Anthropic Introduces Cross-App Persistent Memory for Claude

Anthropic has updated its AI assistant, Claude, to share a single persistent memory across its different platforms, including chat and its Cowork productivity tool. Previously, users often had to restate background details, project goals, or preferences when switching between tools. Now, details saved in a standard chat carry over immediately to autonomous work tasks in Cowork, and vice versa.

The system automatically adds details to a user's memory profile in real time as they converse, rather than waiting for a chat session to end. Users can inspect, modify, or delete specific items in their memory settings at any time. To address privacy concerns, Anthropic turns off memory for sensitive subjects like health, political beliefs, and religious preferences by default, though users can manually opt in to save those details if they find it helpful.

While shared memory reduces repetitive setup for daily tasks, it remains unclear how effectively the AI will handle conflicting or outdated memories as projects evolve over several months. Moreover, as AI assistants gather deeper knowledge across personal and professional environments, users may face increased friction if they ever decide to switch to a rival platform, as their accumulated context cannot easily be exported to a competitor.