Google's AI assistant Gemini appears to be on the verge of significant expansion with two powerful new features discovered in recent app code. These unreleased capabilities could dramatically enhance how users interact with the AI, though both come with notable limitations that users should be aware of.
Video Generation Coming to Gemini
Evidence uncovered in the latest beta version of the Google app (16.9.39.sa.arm64 beta) confirms that Google is actively developing a video generation feature for Gemini. Codenamed Toucan, this feature will allow users to create videos simply by providing text descriptions of their ideas. According to the code strings, the AI will process these requests and generate corresponding videos, though the process may take several minutes to complete.
Upcoming Gemini Features
- Video generation (codename: "Toucan")
- Search history personalization
Video Generation Details
- Uses text prompts to create videos
- Processing takes "a few minutes"
- Likely powered by Veo 2 AI model
- May have daily usage limits
Search History Integration
- Opt-in feature requiring explicit permission
- Accesses several months of search history
- Chats deleted from activity within 60 days
- Requires Web & App Activity to be enabled
Usage Limits May Apply
While video generation represents an exciting advancement for Gemini, it appears the feature may come with usage restrictions. Code discovered in the app includes a message stating: You've reached your Toucan generation limit until tomorrow, suggesting daily caps on video creation. This limitation might only apply to users of the free version of Gemini, with Gemini Advanced subscribers potentially receiving higher or unlimited usage quotas.
Technical Underpinnings
The video generation capability is likely powered by Google's Veo 2 AI model, which was introduced last year with promises of integration into products like YouTube Shorts. This specialized model appears to be dedicated specifically to video creation tasks within the Gemini ecosystem, representing Google's push to compete with other AI platforms offering similar creative tools.
Search History Integration
In a separate but equally significant development, Google is working on a feature called Gemini Personalization that would allow the AI assistant to access users' Google Search history. This opt-in feature, discovered in version 16.8.31 of the Google app, would enable Gemini to provide highly personalized responses based on a user's past search behavior.
Privacy Considerations
Google appears to be implementing privacy safeguards around the search history feature. According to the discovered code, only the specific Gemini Personalization model would connect to search history data. Additionally, chats using this model wouldn't be used to improve Gemini, wouldn't be stored outside the user's chat history, and would be deleted from activity logs within 60 days.
Practical Applications
The search history integration could enable powerful use cases, such as asking Gemini about restaurants you previously researched or requesting recommendations based on products you've searched for in the past. Early evidence suggests the feature can access several months of search history, potentially going back as far as January of this year, though the final implementation might include even more historical data.
Opt-In Requirement
Users will need to explicitly grant permission for Gemini to access their search history and ensure Web & App Activity is enabled in their Google settings. This approach aligns with Google's recent emphasis on transparent AI development and user control over data sharing.
Competitive Positioning
These upcoming features represent Google's continued effort to differentiate Gemini in the increasingly competitive AI assistant market. Video generation capabilities would put Gemini in direct competition with other AI platforms offering creative tools, while search history integration could provide a uniquely personalized experience that leverages Google's core strength in search technology.