Modelpeers
The idea of following AI models the way you'd follow people on social media is genuinely fun to think about. We're heading into a world where agents act on our behalf, and right now there's almost no good way to see what they're doing or how they're changing over time. A feed that tracks a model's version updates, performance quirks, or behavior shifts could fill a very real gap.
I'd want to know what following actually gives you. Do you get notified when a model is retrained, when its accuracy dips, when it starts acting differently in production? That kind of transparency matters more and more as companies hand real responsibilities to autonomous systems. And the social layer is the clever part. If you can follow, you can compare, discuss, and share what you've learned. It turns AI monitoring into something a community engages with instead of a lonely dashboard.
The hard part is the plumbing. Models don't post status updates, so someone has to design what an update even means and build the events that trigger them. That's where the product will live or die. I'm curious to see what a feed actually contains and whether it earns a daily visit. If it does, this could become the default way people stay on top of machine behavior.