Claude Managed Agents gain persistent memory in public beta

- Anthropic put persistent memory into Claude Managed Agents in public beta, allowing agents to retain context across developer‑controlled sessions. - Early users like Netflix and Wisedocs reported up to 97% fewer errors in some use cases when agents kept persistent context, per the EdTech Innovation Hub. - Persistent memory shifts agents from stateless helpers toward ongoing junior operators that will need lifecycle, privacy and security controls. (edtechinnovationhub.com)