Google to Retire Gemini Gems on Nov 17, Switch to Skills

Google will retire Gemini Gems on November 17, 2026, and replace them with skills. TechCrunch

Existing Gems will convert automatically into skills that can be used across different AI tasks. Users will not need to take action to migrate. Gems will remain usable until November 17.

Skills change the interaction model. Instead of opening a separate custom assistant, users type a forward slash "/" in a task thread to select a skill. Skills in Gemini Apps let users create and save custom instructions once and reuse them across tasks, much like reusable templates that can be pulled into any chat.

Gems launched in 2024 as a way to build custom AI assistants for specific tasks. The intent was to let users teach the model to perform tasks without repeating instructions. That design targeted repetitive work such as writing help or simplifying sentences.

Google shipped a set of pre-made Gems for common patterns. The set included a learning coach, brainstorming assistant, career guide, coding partner, and editor. Users could use those pre-made versions immediately or customize them.

Users could also create their own Gems for narrower needs such as a running coach, nutritionist, or vacation planner. Gems could be shared with others. For construction, Google advised providing details about goals, desired behaviours and preferred format to personalize responses, with a magic wand icon to get Gemini's help writing the instructions. Educators could create Gems, described as custom versions of Gemini, for students to interact with.

The wind-down was first reported by 9to5Google over the weekend before September 28, 2026. 9to5Google reported on September 27 that the Gemini app would replace Gems with skills in November, with the formal migration starting November 17.

Looking at daily use, the shift is from discrete assistants to portable instruction blocks. A Gem kept persona, instructions, and task context together in one object. A skill keeps the instructions separate from the thread so they can be called up on demand. For people who keep prompt libraries, short for collections of saved instructions, that reduces duplication. One maintained skill can serve many threads, rather than many near-identical assistants maintained in parallel.

In my view, heavy Gem users should note the trade. Slash invocation is faster for composition, but it puts more responsibility on naming, versioning, and instruction hygiene, meaning keeping directions clear and up to date. A vague skill applied broadly can leak tone or format constraints into tasks where they do not belong. The automatic migration preserves continuity, but users who built elaborate Gems will want to audit what transferred, how each skill triggers, and where defaults now apply.

The broader context here is instruction reuse. The industry has moved steadily away from one-off chats toward saved system-level guidance, reusable templates, and in-flow dispatch, meaning calling up saved help inside a chat. Skills fit that direction. If the migration works as described, the practical gain is less setup overhead and more consistent output across tasks without rebuilding context each time.