Probably one of the coolest interviews of the year: Tibo on Matthew Berman’s show.

And one of his first statements gives me a lot of hope: in the foreseeable future, speeds like 750 tokens per second will be the default, not the exception.

When you consider how much time we spend watching agents work, and how that idle time ultimately costs both time and money, speed becomes increasingly important.

The question of the near future isn’t just how intelligent AI is, but how fast that intelligence can operate.