Non-human identity management is one of those critical blind spots in most organizations. As AI agents and automated workflows proliferate, the number of service accounts, API keys, and machine-to-machine credentials is exploding — often without the same governance we apply to human identities. The zero-trust angle is particularly relevant here. I've been building automation systems where agents need to authenticate across multiple services, and managing those credentials securely at scale is a real challenge. mTLS and workload identity are definitely the right direction. How are you seeing teams handle credential rotation for short-lived agent processes?