🤖 DeepMind's robots 'think' offline

Welcome back, Superhuman. How much should I read into this? Sam Altman just asked his followers to predict when they think an o3-mini-level model will be able to run on a phone, and most said this year. With how fast things are moving, I wouldn’t bet against it.

Also today: Learn how to generate infographics from plain text, and get the latest AI tools and trending social media posts.

TODAY IN AI

1. You can now try Google’s latest image model at no cost: The search giant is opening up access to Imagen 4, an image model that’s especially good at fine details, photorealism, advanced spelling, and capturing a wide variety of art styles. Meanwhile, a more advanced version called Imagen 4 Ultra can handle more complicated text prompts. Both are now available in the Gemini API (a platform geared toward developers) and AI Studio, where you can try them at no cost for “a limited time.”

2. OpenAI is coming for Microsoft and Google: No wonder Microsoft and OpenAI’s tight-knit partnership is starting to fray. The Information reports that OpenAI has been working for at least a year on a collaborative platform similar to Microsoft Office that would let multiple people edit projects and send messages directly within ChatGPT. If OpenAI drops the tool, it could significantly eat into Microsoft and even Google Workspace’s user bases. That’s on top of the news that Microsoft is already losing Copilot clients to ChatGPT.

3. AI beats humans for top slot on prestigious hacker leaderboard: Human red teaming (purposefully infiltrating companies to beef up their security) usually takes weeks and costs around $18,000, according to Oege de Moor, founder of the year-old “hackbot” startup Xbow. The company just raised $75M to automate that process with AI — and it’s already paying off. Security company HackerOne just ranked the tool #1 on its US leaderboard. The ranking measures how many security flaws a hacker finds as well as the significance of each one.

PRESENTED BY AIRTABLE

Put AI to real work: The Airtable AI Playbook

Skip the code. Transform your data into custom interfaces, automations, and agents with Airtable's AI-native app platform.

Get inspired by our most impactful, real-world AI use cases, and try them out yourself to unlock immediate business value.

Explore the playbook today.

FROM THE FRONTIER

Gemini Robotics unveils model that can run locally on robots

Companies have been trying to infuse robots with reasoning capabilities so they can handle more complex tasks — say, doing your chores or automating warehouse work — without prior training. But these robots usually have to rely on a cloud connection to get the job done.

That’s about to change with Google DeepMind’s new vision language action model (VLA), which puts Gemini 2.0’s real-world understanding directly on-device. We talked to DeepMind’s Head of Robotics, Carolina Parada, to find out more.

What paved the way for this release? Because Gemini has watched so many videos of objects in motion, it can now use that knowledge to control a robot’s behavior, with state-of-the-art dexterity. “The same way that Gemini can output text (write poetry), output images, output code, it can also now output actions,” Parada said.

Why focus on a local model? This on-device breakthrough cuts down on latency and lets you use robots without an internet connection. It’s one step toward truly general-purpose robots, which can work in many different environments and don’t need months of training to pick up new skills.

Who’s it for? It already works well with Aloha, Franka, and Apptronik robots. But for the first time, developers and researchers can also fine-tune the model to fit their robotic platform of choice. “The strength of your base model is basically what translates to generalization on the robotics side,” Parada added.

What’s next? “We’re just scratching the surface [in terms of] capturing intelligence from these foundation models onto robotics. There’s all kinds of other aspects, like agentic behaviors and memory” that have yet to be explored, Parada said. “I do imagine that in the next couple of years, the picture is going to look very different.”

THE AI ACADEMY

How to generate infographics from plain text

• Go to Claude and sign up with your account.

• Select ‘Claude Sonnet 4’ as your model, enter your prompt, and press Enter.

Sample Prompt: "You are a world-class visual explainer and technical designer. Your task is to turn this concept into a visual infographic using Mermaid.js or another code-based diagram format: "[INSERT CONCEPT HERE]"

Return the output as:

1. A clear visual breakdown using a format like a flowchart, timeline, concept map, or decision tree — whichever suits best.

2. A plain English caption explaining what the graphic shows.

3. Clean Mermaid code (or HTML/SVG/CSS if more appropriate) that I can copy and paste to render the graphic. Keep it readable, elegant, and minimal like a slide in a consulting deck.”

• You’ll get a visual diagram with all the details ready within seconds.

• You can convert it into a presentation or download the code to use it.

^{Prompt source: godofprompt}

PRESENTED BY DESCOPE

Make your app AI-ready–the secure way

Your APIs weren’t built for AI agents, but Descope fixes that by making APIs and remote MCP servers OAuth-compliant for seamless agent connectivity.

Drag-and-drop simplicity. Scoped access. User consent. Trusted by Databricks and GoFundMe.

Sign up at no cost and make your app AI-ready today.

IN THE KNOW

What’s trending on socials and headlines today

🤳 Insta Automation: A Redditor claims they used AI agents to put an “Instagram account on full autopilot,” generating 4.4M views in just three weeks. Here’s how.

🔁 Secret Weapon: This viral 37-minute tutorial teaches you how to use MCP to make Claude “10x more powerful.”

🧑‍💻 Code Coup: Builder Sherry Jiang thinks it’s wild that single-feature, $1B+ companies like Docusign and Calendly “could now [be] vibe-coded in a weekend.”

✨ Also Worth Checking: Ten handy Claude 4 prompts; AI-generated ASMR; and dogs’ turn on the Olympic diving board.

Sam Altman fired back at a startup’s claims that OpenAI stole its name and ideas, calling its lawsuit “silly, disappointing, and wrong.”

Anthropic’s use of printed books to train its models should be considered “fair use,” according to a first-of-its-kind copyright decision that some are calling a big win for the AI industry.

ElevenLabs launched a mobile app for both iOS and Android, letting users generate “studio-quality voiceovers” right from their phones.

PRODUCTIVITY

5 New AI Tools to Boost Your Productivity

• 🫂 Pally: Bring together connections across all your socials using AI.

• 📱 Slashit: Automate repetitive typing and enhance texts with shortcuts.

• 📁 Dynbox: Organize all your cloud and local files by chatting with an AI.

• 🎨 Pixlr: Edit photos, generate images, and design anything with AI.

• 💼 Supawork: Create professional headshots for your resume.

^{* indicates a promoted tool, if any}

PROMPT STATION

Brutally Honest Growth Advisor

Prompt: I want you to act and take on the role of my brutally honest, high-level advisor.

Speak to me like I'm a founder, creator, or leader with massive potential but who also has blind spots, weaknesses, or delusions that need to be cut through immediately.

I don't want comfort. I don't want fluff. I want truth that stings, if that's what it takes to grow.
Give me your full, unfiltered analysis—even if it's harsh, even if it questions my decisions, mindset, behavior, or direction.

Look at my situation with complete objectivity and strategic depth. I want you to tell me what I'm doing wrong, what I'm underestimating, what I'm avoiding, what excuses I'm making, and where I'm wasting time or playing small.

Then tell me what I need to do, think, or build in order to actually get to the next level—with precision, clarity, and ruthless prioritization.

If I'm lost, call it out.
If I'm making a mistake, explain why.
If I'm on the right path but moving too slow or with the wrong energy, tell me how to fix it.
Hold nothing back.

Treat me like someone whose success depends on hearing the truth, not being coddled.

Source: honestprompts

Real Life Tom and Jerry

ChatGPT Prompt: in real life tom and jerry reading their stories from the book and laughing about it

Whenever you’re ready to take the next step

• Check out our Top 125 AI Tools

• Get better at prompting with our Top 1,000 Prompts

• Learn how to use AI at work and get certified with our free course

Grow customers & revenue: Join companies like Amazon, Hubspot, and Salesforce. Showcase your product to our 1M+ readers and 2M+ followers on socials. Get in touch.

Until next time — Zain & the Superhuman AI team