AI Agents Are a New Kind of Identity & Most Organizations Aren't Ready

This opinion/analysis argues that AI agents represent a new class of non-human identity in development environments that operate autonomously—touching repositories, opening pull requests, running pipelines—and currently lack adequate provisioning, monitoring, behavioral baselines, and audit trails; organizations should inventory all agents, understand their access and behavior, apply lifecycle controls, and extend identity governance into the SDLC to reduce risk and meet compliance requirements.