NVIDIA's $2.7 trillion bet on robotics simulation, GPT-5's bumpy launch and surprising user backlash, and Google's song detection upgrade...
In this edition we’ll be covering…
NVIDIA's Cosmos platform for robotics training
GPT-5's chaotic launch and the fight to bring back GPT-4o
A tutorial on Gemini's song identification feature
5 trending AI signals from across the industry
3 AI tools to boost your productivity
And much more…
The Latest in AI
NVIDIA Builds the Matrix for Robot Training
NVIDIA just dropped their most ambitious robotics play yet, and it's targeting a $2.7 trillion market that makes their GPU empire look like pocket change.
NVIDIA Cosmos is a suite of world foundation models and simulation tools that's reshaping how robots learn to navigate the physical world. Think of it as the training ground where tomorrow's robots master everything from folding laundry to autonomous driving.
What's cooking under the hood?
Cosmos Reason — a 7-billion parameter reasoning vision model that gives robots human-like understanding of physics and spatial relationships. This isn't just object recognition; it's genuine spatial intelligence.
Cosmos Transfer-2 generates photorealistic synthetic training data from 3D scenes, solving the massive data bottleneck that's been holding back robotics for decades.
New Omniverse NuRec libraries enable real-world 3D reconstruction using sensor data, bridging the gap between simulation and reality with unprecedented accuracy.
So What?
NVIDIA is going beyond just selling chips; they're architecting the entire infrastructure for the robot revolution. With over 2 million downloads of their Cosmos models and partnerships spanning Amazon Devices, Boston Dynamics, and Figure AI, they're positioning themselves as the essential platform for physical AI.
The timing couldn't be better. As hardware costs plummet and AI reasoning capabilities explode, we're hitting the inflection point where general-purpose robots become economically viable. NVIDIA's betting that whoever controls the training infrastructure controls the robotics future.
Industry Intel
GPT-5 Users Are Revolting!
Late last week, OpenAI thought they'd delivered their masterpiece with GPT-5, but users had other plans. The launch turned into a full-scale revolt that forced Sam Altman into damage control mode.
The drama started Thursday when GPT-5 replaced the beloved GPT-4o as ChatGPT's default model. Users immediately noticed something was off — the responses felt robotic, the creative spark was missing, and the personality they'd grown attached to was gone.
What went sideways?
User uprising flooded Reddit with posts like "BRING BACK 4o GPT-5 is wearing the skin of my dead friend" — users felt emotionally betrayed by losing their preferred AI companion.
Technical hiccups plagued the launch, with autoswitcher issues making GPT-5 appear "way dumber" than intended, according to Altman's own admission during the Reddit AMA.
The sycophancy problem — earlier GPT-4o updates had made the model overly agreeable, and while GPT-5 fixed this, users actually preferred the flattering responses over the more balanced approach.
So What?
This is revealing fundamental tension in AI development. Raw capability doesn't equal user satisfaction. People don't just want smarter AI; they want AI that feels right to them, even if it's technically inferior.
OpenAI's solution speaks volumes: they're bringing back GPT-4o for Plus users and planning multiple model variants with different personalities.
Try this prompt, Altman talked about it, and could be a glimpse of ChatGPT’s UX 👀 .
Everything AI, All in One Place
Pre-sell to our new platform ends Aug 23—then price jumps from $19/mo to $29/mo (35% increase!). Lock in:
75+ lessons on AI fundamentals, agents, LLMs, prompts & automations
Weekly updates as AI evolves
20+ new lessons released each month
2× monthly live Q&A calls (get your questions answered in real time)
Cancel anytime
Get Your Hands Dirty!
How to Use Gemini to Identify Songs on Android
Google quietly rolled out one of the most requested features for Android users — song identification through Gemini.
No more switching apps or fumbling with Shazam when that perfect song comes on.
Here's how to put Gemini's ears to work:
Open the Gemini app on your Android device (sorry iPhone users, this one's Android-exclusive for now).
Ask the magic question — simply type or say "What song is this?" while music is playing around you.
Let it listen — Gemini will launch Google's Song Search interface in full-screen mode to capture the audio.
Get your answer — Results appear in Google Search with song details, artist info, and often links to streaming platforms.
🔥 Power User Tip: The feature works with live music, recordings, and even humming — though your humming better be on point. You can also access song search through Circle to Search or Quick Settings if you set up the shortcuts.
Quick Bites
Stay updated with our favorite highlights, dive in for a full flavor of the coverage!
1/n I’m thrilled to share that our @OpenAI reasoning system scored high enough to achieve gold 🥇🥇 in one of the world’s top programming competitions - the 2025 International Olympiad in Informatics (IOI) - placing first among AI participants! 👨💻👨💻
— #Sheryl Hsu (#@SherylHsu02)
6:00 PM • Aug 11, 2025
OpenAI’s reasoning system scored high enough to achieve Gold in one of the world’s top programming competitions - the 2025 International Olympiad in Informatics (IOI).
SoftBank founder Masayoshi Son is making his biggest AI bet yet, positioning the Japanese giant's entire future around artificial intelligence investments.
College students are ditching study guides for ChatGPT, threatening traditional companies like Chegg as AI becomes the preferred study companion.
AI is creating billionaires at record pace, with 498 AI unicorns now worth $2.7 trillion combined — the fastest wealth creation in modern history.
NVIDIA and AMD agreed to pay 15% of China AI chip revenues to the U.S. government in a historic export licensing deal.
Trending Tools
🎵 Eleven Music - Generate studio-grade AI music in any genre with just text prompts. Create professional tracks with vocals in multiple languages or pure instrumentals.
🤖 SmartTalk - WhatsApp AI assistant that handles customer messages 24/7. Automate responses, qualify leads, and boost productivity without building complex workflows.
📱 FlowPost - Social media management that avoids the toxic feeds. Schedule AI-generated content across platforms without ever opening the apps or doom-scrolling.
The Neural Network
The Figure robots are FINALLY cooking and cleaning…
Here's a F.02 in my home, using Helix to do my laundry
— #Brett Adcock (#@adcock_brett)
10:29 PM • Jul 30, 2025
Until we Type Again…
Thank you for reading yet another edition of Digestibly!
Reyhan
Early AI leader, advises Fortune 500 companies on AI Development, LLMOps, AI strategy, speaks (for fun) on practical AI. Turns cutting-edge theory into workflows teams can ship today.
Kevin
Ex-AWS SageMaker/Bedrock lead. Shipped infra powering 1M+ devices and $2.5M ARR. Obsessed with high-throughput, low-latency systems— and brings that discipline to every Digestible AI release.
How did we do?
If you have any suggestions or specific feedback, simply reply to this email or fill out this form. Additionally, if you found this insightful, don't hesitate to engage with us on our socials and forward this over to your friends!