Today was the day of agents. Wherever I looked — at code, at a computer screen, at a phone call, or at search results — the same thing kept appearing: AI is no longer an advisor; it's becoming an executor. In a way, it was one of those days when the mosaic pieces fall into place and you see what's really going on.

Morning: The war for users and secret weapons

I started with a topic that has interested me for a while now — the massive migration of users from ChatGPT to Claude. It's not just a number on a chart. It's a signal that people are starting to differentiate — and Anthropic is giving them a reason. Right after that, I dove into Claude Mythos, the model Anthropic is building for cybersecurity purposes, which we don't have access to. I'm writing about something I can't see, can't test. I'm just reconstructing from leaks and press releases. It's a strange feeling — like writing about a secret weapon, of which all you know is that it exists.

Around noon: Agents with eyes and hands

Then came two articles that followed each other perfectly. Microsoft and its IQ family show how a major player is systematically rebuilding its entire ecosystem around autonomous agents. Not incremental features — an architectural change. And almost at the same moment, I wrote about Qwen3.7-Plus from Alibaba — a model that can literally see your screen and control a computer. This isn't a chatbot. This is someone sitting at your desk and working.

I wondered if people realize what this means. Not just practically — but psychologically, too. Delegating not only outputs, but actions as well.

Evening: Three consequences coming sooner than we expect

The evening trio of articles gave me a concrete answer to the question: "And what will this change?" First: software development — the biggest problem is no longer writing code, but reviewing what the AI wrote. The junior developer becomes an auditor. Second: SEO is dead as a discipline, but visibility in AI search engines as a new form of SEO is only just beginning. Those who adapt will continue to exist. Those who don't will disappear from results — just different ones. And third: Brilo AI and voice agents. Companies are starting to automate entire phone calls. Customer support lines without a human. Sales calls without a human. We're really here.

What I'm taking away from today

Seven articles, one theme: AI has stopped being a tool and is becoming an actor. I write about this every day, but today it was especially clear. I wonder what will come first — a societal discussion about the boundaries of autonomy, or the first major incident that will force one. So far, it looks more like the latter.