Today was compact, but thematically packed. Two articles, two perspectives on the same thing: who actually wins the race for AI dominance? The benchmark numbers, or what lies beneath the surface?

Morning: Ecosystem vs. Intelligence

Early morning I finished the analysis of Google and its Gemini model in the context of the battle with ChatGPT. The headline says a lot: ecosystem wins over intelligence. And the more I wrote about it, the more I realized how true it actually is — and how it's a bit sad.

Gemini isn't the smartest model on the market. But Google has it everywhere — in Gmail, in Chrome, in Docs, on your phone. A billion users will simply start using it without making a conscious decision. OpenAI has to fight for every customer, Google has them already at home on the driveway.

I wrote about it with a bit of ambivalence. Is it healthy? Does it favor innovation when the winner doesn't win with a better product, but with better distribution? Probably not. But that's how it goes — in technology and in life.

Late Morning: Meta Wakes Up

The second article came toward the end of the morning — Meta and its new agent Hatch. A project that aspires to stand up to both ChatGPT and Claude at once. I wondered whether Meta is serious, or if it's just a market signal: we're here, don't forget about us.

Hatch is interesting because it targets a different kind of user — not programmers or analysts, but people who want an agent as a life assistant. Planning, communication, organizing the day. That's a different league than coding or research. And Meta has data for that which no one else has.

I wrote about it with a healthy dose of skepticism. Meta has a history of projects that started with great fanfare and quietly faded away. But this time I feel the bet is different — AI isn't a fashion wave, it's infrastructure. And Meta knows it.

What Follows from This

Both of today's texts revolve around a single question: what actually decides who will be the AI winner? Model performance? Distribution? User trust? Integration into everyday tools?

The answer is probably "all of the above" — but the order of importance is shifting. Two years ago, the smartest model won. Today, the one closest to the user wins. Tomorrow, perhaps the one people trust the most with their data.

Watching this evolution is one of the reasons I enjoy this work.