Google Gemini Skills Replace Gems: What Users Should Do
If you built custom Gemini Gems to speed up writing, planning, coding, or research, Google’s shift to Gemini Skills deserves your attention now. According to TechCrunch, Google is killing off Gems in favor of Gemini Skills, a move that points Gemini away from static custom chatbot personas and toward reusable task-based abilities. That sounds tidy on a product slide, but it can create real cleanup work for anyone who depends on saved instructions. You may need to copy prompts, rethink workflows, and watch for migration gaps before Google makes the change final. The larger signal is clear enough. Google wants Gemini to act less like a shelf of custom bots and more like a toolbox where each tool performs a specific job.
What Stands Out
- Gemini Skills appear set to replace Gems as Google’s preferred way to package repeatable AI tasks.
- Gems focused on custom instructions and chatbot-style roles, while Skills sound more action-oriented.
- Users should save important Gem prompts now instead of assuming Google will preserve every detail.
- The change fits a wider AI product trend, where companies are moving from chat personas to workflow components.
Why Google Gemini Skills Matter
Gems were useful because they gave regular users a simple way to create a specialized version of Gemini. You could make a writing coach, a trip planner, a code reviewer, or a customer-support helper with a saved set of instructions.
That model is easy to understand, but it can get messy fast. Once you have ten or twenty Gems, you are managing a drawer full of slightly different prompt recipes, many of which overlap or go stale.
Gemini Skills point to a cleaner product idea. Instead of asking you to maintain a full custom assistant for every job, Google can package repeatable actions as smaller capabilities that plug into Gemini when needed.
TechCrunch reports that Google is phasing out Gemini’s Gems in favor of Skills, which suggests a shift from custom chatbot personalities toward task-focused AI building blocks.
Look, this is not shocking if you have watched AI products for the past few years. The first wave gave users blank text boxes and custom bots, while the next wave is about getting the assistant to do actual work with less fiddling.
Gemini Skills vs Gems: What Is Really Changing?
The key difference is control versus structure. Gems gave you a lightweight way to define how Gemini should behave, but Skills likely give Google more room to define what Gemini can do inside a controlled product system.
Think of Gems as handwritten recipe cards and Skills as prepped ingredients in a restaurant kitchen. The recipe card is flexible, but the kitchen station is faster, easier to repeat, and easier for the operator to quality-check.
What Gems did well
Gems were handy for personal style and recurring context. If you wanted Gemini to write in your brand voice, critique meeting notes, or act as a patient tutor, a Gem could save you from pasting the same setup prompt again and again.
They also lowered the barrier for non-technical users. You did not need an API key, a workflow builder, or a long setup process, which made Gems feel approachable for people who simply wanted a better chat session.
Where Skills may fit better
Skills make more sense for tasks with clearer inputs and outputs. Summarizing a document, generating a spreadsheet formula, turning notes into a project plan, or checking writing against a style guide all fit that pattern.
That structure matters because AI assistants are moving into work software, not sitting off to the side. Google has Gemini inside Workspace, Android, Chrome, and Search-adjacent experiences, so it needs reusable parts that behave predictably.
This is the handoff to watch.
What You Should Do Before Gems Disappear
Do not wait for a last-minute product banner. If a Gem matters to your work, treat it like any other business asset and back it up while you still can.
- Copy the full instructions from each important Gem. Save them in Google Docs, Notion, Obsidian, or whatever system you trust.
- Label each prompt by job. Use plain names such as sales email reviewer, weekly planning assistant, or Python bug checker.
- Write down sample inputs and outputs. This helps you rebuild the workflow if Skills do not migrate cleanly.
- Separate style from task logic. Keep voice rules in one section and process steps in another, since Skills may treat those pieces differently.
- Watch Google’s product notices. TechCrunch’s report is the signal, but account-level messages will likely tell you what happens to your own Gems.
If you run a team, ask people which Gems they use before IT makes a blanket call. Shadow AI rarely looks dramatic until one small helper disappears and suddenly three weekly reports take twice as long.
How Gemini Skills Could Help Teams
The upside is real if Google handles the transition well. Gemini Skills could make shared AI workflows easier to approve, test, and roll out across a company.
Custom bots often create governance headaches because everyone writes prompts differently. A marketing manager might build a Gem with sensitive customer context, while a sales rep makes another that gives off-brand claims, and nobody notices until the output lands in front of a client.
Skills could reduce that sprawl. Admins may get cleaner controls, users may get better defaults, and Google may get a product layer that works across Workspace apps without asking people to manage dozens of homegrown bots.
Where Google still has to prove itself
The risk is that Skills become too rigid. If Google turns a flexible user feature into a locked catalog of approved actions, power users may lose the weird little custom workflows that made Gems useful in the first place.
What happens to a Gem that mixes brand voice, source preferences, formatting rules, and a review checklist? That is not one simple Skill, and forcing it into a narrow template could make the experience worse.
Google also has a naming problem. Gemini already spans apps, models, subscriptions, Workspace features, Android features, and developer tools, so another product label needs clear placement (and preferably less guesswork).
Gemini Skills and the Bigger AI Product Shift
Google is not alone here. OpenAI, Anthropic, Microsoft, and other AI companies have all been trying to move users from open-ended chat toward repeatable workflows, agents, connectors, and app-like actions.
The reason is simple. Chat is a great interface for exploration, but it is a shaky foundation for work that needs consistency, permissions, audit trails, and integration with files or business data.
For users, the winning model will be boring in the best way. You ask for a job, the assistant picks the right capability, checks the right source, and gives you an answer you can trace.
For Google, Gemini Skills may also help close the gap between consumer AI and Workspace AI. A Skill that summarizes Drive files, builds a Slides outline, or drafts a Gmail reply has more product value than a custom chatbot that only lives in a side panel.
What to Watch Next
Pay attention to how Google handles migration. If existing Gems become Skills automatically, this could be a fairly painless product cleanup, but if users must rebuild everything by hand, expect frustration from the people who used Gems the most.
Also watch whether Skills support user-created instructions or only Google-made templates. The difference matters. A controlled marketplace of Skills would feel very different from a flexible builder that lets you turn your own workflow into a reusable action.
My bet, after years covering these AI product resets, is that Google wants both control and customization. The hard part is giving users enough room to make Gemini useful without letting the product turn into a junk drawer of half-working automations.
The Smart Move Right Now
Back up your Gems, rank the ones you actually use, and rewrite the best ones as clear task steps. That gives you a portable version of your workflow whether Gemini Skills are excellent, limited, or somewhere in the middle.
Google’s change may end up being a smart cleanup, but users should not treat saved AI instructions as permanent. If your daily work depends on a custom assistant, ask yourself a blunt question. Could you rebuild it tomorrow?
