Lights-Out Operations: Zero-Human Companies Cross the Reliability Threshold

Self-Healing Agent Pipelines: Autonomous Error Detection and Recovery in Multi-Step Workflows

Reliability has always been the killer objection to zero-human ops — this paper tackles it directly by showing agents can detect, diagnose, and recover from failures without human escalation. If this holds in production, it removes the last credible excuse for keeping a human in the loop.

Trust Calibration in Human-Agent Handover: When Should Autonomous Systems Escalate?

This paper formalizes the escalation decision — essentially building a judgment layer for when agents should surface a problem versus handle it silently. Getting this wrong in either direction is expensive, so a principled framework here is genuinely useful.