Feature Expansion, Not Debut
On September 18, 2025, Google announced an expansion to the sharing functionality of Gemini Gems. This development upgrades a functionality that had existed for some time until possibly August 2024, providing users with a more controlled way of sharing custom AI bots with others. What was missing was thus an interface more analogous to Google Drive’s permission settings, so it’s now a matter of choice as to how one would like to view or edit a Gem.
While the update takes away some controls from users who have used Gems until now, which constitutes a shift in UX, especially for those who use Gemini in collaborative settings, the changes have been conceived to allow safe distribution with less friction.
What Are Gems, and Who Uses Them?
Gems are task-specific AI agents built inside Google Gemini. Users define a task in plain language—such as “summarise articles using peer-reviewed sources” or “draft polite email replies”—and Gemini generates a bot that performs that function.
Creating a gem does not require any coding. Users give their gem a title, describe the task, and test its performance. From there, they can begin using it across tasks or sharing it with others. Instructors, freelancers, marketers, and students have used this method to seamlessly incorporate simple, repeatable AI workflows into their daily processes.
For instance, a university instructor could work on a Gem to give interim feedback on essay drafts with reference to particular grading criteria. A freelancer could work on a Gem for the clients that formats copy into the brand tone and structure. Since 2024, these use cases have been quietly gaining traction among some user groups.
New Controls and Global Sharing
With the September 2025 update, Google introduced expanded sharing permissions. Users can now send links to specific people, limit visibility to only their organisation, or make a Gem publicly accessible. The rights might include viewing and editing, while access can be updated or revoked through the Gem Manager dashboard.
The whole development seems aligned with Google’s general agenda to fully integrate Gemini into its Workspace suite. From the moment Gemini starts entering Gmail, Docs, Sheets, and other Workspace tools, being able to create and share Gems with their respective sharing models will truly pierce through the barrier of automation for personal use versus a tool for team collaboration.
And this enhancement will be for both, for pay and free-tier users. The decision truly makes these productivity tools worldwide accessible, especially for independent workers and students in places with very little exposure to any paid AI platform.
No Public Discovery Layer
Despite the new sharing controls, one key feature remains absent: a public repository or marketplace for browsing Gems. At present, all Gem sharing remains link-based. Users must either know the creator or be given a direct URL to access another user’s Gem.
This limits the broader ecosystem potential, especially when compared to how apps or templates are discovered in Google Workspace Marketplace or similar platforms. Google has not confirmed whether a central Gem directory is in development.
Practical Impact for Daily Users
For daily end users, this update translates to much stronger control over shared AI tools. The teacher can now make sure that their gem is read-only when distributed to the students. The team lead can release into their group a task-specific gem without it ever being accidentally edited.
As Gemini continues to embed itself across Google’s tools, these Gems function less as isolated bots and more like dynamic templates. Tasks that were once repetitive or manual—reformatting text, summarising data, adjusting tone—can be delegated to a consistent AI agent.
Feedback across user forums suggests growing interest, although no official usage data has been published by Google as of September 19, 2025. Anecdotally, users have reported building Gems for everything from summarising PDFs to generating grocery lists. Whether these tools are used daily or occasionally, their accessibility and repeatability are part of their appeal.
Comparison with Other AI Tools
The distinction and difference need to be put down on how Gems are different from other similar tools, say, for example, Custom GPTs by OpenAI. Granted, GPTs can be used in the addition of some advanced features like file uploads and API access, but not all custom GPTs require a developer or a paid plan. In a lot of cases, very simple GPTs are made available for free and can be created with little or no difficulty.
Gems, meanwhile, are much more tightly integrated with Google’s ecosystem. They have the advantage of workspace access and a no-code design paradigm. While they may not yet be able to enable everything a GPT can do, they are arguably simpler to get into for the common user.
Additionally, Gems can be combined with Google Drive files; that sort of thing would certainly add utility to the AI services already employed by many in their daily work. This integration places Gems in the correct light: complementing other AI platforms rather than competing.
Looking Ahead
As of now, Google has not published usage metrics related to the updated Gem sharing feature. Any assessment of adoption remains anecdotal. That said, the structural changes in September 2025 represent a more mature implementation of an existing capability rather than a new innovation.
It is clear that more users now have access to these tools, while the barrier to sharing AI-powered workflows has been lowered. Whether used to standardise a classroom activity, streamline writing tasks, or offer a shared resource to a community, gems are becoming part of the broader toolkit for AI-powered productivity.
Their growth will depend in part on whether Google introduces stronger discovery tools and extends functionality beyond current constraints. For now, the power to build and distribute AI assistance remains with the user—and that may be enough to fuel a new phase of adoption.