AI Daily Digest: Muse adds custom spaces; Anthropic examines agent memory

The Provenance AI Report | November 8, 2026

Meta says Muse now lets users design persistent views around their routines. Anthropic’s latest memory research examines a different problem: deciding which information an agent should retain and correct between sessions. The first changes an app’s interface; the second offers evidence about a design approach, rather than a new product rollout. Two smaller updates concern Grok’s text watermarking and what Copilot’s usage charts actually count.

• Muse: where custom spaces can appear, and what sharing still lacks.
• Agent memory: why checking saved notes against transcripts can matter.
• In brief: Grok’s documented watermarking start date and Copilot’s expanded engagement reporting.

Muse adds views that users can keep and rearrange

Meta’s November 6 announcement introduces spaces inside Muse. A user describes a planner, tracker or other custom experience, then adjusts its appearance and behavior through conversation. The result remains part of the app. Spaces can live in the sidebar or main tabs, be hidden, or appear on a schedule.