The Skills Era Begins for Google Gemini: Farewell to Gems

5 min read

Originally published in Turkish. Read the original →

The Skills Era Begins for Google Gemini: Farewell to Gems - cover image
Summary

Google has rolled out the far more practical Skills system globally in its Gemini assistant, invoked directly with a slash (/), replacing Gems, the customized bots.

The way we interact with large language models is evolving fast, from long and laborious prompt engineering toward quick, modular and task-focused shortcuts. Users' biggest expectation from AI assistants is not having to explain the same context, style rules and formatting guidelines from scratch in every session. To speed up routine work and eliminate prompt repetition entirely in its generative AI assistant Gemini, Google has rolled out a new feature it calls Skills globally.

This move doesn't just add a new capability; it also gradually retires Gems, the customized bot infrastructure the platform has used until now.

From Clunky Gems to an Integrated Skills Architecture

Google had previously launched a mini-bot format called "Gems" that users could customize for specific areas of expertise. When creating a Gem, users could assign it a specific role, tone and set of instructions. But this architecture had a serious drawback in practice: when a user needed a different kind of expertise, they had to open a separate chat window for that Gem and shuttle back and forth between conversations.

The new Skills architecture removes this disconnect completely. Instead of splitting the AI assistant into separate pages or bots, all capabilities are now included in a single central chat flow. Without ever leaving the main conversation window, users can instantly bring in predefined custom skills for different tasks.

Calling Skills Quickly with a Slash (/)

At the center of the Skills system is the slash (/) command structure familiar from modern messaging and collaboration platforms.

    When users simply type the / character into the Gemini chat box, they see an interactive menu listing all defined skills. From here they can select the desired skill and give task-specific parameters:
  • Formatting and Presentation Prep: For example, with a /presentation-draft command, the system automatically applies in the background instructions that turn the given raw notes directly into a specific slide hierarchy.
  • Code Review and Security Audit: Developers can use a /security-scan command to check in a single step whether a code block complies with corporate coding standards and open source licenses.
  • Academic Summary and Translation: For a long PDF or research document, running an /academic-summary command applies predefined summarization rules within seconds.

Thanks to this structure, users no longer need to enter paragraphs of system prompt every time, along the lines of "Answer in this tone, use these headings, keep it this long".

What Will Happen to Existing Gems?

The first question on users' minds was the fate of the custom Gems bots they had put a lot of effort into creating. Google announced a clear transition plan that reassures users on this point.

No Gem saved on the platform will be deleted. The company will automatically convert all of users' existing Gems into the new Skills infrastructure and assign them to their accounts. This way, users will be able to access all the bot instructions they prepared earlier directly through the / command.

    In addition, in the coming weeks the advanced Gems features users relied on most will also move under the Skills umbrella:
  • Being able to connect specific Drive folders or documents directly to a skill through Google Drive integration.
  • Creating shared skill libraries across teams and sharing shared skills with colleagues in a single click.

The "Micro-Agent" Era in AI Assistants

Google's adoption of the Skills approach in place of Gems stands out as part of a larger transformation across the industry. AI systems are moving beyond being passive chatbots that give general answers to questions and turning into modular micro-agents that react instantly to the user's workflow.

Similar steps are visible in rival ecosystems too; the TypeScript-based mod system Anthropic offers for Claude Code and OpenAI's custom actions aim to increase how quickly AI adapts to the user. Google's move to slash ergonomics in Gemini aims especially to minimize the prompt-writing time and cognitive load professionals spend every day.

As the new Skills feature gradually opens up to all Gemini users, skill integrations that go even deeper with Google Workspace tools are expected to arrive in the coming days.