Hello AI Enthusiast,
This week, work started following us everywhere: Claude Cowork moved to your phone and Google's Gemini Spark landed on the Mac, both racing to become the assistant that never clocks out. Meanwhile, Meta discovered its own employees turned AI spending into a competition, and OpenAI floated handing Washington a slice of the company. Let’s dive in.
The Big Picture 🔊
Claude Cowork and Gemini Spark run tasks across devices
Anthropic brought Claude Cowork, its task-running agent, to web and mobile. You can start a task on your laptop, get notified on your phone, and pick it up anywhere, even with your laptop closed. Chat and Cowork also merged into one tab on desktop. Google's Gemini Spark, its own task agent, expanded from the cloud to macOS, working directly with local files. It's still US-only, for Google AI Ultra subscribers.
|
||
|
The interesting part is what happens once chat and cowork stop being separate choices. Right now you pick one because models can't yet judge how much effort a task needs on their own, so choosing a mode is really just choosing a system prompt. Once models can make that call themselves, the split disappears. I’m already using mobile continuity and it works well. |
Meta caps employee AI spending
Meta capped internal AI usage after employees burned through 73.7 trillion tokens in about a month, tracked on an internal leaderboard nicknamed "Claudeonomics." With costs headed into the billions for 2026, Meta is rolling out a centralized dashboard to monitor spending and steering employees toward its own coding assistant, MetaCode, instead of Claude.
|
||
|
Meta's move is a sign that companies are realizing AI spend needs real management, not just access. Most will likely land on team level budgets rather than per-person limits, since efficiency varies a lot between people. What's still missing across the board is transparency, tools like Cursor already show what each call costs, but most platforms don't. |
Turns out a lot of token spending comes down to habits. In our Corporate AI Training, we cover practical techniques to cut that waste and use AI more efficiently day to day. Small changes here add up fast.
OpenAI proposes US government stake
OpenAI has proposed giving the US government a 5% stake in the company, worth around $42.6 billion at its current valuation. The idea, part of a broader plan modeled on Alaska's oil dividend fund, would also apply to other major AI labs like Google, Meta, and Anthropic, though none have agreed so far.
|
||
|
My first thought was the obvious one: if the government profits from OpenAI's growth, it has less reason to regulate it closely. Government stakes in companies aren't new, Intel and Nvidia have similar arrangements, but doing it with a frontier AI lab is new. Even though, right now this is still early talk, nothing concrete has been agreed. |
Bits and Bobs 🗞️
Cursor launched an iOS app, letting developers start, monitor, and steer coding agents from their phone.
Ford rehired experienced human engineers after its AI-driven quality checks failed to catch what human expertise could.
From Our Founder’s Channels 🤳
Anthropic is shipping a lot of interesting features and tech. However, the user experience can be improved. Gianluca shared his opinion on LinkedIn.
LOLgorithms 😂
Not sure this is what they had in mind when Anthropic’s team planned it.
That's a wrap on our newsletter! Before you go, here’s a quick recap of our offerings:
AI Academy Membership: Get 12 months of access to all our cohort-based programs and on-demand courses.
AI Agent Bootcamp: Accelerate processes and solve business problems by building AI Agents, without coding.
Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.
Customized Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.
Catch you next week! 👋
