can traverse different surfaces, from turf to underwater, with the option to operate continuously for three days (with a constant air supply).

Why it matters: These robots have immense potential in environments where electronics may fail, such as areas with strong radiation, disaster zones, or space exploration. Researchers are also exploring ways to store compressed gas inside the bots and integrate manipulators, like grippers, to unlock new possibilities.

TESLA

 🚖 Tesla to launch first robotaxis in Austin

Image source: Tesla

The Rundown: Tesla is set to debut its autonomous rideshare service in Austin starting this June, with CEO Elon Musk describing the service as “unsupervised” and operating with “no one in the car.”

The details:

• Tesla is expected to use its internal fleet of Model 3s and Model Ys rather than the futuristic Cybercab, which won’t be in production at least until 2026.

• The robotaxis will be equipped with what Tesla says is its “unsupervised” Full Self-Driving (FSD) software, which means its Level 2 ADAS jumps to Level 4.

• Musk describes the service as unsupervised with "no one in the car," but the company is likely to employ teleoperation for remote monitoring and support.

• Tesla currently operates autonomous unsupervised FSD at its factory in Fremont, with vehicles driving themselves at low speeds to loading docks.

Why it matters: Operating unsupervised in cities is a major leap for Tesla’s technology, though the company’s messaging remains unclear on how it will unfold. One thing is certain: Austin is the new battleground for robotaxis, with both Waymo and Tesla making their moves in a matter of weeks.

QUICK HITS

📰 Everything else in robotics today

Tesla, Boston Dynamics, and Agility Robotics met with lawmakers on Capitol Hill to demonstrate products and push for new policies boosting robotics in the U.S.

U.S. robotics company Figure shared on X a video of its humanoid 02 robot performing “learned, natural walking,” a step up from its prior bent-knee gait.

China announced a new venture capital fund to boost robotics and AI in the country, expected to attract nearly 1T yuan ($137.8B) over the next 20 years.

Unitree Robotics, a Chinese robotics firm, reportedly confirmed its consistent profitability since 2020, with no plans for a new funding round.

Northeastern University engineers developed a hybrid robot that can accomplish a task that has long challenged roboticists: screwing in a lightbulb.

Chinese social media influencer Zhang Genyuan went viral by renting a Unitree Robotics humanoid for $1.4K a day to cook, clean, and go on dates with him.

Tesla is set to launch in Saudi Arabia on April 10, entering a market where electric vehicle adoption stands at just 1% of all car sales.

Robotics company Dyna Robotics, based in California, secured a $23.5M seed round as it works to make embodied AI robots affordable for smaller businesses.  

Contoro Robotics raised $12M in Series A funding to scale its AI-powered robots for unloading trailers and containers.

German manufacturer igus, known for making motion plastics and robotic components, announced that it made €1.105B ($1.19B) in sales last year.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Thursday at 3 PM EST to learn how to build AI Voice Agents using Vapi, led by Jordan Dearsley, the Founder & CEO at Vapi.

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

Good morning, AI enthusiasts. Another SOTA text-to-image model just dropped — but the only thing on everyone’s mind seems to be turning images into Ghibli-style anime.

Between Ideogram’s 3.0 launch, GPT-4o’s viral image generation capabilities, and Reve’s debut, AI creativity has gone to a brand new level this week.

In today’s AI rundown:

• Ideogram’s advanced 3.0 image model

• BMW, Alibaba bringing AI-enabled cars

• Create custom study assistants for any subject

• Alibaba’s multi-sensory AI for mobile

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

IDEOGRAM

🖼️ Ideogram’s advanced 3.0 image model

Image source: Ideogram

The Rundown: Image generation startup Ideogram just released version 3.0 of its AI model, introducing major improvements in photorealism, text rendering, and style consistency — while outperforming competitors in human evaluations.

The details:

• Ideogram 3.0 brings new text rendering and graphic design capabilities, enabling precise creation of complex layouts, logos, and typography.

• In testing, the model significantly outperformed leading text-to-image models, including Google’s Imagen 3, Flux Pro 1.1, and Recraft V3.

• A new ‘Style References’ feature allows users to upload up to three images to guide the aesthetic of generated content, alongside a library of 4.3B presets.

• The model is now available on Ideogram’s platform and iOS app, with all features accessible to free users.

Why it matters: Ideogram’s new model is very impressive, but the launch timing is unfortunate given the hype around OpenAI’s 4o image capabilities. What’s become apparent from releases from Ideogram, OpenAI, and Reve this week is that graphic design and accurate text generation are all but fully solved for this wave of AI models.

TOGETHER WITH WORKOS

🔐 Protect your AI apps from bad actors

The Rundown: WorkOS Radar is a security solution that shields your AI platform from fake signups, throwaway emails, and brute force attempts — all powered by advanced device fingerprinting and real-time detection.

With WorkOS Radar, you can:

• Rapidly detect and challenge unfamiliar and suspicious devices in real time

• Stop free-tier abuse and fraudulent behavior with advanced detection

• Customize threat responses to fit your app’s exact security needs

Start integrating Radar today.

BMW & ALIBABA

🚗 BMW, Alibaba bringing AI-enabled cars

Image source: Alibaba

The Rundown: Chinese tech giant Alibaba and automaker BMW announced a strategic alliance to develop advanced in-car AI tailored for the Chinese market, bringing cutting-edge vehicle cockpit tech to BMW models as soon as 2026.

The details:

• The partnership centers on a new in-car AI assistant powered by Alibaba's Qwen, featuring enhanced voice recognition and contextual understanding.

• The assistant will feature real-time dining, parking availability, and traffic management, using natural commands rather than touchscreen interfaces.

• BMW also plans to roll out two AI agents: Car Genius for vehicle diagnostics and Travel Companion for personalized recommendations and trip planning.

• The system will also include multimodal inputs like gesture recognition, eye tracking, and body position awareness for more intuitive driving experiences.

Why it matters: BMW has been at the forefront of AI and robotics, making it only a matter of time before advanced AI systems are integrated into new cars. While Tesla, with its internal xAI partnership, remains a strong contender, other automakers are also taking strategic steps to lead in the AI era.

AI TRAINING

📚 Create custom study assistants for any subject

The Rundown: In this tutorial, you will learn how to use Google Gemini's Gems feature to create personalized AI assistants for specific subjects, homework help, and project research — completely free of cost.

Step-by-step:

1. Visit Google Gemini, click the diamond Gem icon on the left sidebar, then select "New Gem."

2. Name your Gem specifically (e.g., "Physics Problem Solver") and write detailed instructions about how it should help with your subject.

3. Add course materials like notes, textbook chapters, or study guides to the Knowledge section.

4. Test your Gem with sample questions and refine its instructions until it responds perfectly.

Pro tip: You can create multiple Gems for different papers instead of one general helper; this keeps each assistant focused on a specific subject.

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.

ALIBABA

🎤 Alibaba’s multi-sensory AI for mobile

Image source: Alibaba

The Rundown: Alibaba released Qwen2.5-Omni-7B, a new multimodal AI capable of processing text, images, audio, and video simultaneously while being efficient enough to run directly on consumer hardware like smartphones and laptops.

The details:

• The model uses a new "Thinker-Talker" system for real-time processing across modalities (text, audio, image, video) with text and speech outputs.

• It shows strong performance in speech understanding and generation, outperforming specialized audio models in benchmark testing.

• Alibaba says Omni-7B can run efficiently on phones and laptops, enabling real-world applications like real-time audio descriptions for visually impaired users.

• It’s immediately available on Hugging Face and GitHub, with Alibaba positioning the model as the foundation for developing practical AI agents.

Why it matters: The age of do-it-all models is nearly here, with omni systems set to unlock completely new experiences and categories of applications. Intelligence that can understand and respond to the full complexity of human environments—while being open-source and easily accessible—is a powerful combination.

QUICK HITS

🛠️ Trending AI Tools

• 🎆 GPT-4o Image Generation - Create and edit photos in ChatGPT and Sora

• 🧠 Gemini 2.5 Pro - Google’s new SOTA reasoning model

• 👋 InfiniteYou - AI portrait generator with high-quality facial accuracy

• 🔎 Perplexity Answer Modes - Enhance searches on specific verticals

💼 AI Job Opportunities

• 💻 UiPath - Software Engineer

• 📊 LabelBox - Data Operations Engineer

• 💰 Runway - Staff Accountant

• 🛠️ xAI - Fiber Superintendent

📰 Everything else in AI today

OpenAI announced it will adopt Anthropic’s open-source Model Context Protocol, enabling ChatGPT and other products to integrate with external data and software.

Microsoft 365 Copilot unveiled Researcher and Analyst, two new AI agents designed to handle workplace tasks with research and data analysis directly in users’ workflows.

A federal judge rejected music publisher UMG’s request to block Anthropic from using song lyrics to train Claude, saying the claim failed to show “irreparable harm”.

xAI announced that its Grok chatbot is now integrated directly into messaging app Telegram, available to Premium users at no additional cost.

Amazon launched ‘Interests,’ a new AI-powered shopping feature that automatically scans its store to notify users about new products based on natural language prompts.

Midjourney revealed in its weekly Office Hours session that its highly-anticipated new V7 model is expected to arrive on Monday, March 31.

The U.S. government added over 50 Chinese tech entities to an export blacklist, targeting firms developing advanced AI, supercomputing and quantum tech.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop today at 3 PM EST to learn how to build AI Voice Agents using Vapi, led by Jordan Dearsley, the Founder & CEO at Vapi.

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

Good morning, AI enthusiasts. Google is back on top of the AI rankings, thanks to the release of its powerful new Gemini 2.5 Pro model — dominating benchmarks across reasoning, math, science, and coding.

But with AI continuing to evolve at lightning speed and rivals like OpenAI gearing up for their next models, how long will Google hold onto the crown?

P.S. Our next workshop is TODAY at 2 PM EST — join and learn how to build an AI-powered sales rep using Zapier Agents. RSVP here.

In today’s AI rundown:

• Google’s Gemini 2.5 Pro tops AI leaderboard

• OpenAI adds image generation to GPT-4o, Sora

• Transform concepts into production-ready pitch decks

• Apple’s billion-dollar bet on Nvidia AI hardware

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

GOOGLE

🏆 Google’s Gemini 2.5 Pro tops AI leaderboard

Image source: Google

The Rundown: Google just announced Gemini 2.5, a new family of AI models with built-in reasoning—starting with the release of Gemini 2.5 Pro Experimental, which tops key benchmarks and represents the company’s most intelligent model to date.

The details:

• 2.5 Pro debuts at #1 on the LMArena leaderboard, showcasing advanced, SOTA reasoning capabilities across math, science, and coding tasks.

• On coding, 2.5 Pro scores 63.8% on SWE-Bench Verified and 68.6% on Aider Polyglot — with specific strengths in web apps and agentic code applications.

• It’s shipping with a 1M token context window, but Google soon plans to double this to 2M for processing entire code repositories and massive datasets.

• The model is available now in Google AI Studio and the Gemini app for Advanced subscribers, with API pricing coming in the weeks ahead.

Why it matters: As major AI labs push forward with reasoning, Google has made "thinking" a standard rather than a premium offering. The tech giant continues to push SOTA models despite lacking the hype of OpenAI — but with how fast AI is moving (and with GPT-5 and others lurking), it remains to be seen how long the new ranking lasts.

TOGETHER WITH TELY AI

📝 Run a lead-generating SEO blog on autopilot with AI

The Rundown: With Tely AI, you don’t need an SEO expert or content team — AI does it all. It researches your industry, competitors, and website to find the best keywords for Google, delivering well-researched, expert-level articles promoting your company.

With Tely AI, you can:

• Publish up to 60 articles a month

• See Google index results in just 2 weeks

• Enjoy full automation for writing, SEO, and publishing

• Convert traffic into valuable leads

Get started and create your first article for free.

OPENAI

🎨 OpenAI adds image generation to GPT-4o, Sora

Image source: OpenAI

The Rundown: OpenAI released image generation within its GPT-4o model and Sora video generator, shifting from separate text and image systems to a fully integrated approach for producing more precise and contextually aware visuals via ChatGPT.

The details:

• GPT-4o treats images as part of its multimodal understanding, enabling more accurate text rendering and contextual awareness.

• The upgrade excels at generations like menus, diagrams, and infographics with readable text, addressing a major weakness of previous models.

• Users can also edit images with natural language, with the model able to maintain consistency between iterations and handle 10-20 objects in prompts.

• The new capability replaces DALL-E 3 as ChatGPT's default image generator for Free, Plus, Pro, and Team users, with Enterprise and Edu coming soon.

Why it matters: OpenAI’s DALL-E lagged far behind other image generators, but this long-awaited native image upgrade looks to be worth the wait. With long-text generation, UI/UX design skills, and natural language editing, visual content generation is entering a completely new era with this next generation of models.

AI TRAINING

🎬 Transform concepts into production-ready pitch decks

The Rundown: In this tutorial, you will learn how to create professional storyboards, videos, and complete pitch decks from simple text prompts in minutes — without any design experience.

Step-by-step:

1. Head over to LTX Studio, click Start with a concept, and enter a simple prompt like "30-second ad for [product]" with location and character details.

2. Customize project settings (aspect ratio and style) and edit scene elements using facial controls and generative fill.

3. Transform static frames into videos with the Motion Editor, and add a soundtrack or voiceovers to enhance your creation.

4. Export your finished project as a video, editing package, or as a professional pitch deck.

Pro tip: We did an extensive workshop showing how to transform your creative concepts into professional videos with LTX Studio here. Additionally, the Rundown University paid members get a free month of the Standard Plan.

PRESENTED BY CONCIERGE

🔑 Unlock the power of AI across your apps

The Rundown: Wish your AI assistant could understand your context and take actions for you? Concierge solves this by connecting seamlessly to your apps (including Gmail, HubSpot, Jira, Linear, Slack, Attio, Sheets, Airtable, and more).

With Concierge, you can:

• Instantly connect your favorite apps and get real-time answers based on your data

• Use top AI models (GPT, Claude, DeepSeek, etc.) without juggling multiple subscriptions

• Draft emails, update CRMs, create tickets, summarize notes, or do deep research — all with a single prompt

Try Concierge today and get AI to truly work for you.

APPLE

💰 Apple’s billion-dollar bet on Nvidia AI hardware

Image source: Reve / The Rundown

The Rundown: Apple is reportedly placing a massive $1B order for Nvidia's advanced servers, partnering with Dell and Super Micro Computer to set up its first generative AI infrastructure—signaling a major shift in the company's AI strategy amid Siri setbacks.

The details:

• Loop Capital analyst Anada Baruah reported the purchase includes roughly 250 Nvidia's GB300 NVL72 systems, with each server costing between $3.7- 4M.

• Both Dell Technologies and Super Micro Computer will reportedly serve as key server partners in building Apple's new large-scale AI cluster.

• While previous reports indicated Apple was developing its own AI chips, this purchase may be a response to slower-than-expected progress in that area.

• The move also comes after Apple’s setbacks in AI development, including delays to the planned AI-powered Siri upgrade and internal restructuring.

Why it matters: After staying on the AI data center sidelines while competitors raced ahead, Apple appears to be acknowledging it needs serious computing power to compete — and must look externally to right some of the issues currently plaguing its in-house AI progress. But the clock is ticking, and AI progress isn’t slowing down.

QUICK HITS

🛠️ Trending AI Tools

• 🎆 Reve Image 1.0 - Image model with advanced realism and prompt accuracy

• 🐳 DeepSeek V3-0324 - V3 upgraded with improved coding and reasoning

• 🧠 Qwen2.5-VL-32B - New vision-language AI with enhanced performance

• 🕺 LHM - Create animated 3D avatars from a single reference image

💼 AI Job Opportunities

• 🌿 Deepmind - Research Engineer, AI for Sustainability

• 📦 Coreweave - Inventory Control Specialist

• 🎨 Anthropic - Brand Designer, Web Experience

• 🤝 Snorkel - Strategic Engagements Lead

📰 Everything else in AI today

OpenAI announced new upgrades to its Advanced Voice Mode, featuring new personality upgrades and fewer interruptions for more natural conversations.

Figure AI published new research and demos of its Figure 02 humanoid achieving natural human-like walking, conducting years worth of simulated training in just hours.

H&M is partnering with 30 models to create AI-based digital twins for ad campaigns, with models maintaining ownership rights and receiving usage-based compensation.

ByteDance released InfiniteYou, an open-source AI portrait generator that produces consistent portraits with enhanced facial accuracy and prompt adherence.

Synthesia launched a $1M equity program for actors with likenesses featured as AI avatars, becoming the first to offer stocks to performers contributing to AI training.

Otter AI unveiled three AI Meeting Agents, including a voice-activated Meeting Agent, a Sales Agent for on-call coaching, and an SDR Agent for autonomous product demos.

Perplexity added new answer modes, enhancing searches on specific verticals with entities like images, videos, and cards with built-in commercial transactions.

COMMUNITY

🎥 Join our next live workshop

Join our next special workshop today at 2 PM EST to learn how to build an AI-powered sales rep using Zapier Agents with Jane Zhang and Anna Marie Clifton from the Zapier team.

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

Good morning, tech enthusiasts. Genetic testing company 23andMe—which was once valued at over $6 billion but was soon left in shambles due to a major data breach—just filed for bankruptcy.

However, if the company is sold, what happens to the sensitive data of its 15 million customers? While people are flocking to quickly delete their accounts, doing so is easier said than done.

In today’s tech rundown:

• 23andMe files for bankruptcy

• Samsung’s AR smart glasses

• Tesla’s European sales take a hit

• Microsoft’s quantum chip sparks debate

• Quick hits on other major news

LATEST DEVELOPMENTS

23ANDME

🔬 23andMe files for bankruptcy

Image source: 23andMe

The Rundown: Genetic testing company 23andMe filed for Chapter 11 bankruptcy protection on Sunday, marking a dramatic turn of events for the once-pioneering biotech firm—and raising concerns about the genetic data of its 15M customers.

The details:

• 23andMe, which allowed customers to explore their genetic makeup through saliva samples, cited ongoing financial struggles as the reason for the decision.

• Co-founder and CEO Anne Wojcicki resigned this week following unsuccessful attempts to privatize the company amid ongoing financial struggles.

• The company’s trouble started in 2023, when it suffered from a massive data breach that exposed the personal information of nearly 7M customers.

• 23andMe was valued at $6B after going public in 2021, but it all fell to less than $50M by the time the company filed for bankruptcy.

Why it matters: While 23andMe assures it will protect the data of all customers during the transition, those seeking to remove it could face challenges. Even after receiving deletion requests, the company has to retain at least some data for up to three years—including your genetic info, date of birth, and sex—to comply with legal obligations.

SAMSUNG

👓 Samsung’s AR smart glasses

Image source: Ideogram/The Rundown

The Rundown: Samsung is reportedly finalizing a new line of augmented reality smart glasses, codenamed “Haean,” with a built-in display and advanced AI features—all set to launch later this year and take on Apple and Meta’s AR devices.

The details:

• While Samsung has yet to confirm Haean, rumors suggest the glasses will come with AI features, including gesture recognition and payment capabilities.

• They are expected to use multiple cameras and sensors to track user movements and interact with the environment.

• On the software side, the glasses are anticipated to run Google's Android XR OS, coupled tightly with Samsung’s smart devices and wearables.

• Samsung reportedly plans to launch the glasses by the end of 2025, alongside its upcoming extended reality (XR) headset—Project Moohan.  

Why it matters: Positioned to compete with AR industry giants, Samsung’s version promises advanced features like facial recognition, gesture-based controls, and real-time text translation. This launch is also a key move in Samsung’s broader strategy to expand its presence in the XR market.

TESLA

🚘 Tesla’s European sales take a hit

Image source: Tesla

The Rundown: Tesla's European sales dropped by nearly 45% in February, resulting in a market share of just 9.6%—the lowest in five years. Meanwhile, Chinese rival BYD has overtaken Tesla in global revenue, clocking more than $100B in 2024.

The details:

• Chinese EV brands sold nearly 20,000 vehicles in Europe last month, compared to Tesla's 15,700 units.

• Analysts attribute this downturn to CEO Elon Musk's political involvement with Donald Trump and anti-labor sentiment, which has led to consumer backlash.

• Additionally, delays in updates to the Model 3 and Model Y may have caused potential buyers to delay purchases, further impacting sales. ​

• BYD, one of Tesla’s biggest Chinese rivals, recently took a lead in EV revenues and reported $107B in 2024 — a 29% increase from the previous year.

Why it matters: Tesla is feeling the heat from European and Chinese competitors, who are quickly closing the gap in terms of price, features, and overall availability. BYD, despite the newly imposed EU tariffs, is pushing its strategy of low-priced EVs and plug-ins, with sales projections pointing toward rapid growth this year.

MICROSOFT

☄️ Microsoft’s quantum chip sparks debate

Image source: Microsoft

The Rundown: Microsoft's recent announcement of Majorana 1 quantum chip, which is said to be powered by topological qubits, ignited a fierce debate among physicists this week, who claim it lacks conclusive evidence.

The details:

• At a global physics summit, Microsoft’s quantum lead presented data claiming the chip uses elusive Majorana zero modes to create error-resistant qubits.

• Physicists criticized the lack of evidence, noting that Microsoft’s Nature paper explicitly states the results “do not represent evidence for topological modes.”

• Skeptics argued that the company’s protocol test could produce false positives, calling the data provided “unconvincing.”

• Microsoft’s claims, which CEO Satya Nadella likened to “quantum’s transistor moment,” have drawn comparisons to its 2021 retraction of similar research.

Why it matters: Amazon’s quantum execs have reportedly dismissed the claims as “hype,” and critics warn that overstated breakthroughs risk undermining public trust in quantum computing, especially as Microsoft’s stock surged post-announcement. However, Microsoft maintains its findings are valid and plans to release additional data.

QUICK HITS

📰 Everything else in tech today

Apple announced iOS 18.4 will start rolling out next month with several new features, including lossless audio and ultra-low latency audio to AirPods Max.

Waymo, Alphabet’s autonomous vehicle subsidiary, announced plans to launch its robotaxi service, Waymo One, in Washington, D.C., by 2026.  

Han Jong-hee, co-chief executive officer of Samsung Electronics, passed away at the age of 63 due to a heart attack, the company announced.

Tech giant IBM is set to cut around 9,000 U.S. jobs this year, with a significant portion of the roles reportedly shifting to India.

Apple is reportedly exploring the integration of AI-powered cameras into future versions of both the standard and Ultra versions of the Apple Watch.  

Fintech startup Alchemy launched a $100M fund to provide content creators with upfront payments based on projected future income.

Revel, a NYC-based EV charging infrastructure startup, launched its first fast-charging station in San Francisco, with plans to expand across the Bay Area soon.

Yale researchers developed a CRISPR technology that can assess genetic interactions on a host of immunological responses to diseases, including cancer.

German startup Isar Aerospace is gearing up to launch its Spectrum rocket, marking the inaugural flight of an orbital launch vehicle from continental Europe.

Yahoo struck a deal to sell TechCrunch, the 20-year-old tech journalism site, to media investment firm Regent for an undisclosed amount.

Twitter's iconic bird logo, which was removed from the company's former San Francisco headquarters, sold for nearly $35,000 at an auction.

A new survey finds that two in five tech workers quit in the past year due to a lack of flexibility concerning hours, location, and the “intensity of work.”

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. You'll learn how to use the latest AI tools to take your content creation to the next level.

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

Good morning, AI enthusiasts. The text-to-image race just got a surprising new frontrunner — with Reve emerging from stealth with a model that instantly topped global rankings under the codename "Halfmoon."

With next-level design and text capabilities, indistinguishable-from-reality photorealism, and world-class prompt adherence, a new AI creative powerhouse just announced itself in a BIG way.

In today’s AI rundown:

• Reve’s new leading image model

• DeepSeek quietly drops V3 upgrade

• Turn any YouTube video into your tutor

• ARC Prize returns to challenge AI reasoning

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

REVE

🌜 Reve’s new leading image model

Image source: Reve / The Rundown

The Rundown: Reve just emerged from stealth with Reve Image 1.0, a new text-to-image AI model that topped global rankings with the codename “Halfmoon” over the last week—showcasing exceptional prompt accuracy, text rendering, and image quality.

The details:

• The model claimed the #1 position in Artificial Analysis' Image Arena, outperforming rivals like Google's Imagen 3, Midjourney v6.1, and Recraft V3.

• Reve said its mission is to “enhance visual generative models with logic,” with 1.0 showing impressive prompt adherence and long text rendering in tests.

• The platform also features natural language editing, photo uploads, and an ‘explore’ tab to view community prompts and generations.

• A preview of Reve Image 1.0 is currently free to try (though no API access yet), with the company saying that “much more is coming soon”.

Why it matters: What a stealth debut from Reve, with their first model already topping the leaderboards against established giants in the text-to-image arena. 1.0 seems to combine the best of the SOTA image models — with extreme photorealism, world-class prompt following, editing tools, and absolutely next-level text capabilities.

TOGETHER WITH VANTA

🦺 Simplify compliance with automation and AI

The Rundown: Navigating new compliance requirements can be a daunting task, but with the right automation tools, it doesn't have to be. Whether you’re a fast-growing startup or an established 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

Secure your spot today.

DEEPSEEK

🐳 DeepSeek quietly drops V3 upgrade

Image source: Deepseek

The Rundown: Chinese AI startup DeepSeek just released an updated version of its V3 model, a massive 641GB model capable of running on high-end personal computers — also featuring a highly permissive open source MIT License for broad use.

The details:

• The V3 update, V3-0324, uses a Mixture-of-Experts architecture that activates only 37B parameters per token, dramatically reducing compute demands.

• Testers have shown it can run smoothly on Apple's Mac Studio computers, making it the first model of this caliber accessible outside data centers.

• Early users have also reported upgraded math and coding capabilities, with another calling it the best non-reasoning model available.

• V3-0324 can be accessed with an open-source MIT License, a change from the previous V3 model that had a more restrictive custom license.

Why it matters: China’s AI darling continues to ship, with a supposedly minor update bringing some big upgrades. Rumors about the upcoming R2 release are also gaining momentum, hinting at another 'DeepSeek moment' that could shake the AI world—potentially signaling a new leader in the field.

AI TRAINING

🎓 Turn any YouTube video into your tutor

The Rundown: In this tutorial, you will learn how to convert YouTube videos into personal tutors with Google AI Studio, which lets you ask questions about any video by simply pasting the link—making complex content instantly accessible for learning.

Step-by-step:

1. Visit Google AI Studio and log in with your Google account.

2. Select "Gemini 2.0 Flash" from the Model dropdown menu on the right side of the screen.

3. Paste your YouTube video link in the prompt area, followed by your specific question about the content.

4. Ask follow-up questions to explore the video content more deeply, referencing specific timestamps if needed.

Pro tip: When exploring videos on complex educational subjects, you can ask the AI to "Create a simple explanation of the [concept]" or "Generate a quiz based on this lecture to test my understanding."

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.

ARC FOUNDATION

🧠 ARC Prize returns to challenge AI reasoning

Image source: ARC Foundation

The Rundown: The ARC Prize Foundation just launched ARC-AGI-2, a new benchmark designed to push the frontier of AI reasoning capabilities — alongside a $1M competition to drive research toward more efficient general intelligence systems.

The details:

• ARC-AGI-2 targets skills that remain difficult for AI while being easy for humans, with tasks solvable by at least two humans in under two attempts.

• Current AI reasoners perform poorly on ARC-AGI-2, with even OpenAI's o3-low scoring only an estimated 4%, compared to 75.7% on the previous version.

• ARC also introduced an efficiency metric alongside raw performance, measuring cost per task to test both capability and resource efficiency.

• The ARC Prize 2025 competition launches this week with $1M in prizes, including $700k for the first to achieve 85% accuracy within efficiency limits.

Why it matters: Many of the top AI labs feel that AGI is just around the corner, and ARC’s original AGI-1 benchmark was already seeing top models like the o3 score in the 80s (not without controversy). While the new iteration may be a better test, it’s still hard to imagine it lasting long — but the AGI goalposts will likely continue to shift.

QUICK HITS

🛠️ Trending AI Tools

• 🪄 Promptimize - Browser extension that gives you the power of an expert prompt engineer in a single click*

• 🧠 Hunyuan T1 - Tencent AI matching DeepSeek in performance and pricing

• ⚙️ Zapier MCP - Connect AI across 8,000+ apps without API integrations

• 🤔 Claude “Think” - Tool to improve Claude’s problem-solving performance

*Sponsored Listing

💼 AI Job Opportunities

• 🔍 Anyscale - Technical Sourcer

• 🎬 Captions - Software Engineer, Video Processing

• 🗄️ Anthropic - Data Infra Engineer, Pretraining

• 👥 xAI - People Coordinator

📰 Everything else in AI today

Sam Altman announced new internal leadership promotions at OpenAI, with Mark Chen appointed as Chief Research Officer and Brad Lightcap’s role expanding as COO.

Alibaba’s Qwen team open-sourced Qwen2.5-VL-32B-Instruct, a new vision-language model featuring enhanced mathematical reasoning and visual capabilities.

Netflix co-founder Reed Hastings donated $50M to his alma mater, establishing the Hastings Initiative for AI and Humanity to study the tech’s risks and future impact.

Alibaba-affiliate Ant Group reportedly adopted a hybrid chip strategy combining Chinese and American chips, which reduced AI development costs by 20%.

Google started rolling out ‘Project Astra’ features that give advanced vision, live video, and screen reading capabilities to Gemini.

Alibaba released LHM, a new AI model that creates animated 3D avatars from just a single reference image.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. You'll learn how to use the latest AI tools to take your content creation to the next level.

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

Good morning, robotics enthusiasts. Kyle Vogt, who co-founded Cruise and Twitch, is now racing ahead with The Bot Company—a robotics startup that just raised $150M, reaching a $2B valuation in less than a year since its inception.

The San Francisco-based company is tapping a market hungry for AI-powered robots with a “non-humanoid” bot for helping with household chores. But with no design yet on the table, can it scale up in time to compete with Amazon, Figure, and 1X?

In today’s robotics rundown:

• Ex-Cruise CEO’s robot startup valued at $2B

• Undersea bots to search for missing MH370

• Tesla to build ‘legions’ of robots

• 1X launches NEO Gamma’s home trials

• Quick hits on other robotics news

LATEST DEVELOPMENTS

THE BOT COMPANY

🦄 Ex-Cruise CEO’s robot startup valued at $2B

Image source: Ideogram/The Rundown

The Rundown: Former Cruise CEO Kyle Vogt’s new San Francisco-based robotics startup, The Bot Company, has reportedly raised $150M in funding at a $2B valuation—without even launching a product to date.

The details:

• Vogt founded the company last year alongside Paril Jain, who led Tesla’s AI team, and former Cruise software engineer Luke Holoubek.

• Sources who spoke to Reuters say the company is developing non-humanoid robots equipped with a base and grips to assist with household chores.

• The company plans to use advanced AI models, including large language models, to enable its robots to understand natural language commands.

• Vogt, who also co-founded Twitch, launched the startup five months after he resigned as CEO of the Cruise, the robotaxi startup acquired by GM.

Why it matters: The Bot Company is tapping the market for domestic automation, drawing strong investor backing from the outset. Yet with few design details revealed, investors seem to be betting solely on its potential—leaving open the question of whether it can keep pace with competitors like 1X, Amazon, and Figure.

OCEAN INFINITY

🚤 Undersea bots to search for missing MH370

Image source: Ocean Infinity

The Rundown: Texas robotics company Ocean Infinity has been granted a permit to renew the search for Malaysia Airlines Flight 370, which is thought to have crashed in the southern Indian Ocean more than a decade ago, killing all 239 people onboard.

The details:

• Ocean Infinity will deploy a fleet of autonomous underwater vehicles capable of operating at depths up to 6,000 meters for as long as 100 hours at a time.

• These vehicles are equipped with advanced sensors, including AquaPix sonar, which provides high-res imagery and 3D digital terrain maps of the seafloor.

• Multiple AUVs can be deployed simultaneously, allowing for efficient coverage of the 5,800-square-mile search area.

• Ocean Infinity's new state-of-the-art Armada 7806 offshore support vessel will serve as a base and can be operated remotely, requiring no onboard crew.

Why it matters: The Malaysian government will pay Ocean Infinity a fee of $70M if its 18‑month search locates the wreckage. This latest effort follows previous unsuccessful attempts—including a search that ended in 2018—though Ocean Infinity claims that its new technological advancements now position it for success.

TESLA

🤖 Tesla to build ‘legions’ of robots

Image source: Tesla

The Rundown: At an emergency all-hands meeting last week, Tesla CEO Elon Musk announced that the first Optimus humanoid has been completed at the new pilot production line at the Fremont Factory—with "legions" of robots expected soon.

The details:

• Musk’s ambitious production targets for Optimus include 5,000 units this year, with the capacity to produce up to 12,000 units if necessary.

• Looking ahead, the company plans to increase production to 50,000 units in 2026, with the possibility of using snap-on parts for customized use cases.

• He also mentioned the potential for Optimus to be used outside of Tesla-controlled environments by the second half of 2026.

• Musk predicted that Optimus would be the “biggest product of all time” and could be 10 times or even 20 times larger than any other product ever made.

Why it matters: Musk has said that Optimus has the potential to reach up to 10,000 different tasks in future updates and significantly increase Tesla’s profits by trillions of dollars. While competitors like Apptronik and Figure have a head start in real-world applications, none have publicly announced production goals to match Tesla’s.

1X

🧹 1X launches NEO Gamma’s home trials

Image source: 1X

The Rundown: Norwegian robotics startup 1X is set to launch early in-home trials of its NEO Gamma humanoid by the end of 2025, aiming to place the beige knit-suited bot in "a few hundred to a few thousand" homes.

The details:

• While NEO Gamma uses AI to walk and balance, it is not yet fully autonomous and will rely heavily on human teleoperators during these early trials.

• This “early adoption" program aims to gather data for developing autonomous capabilities and to teach the robot how to behave in real-world contexts.

• NEO Gamma, unveiled in February, is designed to perform various household tasks such as vacuuming, watering plants, and making coffee.

• The company is outspoken in saying its humanoid is still a “long way away from commercial scaling and deployment.”

Why it matters: 1X CEO Bernt Børnich says NEO Gamma needs to "live and learn among people" to advance its development, but collecting data from microphones and cameras in homes raises privacy flags. Still, 1X says moving NEO Gamma out of the lab and into homes is a crucial step to advancing the tech.

QUICK HITS

📰 Everything else in robotics today

Global ad firm WPP is partnering with Nvidia, Boston Dynamics, and Cannon to explore robotic-assisted cinematography, releasing a video of Atlas filming on set. 

Mercedes-Benz is trialing a handful of Apptronik’s Apollo robots in two of its auto plants for tasks such as moving components to the production line and quality checks.

California-based Dexterity launched a massive new robot called Mech, touted as the “world’s first industrial superhumanoid,” which can lift 136 lb. with its arms.

BlueHalo announced that it has won a contract worth more than $30M to provide its advanced underwater robotic tech for U.S. Navy expeditionary and security missions. 

Lyft announced it is working with May Mobility to introduce autonomous vehicles on its platform, with the rollout set to begin in Atlanta this summer and Dallas next year.

Tokyo Electric Power Company is set to deploy robots to retrieve highly radioactive sandbags at the Fukushima Daiichi nuclear power plant this week.

Swedish AI startup IntuiCell created a robot dog named Luna that it says has a functional digital nervous system capable of learning and adapting like humans.

SwitchBot is integrating more than 45 of its products with Home Assistant in the coming months, including robot vacuums, smart shades, and its garage door opener.

Stockholm-based Rerun, a company developing physical AI for robotics, drones, and autonomous vehicles, has raised $17M, bringing its total funding to $20.2M.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop on Wednesday at 2 PM EST to learn how to build an AI-powered sales rep using Zapier Agents — with Jane Zhang and Anna Marie Clifton from the Zapier team.

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

Good morning, AI enthusiasts. AI just crossed a critical medical threshold — blowing away human experts with a 99.26% accuracy rate in detecting certain types of cancer.

With breakthrough treatments, rapidly evolving patient care, and now near-perfect diagnostic capabilities, AI’s future in stomping out cancer is looking brighter than ever.

In today’s AI rundown:

• New AI’s near-perfect cancer detection

• Tencent’s Hunyuan T1 reasoning model

• Create custom AI videos to boost engagement

• Perplexity’s bold bid to take over TikTok

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

AI RESEARCH

🔬 New AI’s near-perfect cancer detection

Image source: Daffodil International University and Charles Darwin University

The Rundown: Researchers just unveiled an AI model called ECgMLP that identifies endometrial cancer with 99.26% accuracy from microscopic tissue images—drastically outperforming human specialists and current automated methods.

The details:

• ECgMLP uses specialized attention mechanisms to spot cancer cells in microscopic tissue images that doctors might miss during standard analysis.

• Current human diagnostic methods for endometrial cancer only achieve 78-81% accuracy, far below this model’s accuracy of more than 99%.

• Researchers also tested its versatility across other cancers, detecting colorectal (98.57%), breast (98.20%), and oral (97.34%) with high accuracy.

Why it matters: Medical diagnostics are undergoing a major shift, with AI now consistently outperforming humans in life-saving detection tasks. With many cancers being highly treatable when caught early, these models will save a lot of lives — and eventually democratize access to expert-level cancer screening worldwide.

TOGETHER WITH VAPI

🗣️Voice AI agents for developers

The Rundown Vapi delivers natural voice interactions with flexible, secure integration. Vapi’s platform powers both simple inbound agents and complex interactive AI solutions, integrating any LLM, voice model, or your own tools.

Over 120k developers — from startups to Fortune 500 companies — trust Vapi for:

• Flexible integration with your own data, APIs, and models

• Human-like agents for natural conversations

• Scaling to millions of calls at <500ms latency

• Enterprise-grade security with LLM guardrails with HIPAA/SOC-2

Call 1‑844‑439‑8274 or click here to talk to Vapi.

TENCENT

🧠 Tencent’s Hunyuan T1 reasoning model

Image source: Tencent

The Rundown: Tencent just released Hunyuan T1, a new reasoning model that matches DeepSeek's R1 in performance and pricing—while tapping the industry's first hybrid Transformer-Mamba architecture for improved efficiency.

The details:

• T1 matches or surpasses rivals like DeepSeek R1 and OpenAI’s o1 and GPT 4.5 across benchmarks, excelling especially in math and Chinese language evals.

• Tencent claims the model is the first to combine Google's Transformer architecture with Carnegie Mellon and Princeton researchers’ Mamba system.

• The hybrid approach reportedly delivers 2x faster speeds while reducing computing demands, particularly when handling long-text reasoning tasks.

• As for pricing, T1 matches DeepSeek's competitive rates at 1 yuan ($0.14) per million tokens for input and 4 yuan ($0.55) per million tokens for output.

Why it matters: Between DeepSeek, Tencent, and Alibaba, China’s AI labs have almost completely closed the gap with the U.S. leaders — something that felt extremely far off just a year ago. With the next-gen R2 also coming soon, China feels closer than ever to officially taking the lead for the world’s top AI models.

AI TRAINING

🎬 Create custom AI videos to boost engagement

The Rundown: In this tutorial, you will learn how to use Synthesia's AI video platform to create personalized videos using AI avatars and add them to your emails to increase response rates.

Step-by-step:

1. Create a free Synthesia account, select a template, and add your script and on-screen text to create a basic video.

2. Add personalization variables using double curly brackets in both your script and on-screen text, then convert to a template.

3. Use your template to create individual videos by filling in the variable fields for each recipient.

4. Copy video thumbnails as GIFs directly into your emails and add call-to-action buttons to drive conversions.

Pro tip: You can also check out our extensive workshop showing how to create hyper-personalized video messages that drive engagement with Synthesia here. The Rundown University’s paid members will also get a 30% discount on the Creator Plan.

PRESENTED BY RECRAFT

🎨 Elevate your creative process with Recraft

The Rundown: Recraft is an AI-powered image generation and editing tool built to give creators total control, not replace their creativity. Trusted by 3M+ users, Recraft’s V3 model sits atop the Hugging Face text-to-image leaderboard — ahead of alternatives like Midjourney and DALL·E.

With Recraft, you can:

• Generate lifelike images (with correct anatomy) indistinguishable from real photos

• Enjoy real-time editing on an infinite canvas for true creative freedom

• Create precise, accurate text overlays for posters, banners, and layouts

• Keep complete ownership of your creative process

Incorporate Recraft into your design process today — use code RUNDOWN12 for $12 off any plan, exclusive to The Rundown AI readers.

PERPLEXITY

🔍 Perplexity’s bold bid to take over TikTok

Image source: Perplexity

The Rundown: AI search startup Perplexity just published its proposal to acquire TikTok's U.S. operations, promising to rebuild the platform's algorithm with transparency and American oversight while integrating its own search tech.

The details:

• Perplexity plans to reconstruct TikTok's recommendation system in American data centers, promising full transparency by making the algorithm open-source.

• The company would integrate its AI citation capabilities with TikTok videos, enabling users to cross-reference info in real time while watching content.

• Perplexity also proposed enhancing TikTok with Nvidia Dynamo technology to scale recommendation models "100x" while improving inference speed.

• The vision includes cross-platform benefits, with TikTok videos in Perplexity search results, and Perplexity's information engine powering TikTok searches.

Why it matters: Perplexity has had a wild few years, evolving from an AI search startup to developing its own models, partnering on an AI phone, building an AI browser, antagonizing Google in commercials, and now bidding on TikTok. It could be another publicity stunt, but the ban deadline is April 5 — so we’ll find out soon enough.

QUICK HITS

🛠️ Trending AI Tools

• 🛜 Web Search - Access the web and real-time info with Claude 3.7 Sonnet

• 🗣️ GPT-4o-tts and Transcribe - OpenAI’s text-to-speech & speech-to-text AI

• 🗺️ Mind Maps - New interactive visual creation tool in Google’s NotebookLM

• 🎥 KREA Video Training - Train video AI with your videos for custom styles

💼 AI Job Opportunities

• 🚀 The Rundown - Enterprise Partnerships Lead

• 📋 Snorkel - Technical Delivery Manager

• ✏️ Shield AI - Associate Visual Designer

• 🤖 Captions - Member of Technical Staff, ML Ops

📰 Everything else in AI today

Anthropic introduced a "think" tool for Claude, enabling the AI assistant to perform structured reasoning when handling complex tool use tasks.

OpenAI and Meta are seeking a partnership with India's Reliance Industries, with OpenAI considering a price cut of up to 85% on ChatGPT to break into the market.

Kai-Fu Lee said that "Sam Altman is probably not sleeping well,” with his startup 01.AI pivoting to DeepSeek's open models and operating at just 2% of OpenAI's annual costs.

Apple is developing a camera-equipped Apple Watch with AI Visual Intelligence features using its own models — hoping for implementation by 2027.

Zapier launched its own MCP protocol, enabling AI assistants to perform actions across 8,000+ apps without complex API integrations.

New studies from OpenAI and MIT research found that increased ChatGPT usage correlates with higher loneliness and emotional dependence in users.

Browser Use, which makes websites easier to navigate for AI agents, raised $17M in seed from Felicis’ Astasia Myers, Paul Graham, A Capital, and Nexus Venture Partners.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop on Wednesday at 2 PM EST to learn how to build an AI-powered sales rep using Zapier Agents — with Jane Zhang and Anna Marie Clifton from the Zapier team.

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