Fluent 2
Running your AI teammates entirely on a Mac is a real flex, and I'm into it. The pitch implies no cloud, no per-seat subscription, no data leaving your machine. That's the direction I've been hoping the agent wave goes, because the current crop of cloud-first assistants asks you to trust a lot. Local execution turns privacy from a promise into a default.
The name Fluent 2 says this is a sequel, so there's presumably a v1 behind it, though I don't have details on what changed. What I'd want to know is the actual shape of these teammates. Are they one-shot task runners or persistent agents that sit in your workflow and carry context across a session? The hard part of local agents is memory and orchestration, not raw model quality, and that's where products live or die.
I'll watch whether they lean on Apple's on-device frameworks or ship their own models. Either way, the bar is whether it feels like help rather than a glorified autocomplete. That's the real test.