Scale & Strategy
together with
This is Scale & Strategy, the newsletter that gives you the cheat sheet before the test.
Here’s what we’ve got for you today:
- Google’s AI glasses might finally crack the category
- Anthropic lands OpenAI co-founder Andrej Karpathy
Google’s AI glasses might finally crack the category
AI glasses have spent the last two years stuck somewhere between “cool demo” and “guy at the coffee shop secretly recording everyone.” Google may finally have the stack to push them into something mainstream.
At I/O this week, Google showed off its first polished AI smart glasses built with Samsung, Gentle Monster, and Warby Parker. And unlike most wearables launches, this one actually feels strategically coherent instead of five companies stapling buzzwords together and praying investors stay conscious.
The pitch is simple: put Gemini directly into your field of view and make AI ambient instead of app-based.
The glasses come in multiple styles and lean heavily on Android integration. That matters more than most people realize. Hardware alone was never the bottleneck for smart glasses. Ecosystem integration was. Meta proved people will wear cameras on their face if the product is useful enough. Google is now trying to turn that into a real computing platform.
The feature set is deeper than previous attempts:
Voice-guided navigation with directions appearing directly in the lens instead of forcing users to constantly check their phone.
Live translation that matches the speaker’s voice while also translating text in your surroundings in real time.
Notification summaries, recommendations, ordering, and assistant actions powered through Gemini.
Hands-free photo capture and contextual AI assistance based on what the user is actively seeing.
The key difference is the display layer. Meta’s Ray-Bans succeeded largely because they were good wearable cameras with decent AI attached. Google’s in-lens interface changes the interaction model entirely. You’re no longer asking AI something abstractly. The system sees what you see and responds contextually inside your environment.
That’s a much bigger shift than adding ChatGPT to another gadget nobody charges after week two. Humans truly looked at smartphones destroying their attention spans and decided the next logical step was placing notifications directly onto their eyeballs. Tremendous species behavior.
Strategically, this is also one of the clearest examples yet of why Google still has structural advantages in AI despite OpenAI dominating mindshare. Gemini tied into Android, Maps, Search, YouTube, Gmail, and Samsung hardware creates a distribution machine competitors simply do not have.
The partnership setup is smart too. Samsung handles hardware. Google handles AI and ecosystem. Gentle Monster and Warby Parker handle the part tech companies historically fail at: making wearable products people don’t feel ridiculous wearing in public.
That last part matters more than model benchmarks.
The remaining hurdles are predictable. Privacy concerns will be nonstop. Social acceptance still isn’t solved. Battery life, weight, heat, and comfort all become brutally important when something sits on your face all day. But based on early hands-on impressions, Google seems much closer on comfort than previous generations.
The bigger takeaway is that AI hardware is finally moving beyond “chatbot in a rectangle.” Phones won’t disappear anytime soon, but ambient AI interfaces are starting to look inevitable. And right now, Google suddenly looks further ahead than most people expected.
Risk and regulations are ramping up, and customers expect proof of security just to do business. That’s why Vanta is a game changer.
Vanta automates your compliance and security processes in one AI-powered platform.
Whether you’re prepping for SOC 2, or running an enterprise GRC program, Vanta keeps you secure and your deals moving.
Companies like Ramp and Writer spend 82% less time on audits with Vanta. Faster compliance—and more time for growth.
Check out the on-demand demo to learn more. – trusted by top startups like Cursor, Linear, and Replit.
Anthropic lands OpenAI co-founder Andrej Karpathy
Andrej Karpathy is officially joining Anthropic, giving the company one of the most respected researchers in modern AI and adding even more fuel to the increasingly absurd frontier-lab talent war.
Karpathy announced the move on X, saying he’ll join Anthropic’s pre-training team under Nick Joseph while also helping build a new internal initiative focused on using Claude to automate parts of Anthropic’s own AI training pipeline.
Which is a sentence that should probably make every researcher in the industry slightly nervous.
Karpathy’s résumé is about as stacked as it gets:
Co-founded OpenAI in 2015.
Led Autopilot at Tesla during the company’s most aggressive AI scaling years.
Briefly returned to OpenAI before leaving again in 2024 to focus on AI education and independent research.
Became arguably the single most trusted technical educator in the AI ecosystem while the rest of the industry discovered podcast microphones and started hallucinating about AGI timelines.
The important part here isn’t just that Anthropic hired a star researcher. Frontier labs hire elite talent constantly. The interesting part is what Karpathy is reportedly working on.
Every major AI lab is now racing toward the same destination: systems that help build better systems. Training infrastructure, evaluation loops, data pipelines, experimentation, debugging, research assistance. The industry is increasingly trying to automate the process of AI development itself.
That’s where the leverage is.
The bottleneck in frontier AI is no longer just model intelligence. It’s research velocity. The lab that can compress iteration cycles fastest gains an enormous advantage. If Claude can meaningfully assist in designing experiments, improving training efficiency, surfacing failures, or accelerating model development internally, that compounds hard.
Anthropic has been quietly leaning into this direction for months. Claude Code, internal tooling, long-context workflows, agentic systems, research automation. Karpathy joining specifically to push on training-pipeline automation makes the strategy impossible to miss now.
It’s also a symbolic loss for OpenAI whether people admit it or not. Karpathy is one of the few figures broadly respected across essentially every faction of the AI world: researchers, founders, engineers, open-source communities, academics. Getting him back into the frontier-lab game at all was significant. Getting him onto Anthropic’s side is even more so.
And underneath all of this is the larger pattern: the labs are increasingly trying to remove humans from the slowest parts of AI development. Which sounds efficient right up until the models start reviewing their own homework and promoting themselves to senior engineer. Small detail. Humanity loves deploying systems first and figuring out governance sometime around phase six of the disaster movie.
Still, from a pure talent and strategy perspective, this is a huge move for Anthropic.
Was this email forwarded to you?
That’s it for today and as always It would mean the world to us if you help us grow and share this newsletter with other operators.
Our mission is to help as many business operators as possible, and we would love for you to help us with that mission!
