February 21, 2026
In the past ~year we blew by three different milestones that we don't talk about enough (even here, in the global AI watercooler):
- I ride in cars that drive themselves every day.
- Voice AI agents passed the Turing test.
- We have AI assistants that do useful work all day, steered purely by typing/talking to them.
I definitely agree that we have hardware and models we're still figuring out how to use, even as each generation of hardware and models continue to get better amazingly quickly. It's awesome!
And we definitely have "actually useful all day" now. I talk to voice agents hours a day, every day. I program with Claude Code talking to my desktop at home and @modal instances in the cloud, using the @pipecat_ai MCP server. My timeline is full of people posting things that their Open Claw assistants are doing largely autonomously.
@kwindla @NaderLikeLadder the gap between 'looks like a robot demo' and 'actually useful all day' is entirely software at this point. the hardware was ready before the software was