AI images meet ultrasound spas. .
Free AFIRE Guide | AI Academy | Advertise | AI Mastery A-Z
Plus: 6 Methods that Anthropic Team Used to Boost Claude Output Quality (Full Breakdown)
Midjourney just made its most unusual move yet: a full-body scanner built with Butterfly Network, designed to deliver 60-second ultrasound scans inside future wellness spas.
What’s on FIRE 🔥
IN PARTNERSHIP WITH HUBSPOT
LLM traffic converts 3× better than Google search
58% of buyers now start their research in ChatGPT or Gemini, not Google. Most startups aren’t showing up there yet.
The ones that are get cited by the AI tools their buyers, investors, and future hires already use. And they convert at 3×.
Download the free AEO Playbook for Startups from HubSpot and get the exact steps to start showing up. Five minutes to read.
AI INSIGHTS
💧 Midjourney’s Unprecedented Move to Medical Hardware
Midjourney, the company we all know for generative AI images, just announced the “Midjourney Scanner“, a medical hardware device focused on full-body imaging, with plans to deploy it in consumer wellness spas starting in 2027.
-
The core of the hardware is an underwater ultrasonic sensor system. This scanner physically lowers the user into water, passing them through a ring of ultrasound sensors. The goal is to generate a complete full-body scan in just 60 seconds.
-
The device is designed to rival the detailed imaging of a standard MRI, but at a fraction of the time. To build the machine, they partnered directly with the ultrasound-chip manufacturer Butterfly Network.
-
Midjourney is framing this as a wellness experience. They are planning to open the first “Midjourney Spa” in San Francisco’s Union Square in 2027.
While the pivot from software to medical hardware is a massive leap, the strategy here is clear. Midjourney is attempting to completely reframe how we approach preventative health tracking.
Transitioning from AI model development to manufacturing physical, sensor-rich devices comes with a completely different set of logistical challenges. However, if they can successfully execute on the 60-second ultrasound technology, it could offer a highly accessible alternative to traditional imaging.
PRESENTED BY DEEL
Hiring in 8 countries shouldn’t require 8 different processes
This guide from Deel breaks down how to build one global hiring system. You’ll learn about assessment frameworks that scale, how to do headcount planning across regions, and even intake processes that work everywhere. As HR pros know, hiring in one country is hard enough. So let this free global hiring guide give you the tools you need to avoid global hiring headaches.
AI SOURCES FROM AI FIRE
1. Free Guide: How to Use Gemini + NotebookLM to Keep Your AI Accurate Every Time. Gemini NotebookLM pairs live research with source backed answers in one chat. See the setup steps & prompts built on real documents.
2. 6 Methods that Anthropic Team Used to Boost Claude Output Quality (Full Breakdown). These methods are buried in their workflow, but you can see the full breakdown and apply them to your own AI sessions to improve accuracy!
3. Gemini Plus vs ChatGPT Go: Which $5–$8 AI Plan Actually Fits You After 30 Days. Full budget AI plan breakdown for 2026. Compare pricing, usage limits, memory, Canvas, image quality, Google tools, third-party apps, and who should choose each plan before paying monthly.
TODAY IN AI
AI HIGHLIGHTS
👨💻 Anthropic studied ~400K Claude Code sessions and found the average task value grew 27% in six months. People still decide what to build, while Claude does most of the execution.
🩺 OpenAI says GPT-5.5 Instant now matches its frontier health models on tough medical tests. It was shaped by physician-led evals and is available to free ChatGPT users.
🧠 OpenAI just hired Noam Shazeer from Google, a Gemini co-lead and co-author of the 2017 Attention Is All You Need paper. Big talent-war signal after Google reportedly paid $2.7B to bring him back from Character.AI.
👀 Anthropic says Mythos and Fable 5 could return “in the coming days” after being blocked under U.S. security restrictions. The models are tied to Project Glasswing, Anthropic’s high-security AI access program.
🏆 John Jumper, the Nobel-winning AlphaFold co-creator, is leaving Google DeepMind for Anthropic. Another huge AI talent move, just days after OpenAI hired Noam Shazeer from Google.
💰 Big AI Fundraising: Baseten is reportedly nearing a $1.5 billion raise at a $13 billion valuation, up 160% in 5 months, as investors rush into AI inference infrastructure.
HOT PAPERS OF THE WEEK
1/ World models may get deeper without getting much bigger
Looped World Models from FaceMind Research Asia introduces LoopWM, a looped architecture for world simulation. Instead of making the model larger, it reuses the same transformer block to refine hidden world states again and again. Big shift: world models like those behind OpenAI Sora and Google DeepMind Genie may get longer, more stable simulations through iterative depth, not just more parameters.
2/ A 3B model is matching much larger reasoning systems
VibeThinker-3B from Sina Weibo shows how far small language models can go on verifiable reasoning. With only 3B parameters, it reaches 94.3 on AIME26, 80.2 Pass@1 on LiveCodeBench v6, and strong results on unseen LeetCode contests. Key result: it matches or beats much larger models like DeepSeek V3.2, GLM-5, and Gemini 3 Pro on several reasoning tasks.
3/ Microsoft wants coding agents to search codebases with less waste
FastContext from Microsoft and Shanghai Jiao Tong University introduces a dedicated repository exploration agent for coding tasks. It searches files, gathers evidence, and returns only the useful paths and line ranges, instead of filling the main agent’s context with noise. Big impact: coding agents like Mini-SWE-Agent can solve more tasks while using up to 60% fewer tokens.
NEW EMPOWERED AI TOOLS
-
💻 Claude Code Artifacts lets you preview in-progress Claude Code work as a live interactive artifact and share it with your team.
-
📱 Zernio WhatsApp API gives developers one API for WhatsApp messaging, broadcasts, calls, AI agents, chatbots, and MCP workflows.
-
🩺 Midjourney Scanner claims to offer 60-second ultrasound-based body scans using gentle sound waves, with no radiation and repeatable health tracking.
-
🔥 Firecrawl Research Index gives AI agents access to 3M+ arXiv papers and GitHub research artifacts, refreshed daily for AI/ML research.
AI BREAKTHROUGH
📊 Anthropic Study: Domain Knowledge Beats Coding Skills in Claude Code
Anthropic released some fascinating data after analyzing 400,000 Claude Code sessions. So, your actual expertise in your field matters significantly more than your raw coding ability when getting results from AI agents.
-
Humans act as the architects, driving roughly 70% of the decisions, while Claude handles the heavy lifting, executing about 80% of the choices.
-
Beginners average 5 actions and 600 words per prompt, whereas experts pull out 12 actions and up to 3,200 words.
-
Verified success jumps to 28–33% for intermediate and advanced users, which is more than double the 15% success rate for novices.
-
Professionals with zero coding in their job title, like lawyers, managers, and scientists. nearly matched actual software engineers on coding tasks, finishing within just 7% points of them.
Similar to recent findings showing that agents push users toward harder, cross-disciplinary tasks rather than just speeding up simple ones, this Anthropic study confirms a clear trend.
The true value of an AI agent is no longer capped by the model itself, but rather by how deeply the user understands the actual job they are trying to get done.
We read your emails, comments, and poll replies daily
Hit reply and say Hello – we’d love to hear from you!
Like what you’re reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Leave a Reply