Agent Autonomy

“By narrowing the domain of autonomy, you can have more robust guardrails and even with current models actually achieve something that's effective enough to be commercially viable today.”