Claude's internal Hub now asks why you switched back

Anthropic is still testing Hub, a mode flagged as internal inside Claude's mobile app, where it sits next to a planet icon. The latest build asks users to provide feedback when they leave Hub and switch back to a normal chat, according to Threads.

A prompt that fires on the way back to normal chat reads as retention measurement: Anthropic appears to be checking whether an agent-oriented surface holds users better than a single conversation thread. The Connected Devices move points in the same direction, since an orchestration view is only useful if the things it orchestrates are easy to attach. Both changes fit a year in which agent management, not raw model output, has become the interface problem.

According to Threads, the request is tied to the switch itself, appearing when a user drops out of Hub and back into the standard chat. Hub remains flagged as internal, and Anthropic has not put it in front of general users.

The planet icon has appeared in Claude before under a different name. A similar icon was labelled Orbit and later removed, according to Testing Catalog and Memeburn, which both place the current Hub entry point in the same part of the mobile app. Testing Catalog tied the mode to sub-agents in an earlier report.

Threads describes Hub as a task orchestration component that Anthropic may eventually release broadly, alongside the widget change. The Testing Catalog write-up on Hub and sub-agents was last updated on August 26, and the mode stays inside builds marked for internal testing.

Hub's route to a public release

No release date has been given for Hub, and Anthropic has not announced the mode as a product. Whether it will coordinate work across the devices now exposed in the tool widget is unclear from the current builds, as is whether the planet icon survives this time after the Orbit entry was pulled. The switch-back feedback prompt suggests the internal phase is not finished.