Good morning, {{ first_name | AI enthusiasts }}. Midjourney has spent years turning text prompts into surreal images. But its next big image project is a little more personal: the inside of your body.
The company just revealed a full-body scanner that lowers users through an ultrasound ring for fast body maps, and wraps the whole thing in a spa rollout — becoming easily the most surprising AI launch of the week.
P.S. — We’re excited to roll out a new weekly feature called Rowan’s Corner, where our founder and CEO Rowan Cheung will share his more personal takes on AI, building The Rundown, and more! Check it out below.
In today’s AI rundown:
Midjourney’s wild medical hardware pivot
Rowan’s Corner: How I’m preparing for an AI doctor
Save hours on meeting prep with Google Gemini
OpenAI poaches a transformer pioneer from Google
4 new AI tools, community workflows, and more
LATEST DEVELOPMENTS
MIDJOURNEY

Image source: Midjourney
The Rundown: Midjourney pivots from image generation to medical hardware: the Midjourney Scanner uses underwater ultrasonic sensors to image the full body in 60 seconds, with plans to house the device inside its own spas starting in 2027.
The details:
The scanner lowers users through water and a ring of ultrasound sensors, with Midjourney aiming to finish a full-body scan in just 60 seconds.
Midjourney founder David Holz says it rivals an MRI’s detail in a fraction of the time, and built the machine with ultrasound-chip maker Butterfly Network.
The first Midjourney Spa opens in 2027 in San Francisco’s Union Square, pairing around 10 scanners with saunas, cold plunges, and hot tubs.
Why it matters: Although there is still plenty to prove out, sign us up for an eventual spa day that doubles as a full-body health scan. Midjourney’s hardware vision is the type of device the future has promised but seemed far-fetched. After this wild reveal, the startup’s other mysterious “TBA” products gain even more intrigue.
TOGETHER WITH MERCURY
The Rundown: Most founders only understand their finances by exporting data into spreadsheets, because their bank* can’t do the work itself. Mercury Command flips that — it’s AI that takes action for you inside your account. You review and approve everything.
With Mercury Command, you can:
Ask “what’s my cash flow” or “follow up on that invoice” and it’s done
Get actionable insights from your live account data
Turn natural language into completed work across Mercury
*Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column N.A., Members FDIC.
ROWAN’S CORNER
The Rundown: Rowan’s Corner is a new weekly feature where our founder and CEO Rowan Cheung shares his more personal takes on AI, his conversations with leaders in the industry, building The Rundown, and more.
Rowan: It’s been weeks since I spoke with Demis Hassabis, but I can’t stop thinking about what he called “the biggest watershed moment in AI.”
A year ago, he told me AI could help cure all diseases within our lifetimes. This time, I asked if his timeline had changed. “They’ve not shifted, but they’ve hardened, they’ve tightened,” he said. “I'm very confident.”
Let that sink in: the Nobel laureate running Google DeepMind is growing confident all disease could be cured within 10–20 years.
So I started thinking: if medical superintelligence is approaching, what’s the smartest way to get ready for it today? (My best theory below, but hit reply and tell me yours if you disagree!)
My answer is to collect as much data as possible to feed the eventual medical superintelligence. Wearing fitness trackers, tracking workouts, and getting blood tests that track as many biomarkers as possible.
Today, this data can help me with little things. Vitamin D3 low? Cool, I’ll supplement. Sleep score dropping? Let’s try magnesium. But when medical superintelligence has arrived, all this data I’ve collected for years will be worth its weight in gold.
My bet is that drag-and-dropping 5–10 years of health data will be literally life-changing for folks disciplined enough to start collecting today.
AI TRAINING
The Rundown: In this guide, you will learn how to use Google Workspace Studio to turn upcoming meetings into automatic prep time and a Gemini-written brief.
Step-by-step:
Go to Google Workspace Studio and create a new flow with Based on a meeting.
Set the trigger to run 10–15 minutes before the meeting so the prep block lands before the call starts.
Add an Ask Gemini step that creates a short prep brief with meeting context, open loops, suggested agenda, and questions to ask.
Add Create a doc next. Use the meeting title variable + “Meeting Prep” for the doc name and the Gemini output for the body.
Add Block time last. Use the meeting title and meeting start time variables, then place the doc link variable in the prep block description.
Add a fake meeting to Google Calendar, click Test workflow, and select that meeting from the dropdown before you turn the flow on. Then confirm the prep block lands before the meeting and the brief link works.
Going further: Start with one recurring meeting type, test the flow, then expand the same Studio setup to client calls, hiring interviews, and exec reviews.
PRESENTED BY INCOGNI
The Rundown: Scammers don’t guess your number — they buy it from data brokers selling your phone, home address, and relatives’ info online. That’s how a random call turns into phishing, impersonation, or worse. Incogni removes your data from these databases and keeps removing it automatically.
With Incogni working in the background, you can:
Keep phone numbers and addresses off broker databases
Shrink exposure to identity theft and impersonation
Let automated removals do the work continuously
Protect your family’s details alongside your own
Try Incogni and get 55% off with code RUNDOWN.
OPENAI

Image source: Noam Shazeer
The Rundown: OpenAI just hired Noam Shazeer away from Google, pulling in a Gemini co-lead whose 2017 transformer work helped shape modern AI — just two years after Google paid $2.7B to bring him back from his own startup, Character.AI.
The details:
Shazeer started at Google in 2000, and co-authored “Attention Is All You Need” in 2017, a transformer paper that helped shape nearly every modern chatbot.
Google reportedly spent $2.7B in 2024 to win Shazeer back from Character AI, the startup he built after Google snubbed his initial ChatGPT-style pitch.
Shazeer was a VP and co-lead on Google’s Gemini, with his expertise helping bring the models near the frontier after falling behind OAI’s ChatGPT.
Why it matters: The talent wars have been quieter than last summer’s chaotic Meta poaching spree, but this is a big move. Where top AI researchers and engineers go is always a signal — and in 2026, the answer has consistently been Anthropic or OpenAI.
QUICK HITS
⚙️ Codex - OpenAI’s agentic coding tool, with new Record & Replay for creating reusable skills
🧠 Brain - Perplexity’s self-improving memory for its Computer agent
🎨 Firefly Studio - Adobe’s upgraded all-in-one platform to generate and edit with AI
📑 Crosby - Agentic law firm for sales teams to speed up time to signature
Anthropic’s Chris Ciauri told reporters during an event in Korea that the company is “very confident” that its Mythos and Fable models will become available again in the “coming days.”
Adobe rolled out new agentic skills for Firefly AI Assistant, also extending its creative agent into public beta across Photoshop, Premiere, Illustrator, InDesign, and Frame.
Databricks launched a slate of new agentic tools at its Data + AI Summit, including LTAP for running AI apps and analytics, and an AI-run customer data platform called CustomerLake.
Former White House AI advisor Dean W. Ball is joining OpenAI to lead Strategic Futures, a new team that will help shape frontier AI policy.
COMMUNITY
Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.
Today’s workflow comes from reader Kristin B. in Los Altos, CA:
“I teach classic Chinese mah-Jong in the Bay Area. The Chinese version includes a two-step scoring process that can feel complicated at first but is essential to learning the game.
With Claude Code, I created a scoring app for my students to check if they correctly scored their hands. It was fun to create the app, and the best feeling was when I showed my scoring app to my son, and he responded with both surprise and respect."
How do you use AI? Tell us here.
Read our last AI newsletter: Inside the deadlock keeping Mythos offline
Read our last Tech newsletter: Xbox’s studio crisis gets bigger
Read our last Robotics newsletter: Meet Eno, the anti-humanoid robot
Today’s AI tool guide: Save hours on meeting prep with Google Gemini
RSVP to next workshop on June 25: Get consultant-grade strategy from AI
That's it for today!
See you soon,
Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown











