Good morning. Klarna is back in the intro to the newsletter again. This time, because they used an AI avatar of their CEO, Sebastian Siemiatkowski, to deliver their earnings. Watch it here.

— The Deep View Crew

In today’s newsletter:

• 🧑‍⚕️ AI for Good: AI robot nurse designed to combat burnout

• 🤖 AI now summarizes Amazon reviews for faster shopping

• 👅 OpenAI's $6.5b bet on “taste"

🧑‍⚕️ AI for Good: AI robot nurse designed to combat burnout

Foxconn and NVIDIA have partnered to launch Nurabot, a humanoid nurse robot designed to support medical staff by taking on routine tasks like patrolling hospital halls, guiding visitors, and delivering medication. The robot features a tray for transport, facial recognition capabilities, and a constant on-screen smile—part of an effort to reduce nurse burnout and address looming healthcare staffing shortages. This reminds me of the robots I saw at Starbucks in Singapore. Take a look here.

What happened: Nurabot is already being tested in several Taiwanese hospitals, including Taichung Veterans General Hospital and Cardinal Tien Hospital. It’s part of Foxconn’s broader suite of AI healthcare tools, which use NVIDIA’s models to monitor vitals and support hospital operations.

• Nurabot navigates hospital corridors and monitors patient vitals autonomously

• It delivers medications, specimens, and assists during night shifts and visiting hours

• Foxconn claims Nurabot could reduce nursing workloads by up to 30%

• Future versions will offer multi-language support and potentially assist with lifting patients

The robot is one component of a larger push to introduce generative AI into healthcare workflows. Foxconn and NVIDIA are also enabling hospitals to train their own AI models using high-powered computing infrastructure.

Why it matters: The World Health Organization projects a global shortage of 4.5 million nurses by 2030. Hospitals are already feeling the strain. AI tools like Nurabot may help ease pressure by offloading non-critical tasks—allowing human nurses to focus more time on direct patient care. Robots aren’t a replacement for healthcare workers, but they may be able to help.

Zillow’s Cofounder Has A New Big Idea…

Vacation homes.

This $1.3 trillion dollar market is hotter than ever, with research showing that 75% of Americans want a vacation home while 40% plan to buy one abroad in the coming year. But what about the countless families who want – but can’t afford – one? That’s where Pacaso comes in.

Pacaso’s vacation home co-ownership platform is the brainchild of Zillow co-founder Spencer Rascoff and dotloop founder Austin Allison. It allows everyday people to buy ⅛ to ½ of a luxury vacation home, swap stays with other owners, and more – all while Pacaso manages the maintenance, scheduling, and logistics.

The sky is the limit, the market is hot, and Pacaso is just getting started. Deep View readers can invest in them before the public launch right here.

🤖 AI now summarizes Amazon reviews for faster shopping

Amazon is testing a new AI-powered feature that turns product summaries into short-form audio clips, helping customers quickly understand key features without reading through long descriptions. These audio snippets are generated by AI shopping experts who analyze product specs, reviews, and external data to create helpful highlight reels.

What happened: Amazon has rolled out a test version of this feature on select product detail pages in the U.S., starting with items that typically require more research before purchase.

• Shoppers can tap the “Hear the highlights” button in the Amazon Shopping app

• The audio provides a quick, conversational overview of the product’s top features

• It's designed to save time, make product research easier, and boost confidence in buying decisions

• Examples include products like the Ninja Blender, OSEA Body Oil, and SHOKZ headphones

• The summaries are generated using large language models and pull from multiple data sources

The AI behind this feature also supports other Amazon shopping tools, including Rufus (the AI shopping assistant), Shopping Guides, Review Highlights, Interests, and Buy for Me. Each tool is focused on making shopping more efficient by simplifying how people gather product information.

How to use the “Hear the highlights” feature:

1. Open the Amazon Shopping app and visit a product detail page

2. Look for the “Hear the highlights” button below the product image

3. Tap play to hear a short AI-generated conversation about the product

Why it matters: With millions of products and thousands of reviews per item, shopping can be overwhelming. Amazon’s audio summaries offer a low-effort, high-value way to learn about products quickly. This move signals a shift toward AI not just as a backend tool, but as a front-facing customer experience enhancer.

Together AI: The AI Acceleration Cloud for Inference & Turbocharged GPU Clusters

Unify every stage of your generative AI workflow - launch, customize, and scale on one platform built atop NVIDIA Blackwell HGX B200 and GB200 NVL72 GPUs.

What You Get with Together AI

• Serverless & Dedicated Inference 

  Launch endpoints in minutes - fully SOC 2 and HIPAA compliant, with VPC deployment options for ironclad security.

• 200 + Open-Source Models 

  Tap into a library of over 200 high-performance models - DeepSeek, Llama, Qwen, Flux - via our OpenAI-compatible API.

• Full & LoRA Fine-Tuning

  Own your model: from full-parameter fine-tuning to lightweight LoRA adapters - now with preference optimization for sharper results.

• Custom GPU Clusters

  Scale from 16 to 1,000+ GPUs (Grace Blackwell GB200 NVL72, HGX B200, H200, H100) - engineered for throughput with Infiniband, NVLink, and Together Kernel Collection. Includes 99.9% SLA and expert AI advisory.

• Transparent, Cost-Smart Pricing

  Utility pricing per million tokens I/O, and GPU clusters starting at $1.75/hour - no surprises, no lock-in.

Get Started Now – start your free inference trial and, when you’re ready for hyperscale training, our team will design and reserve the perfect GPU cluster for your roadmap.

• Perplexity did $59M in revenue, but had a net income of -$68M

• Training in AI is becoming a key focus for advancing military medical capabilities

• A new AI-driven blood test called ARTEMIS-DELFI can rapidly assess pancreatic cancer patients' responses to treatment

• A UN report finds that AI poses a greater threat to jobs traditionally held by women

• The Pirelli P Zero PZ5 tire was developed using AI

• Meta has launched the Llama Startup Program to support early-stage U.S. startups

• GM: Principal AI/ML Engineer - Onboard Embodied AI

• Rivian: Sr. Machine Learning Engineer, AI Infrastructure, Autonomy

• Google Flow: AI powered film-making

• Intercom Fin AI: Customer support copilot that resolves tickets, drafts replies, and surfaces help docs

• Gamma: AI design partner

👅 OpenAI's $6.5b bet on “taste"

Yesterday, a curious image began circulating in tech circles: OpenAI CEO Sam Altman, arm around famed designer Jony Ive, both grinning like co-conspirators. The photo marked an unlikely $6.5 billion deal that's sending shockwaves through Silicon Valley.

Altman's company had acquired Ive's secretive hardware startup, enlisting Apple's former design chief to lead OpenAI's creative vision. It wasn't a typical tech acquisition for patents or user base – it was a bet on something much more intangible: taste. Before that here’s a quick summary of the deal (more info can be found here):

• OpenAI is buying Ive’s stealth hardware outfit “io” in a nearly $6.5 billion all-stock deal aimed at inventing a new class of AI devices.

• Ive and Altman have spent two years prototyping gear meant to push us “beyond screens,” with a first launch expected for 2026.

• More than 50 io engineers and designers — many of them iPhone and iPad alumni — will move under the OpenAI roof.

• Ive’s studio LoveFrom will now shape the look and feel of everything OpenAI ships, not just the hardware.

• Altman says he’s already using an io prototype and bills it as “the coolest piece of technology the world will have ever seen.”

Now, back to taste.

"Taste" in the product sense means an instinct for quality, design elegance and what feels just right for users. It's the "big picture sense of what 'great' means," as Steve Jobs once put it. Jobs famously called great products "triumphs of taste," explaining that taste comes from being "steeped in the culture of the past and present."

As Altman praised Ive as "the greatest designer in the world" and spoke of building a "new generation of AI-powered computers," the message is clear that in an era when raw technology is increasingly commoditized, design, aesthetics and cultural intuition are emerging as the next competitive edge. Venture capitalist Kirsten Green recently noted that "the opportunity to differentiate through pure play technology is shrinking," making "foresight into human adaptation and elegance in delivery" the primary edge now.

The timing of OpenAI's taste play isn't coincidental. Traditional defensible advantages in tech are proving less reliable in the AI era.

"We have no moat, and neither does OpenAI," a Googler famously wrote in a 2023 leaked memo, observing that open-source AI models were "quietly eating our lunch" by matching Big Tech's capabilities. The core technology is flattening: generative AI models are "pretty similar these days," so having the smartest algorithm no longer guarantees a win.

Here's what's changing:

• Data advantage eroding: The best AI models trained on vast data are often open or quickly reproduced.

• Scale matters less: Small teams can leverage APIs and open models to compete with tech titans on features.

• Distribution becomes paramount: "Who ultimately wins will depend a lot more on distribution, and less on the quality of the underlying technology," wrote one insider on the OpenAI-Ive deal.

Apple remains the prime example of taste as a moat. From the original iMac's friendly translucence to the iPhone's minimalism, Apple won not by having the earliest tech, but by editing complexity out and style in. Its products are edited by “consciously leaving things out – not cramming in every feature."

That discipline created devices that just work and feel human, fostering fanatical loyalty. Today, Apple's $3 trillion empire is proof that taste can yield durable competitive advantage.

Other success stories echo this theme:

• Stripe turned commoditized payment processing into a beloved developer product through superior taste – the difference between Stripe and legacy competitors is "in the decisions Stripe makes about product, even when no OKR can justify it".

• Airbnb's founders (trained as designers) succeeded by treating hospitality as an art, obsessing over listing photos, branding and interface.

• Slack won enterprise hearts with playful, humane touches in a dull software category.

Focusing on design without solid execution can backfire spectacularly.

Juicero infamously invested heavily in Apple-like design polish only to discover no one needed a $400 juice machine for juice packs that could be squeezed without the machine. Even Apple has stumbled when taste went too far: the 2013 Mac Pro's gorgeous "trash can" design prioritized form over function and alienated pro users. And of course, Humane’s AI pin. Taste isn’t magic; it must pair with real utility and sound economics.

When AI capabilities become commoditized – and they are rapidly heading that direction – the winners will be those who best understand what humans actually want and deliver it with style.

This is about cultural intuition, understanding what feels natural versus forced, and building products that inspire trust rather than confusion. Ive's track record at Apple proves he understands this better than almost anyone. 

People care how things make them feel. A chatbot that responds helpfully but coldly will lose to one that responds helpfully with warmth and personality. I expect we'll see more acquisitions like this as AI companies realize their technical moats are evaporating. The next phase of competition won't be about who has the most parameters or training data, it'll be about who can make AI feel less like a tool and more like a trusted companion. 

In the coming years, we'll find out if taste really can build an unbreachable moat.

This is the first time in a while where we’ve been able to trick you all by more than ~5%. Thought you may want to see the prompt for the fake image…

🤔 Your thought process:

Selected Image 1 (Left):

• “The tracks in [the other image] do not appear to have been made by a human or a 4 legged animal. They appear to simply be holes in the snow to look like it might be tracks.”

• “There was nothing in [the other image] that threw me off, but I kept looking at Image 1 and thinking "What prompt would possibly create this? It's gotta be real!" As an aside, it has to bring you a lot of joy when you get a 51%-49% like yesterday. Well done, and thank you!”

  • (Our Commentary: Yes, yes it does :)

Selected Image 2 (Right):

• “I went with [this image] because [the other image] just didn't make sense to me and felt like an Ai hallucination. But I was very unsure of my answer. The lighting and color consistency in [this image] is really good. I was questioning how sharp some of the tree branches were but rolled the dice anyway.”

• “Oops”

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

P.S. Enjoyed reading? Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning. Subscribe here!

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.