Hey there, 

This week, two of the biggest AI companies shipped features that change how we use AI. Instead of you opening an app and asking a question, the AI now comes to you. Google also announced a new frontier model, just two weeks after Gemini 4 Argon. Let’s jump into everything that launched this week.

Meet Gemini 4 Carbon, Google’s Strongest Model Yet

On Wednesday, Google announced Gemini 4 Carbon, which is its strongest model yet. It currently ranks first on the Artificial Analysis Intelligence Index, and Google says its teams have seen it continuously working for days on complex engineering problems. Carbon isn’t a faster or cheaper version of Argon. It’s a full upgrade, and Google’s new flagship model has set new records on tests for finding and fixing security flaws. But the most interesting part for everyday users is honesty. Google says Carbon makes about half as many unsupported claims as Argon, and when evidence is missing, it points out what’s missing instead of guessing. Like Argon, it’s going to a small group of trusted users and cybersecurity experts first, with access for developers, businesses, and everyone else coming “as soon as possible.”

Claude Duet Messages You First

Anthropic opened a research preview of Claude Duet, a new app where Claude works as a proactive agent. Each morning, it sends you a short brief on your connected apps, then checks in throughout the day and messages you only when something needs your attention. In one example from early testers, it noticed a client call was scheduled right before a user’s flight and offered to move it. Before it sends a message, makes a payment, signs a document, or deletes anything, it shows you exactly what it plans to do and waits for your approval. It’s available by waitlist for Pro and Max subscribers in the U.S.

ChatGPT Can Now Text You

OpenAI is moving in the same direction. ChatGPT’s “dots,” the always-on agents OpenAI announced at DevDay, can now be reached by text and they’ll reach out to you with updates on their own. Each dot has its own cloud computer, so it can keep working on a task while you’re away and send the results back for you to review. Your dot can only text you, and OpenAI says this doesn’t give it access to your other conversations. It’s starting as a beta for Pro users in the U.S.

My Thoughts

It’s hard not to notice how fast Google is moving right now. Argon came out just two weeks ago, and most people still can’t use it without a paid plan, yet Google is already announcing the next model. That pace is exciting, but it also makes it harder to keep track of which model is worth using. Still, the bigger story for me this week is how AI is trying to fit into our everyday lives. With Claude Duet and ChatGPT’s dots, AI is moving from an app you open into something that checks in on its own, even through regular text messages. That could save a lot of time, but only if it gets good at deciding what’s actually worth interrupting you for. So I’d want to test these for a while before deciding whether they’re genuinely helpful or just another stream of notifications.