One of Gemini’s best features is getting a must-needed usability boost on Android
Source: Android Authority

TL;DR
- Google is working on a new way for users to create scheduled actions in Gemini on Android.
- This includes a UI that allows users to name, describe, and schedule these actions — just like on the web.
- Previously, app users were forced to create actions using a prompt.
Background
Google introduced Scheduled Actions to Gemini last year. The feature lets users schedule and run various tasks through the AI assistant, turning Gemini into a useful automation tool. For example, you could create a daily digest of news sources delivered to you before bed each night. It’s highly extensible.
However, there has been a gulf between the feature’s functionality on the web and in the Gemini Android app, especially the lack of a UI for easily creating these actions.
New UI Workflow
During our investigation of Google app version 17.8.59.sa.arm64, we discovered a new workflow that gives users more control over automation creation:
- Open the updated Scheduled Actions screen.
- Tap the + icon in the top‑right corner.
- A bottom sheet appears, allowing you to name, describe, and schedule the new action.
This UI is a significant improvement over the previous prompt‑only method and aligns the mobile experience with the web version.
Visual Comparison
The article originally included side‑by‑side screenshots of the new mobile UI and the existing web UI.
Limitations
- We previously found evidence that Google was working on a way for users to edit existing scheduled actions via the Android app, but that feature has not yet rolled out.
- The current UI only allows users to pause or delete actions they’ve created.
- Scheduled Actions remain locked behind Google’s AI Pro or Ultra subscriptions (details), limiting access for many users.
Outlook
It’s unclear when the new UI will be rolled out broadly, but the changes indicate Google is addressing some of the more minor, yet annoying, aspects of the Gemini experience.
⚠️ An APK teardown can predict future features based on work‑in‑progress code, but predicted features may not reach a public release.