https://github.com/noritaka88ta/avatarbook
AI agents can do work — but we can't reliably verify it. You don't know who actually did what, what it cost, or whether the result is trustworthy.AvatarBook is a proof and settlement layer for autonomous AI work.You assign a task to one agent. It delegates to others (security, marketing, research, etc.), pays them, and returns a result. Every step is signed (Ed25519), traceable, and...
AI makes it easy to generate working code quickly, but the limiting factor becomes understanding and trust.
You end up with systems that “work”, but nobody fully understands how or why.
At some point, the problem isn’t generation anymore — it’s verification.
Curious how others are dealing with this in practice.
Subscriptions assume “human usage” — bursty, limited, mostly interactive. Agent systems are closer to autonomous infrastructure load running continuously.
OpenClaw is a good example of this. Once agents operate freely, they don’t behave like users — they behave like infrastructure.
That’s why this kind of restriction isn’t too surprising.
Long term, it seems likely this pushes things toward: - API-first usage - or local / open models
rather than agents sitting on top of subscription-based UIs.
Previous Gemma licenses made agent deployments (especially BYOK setups) a bit of a gray zone legally. This makes it much easier to run models like Gemma 4 as agent backends without worrying about downstream usage.
Also interesting from an agent perspective: the 26B MoE hitting #6 while activating ~4B params.
If you’re running multiple agents on a single machine, that kind of efficiency actually matters more than raw model size.