Product Launch2026-09-29TechCrunch AI

Google kills Gemini Gems for skills

Google is ending Gemini's Gems feature, which allowed users to build task-specific AI agents, and replacing it with a new skills system. The change comes as all-in-one AI agents such as Meta's Muse and Instinct gain traction, and as Google appears to favor built-in, reusable skills that integrate more seamlessly across its AI ecosystem. Gems gave users a way to customize chatbots for particular purposes, from writing help to research assistance. But the feature also created separate, fragmented experiences that could be hard to manage. By moving to skills, Google may simplify how people access specialized AI capabilities. Instead of building and maintaining many individual Gems, users could rely on modular skills that work across Gemini and related products. That approach could make advanced functions easier to discover and reuse. It may also give Google more control over quality, safety, and consistency, since skills would be part of the core platform rather than user-created agents. The shift reflects a broader debate in the AI industry about how autonomous assistants should be organized. Some companies are pushing toward all-in-one agents that can handle many tasks without requiring users to configure separate bots. Others still see value in specialized assistants tailored to narrow jobs. Google's decision suggests it believes the future lies in composable capabilities that feel native to the assistant, not in a marketplace of separate Gems. For existing users, the transition may mean losing some custom configurations or needing to adapt workflows. Google has not detailed every step of the migration, but the direction is clear. The company wants Gemini to feel more unified and less fragmented. If the skills system works well, it could make Gemini easier to use and more competitive against rival agent platforms. If it does not, Google risks alienating users who invested time in building Gems. The outcome will be an important test of how AI assistants should balance customization with simplicity.

Related news