Google to Force-Migrate Gemini Gems to Skills on November 17

On November 17, 2026, Google will automatically convert Gems in personal Google accounts to a new format called Skills. According to reporting by Ben Khalesi of Android Police, the conversion is a mandatory measure that does not involve user choice, and Gems will no longer be available on personal accounts. Migration for work and school accounts is postponed to 2027.

Gems were designed as standalone custom bots. Users selected any Gem from the sidebar and conversed with it in a dedicated window. Skills, by contrast, bring custom instructions into the main chat. Users type a slash in the prompt bar and select the Skill they want to apply. In some cases, Gemini determines the appropriate Skill from the prompt content and applies it automatically.

At the same time, Google is moving the background agent Spark into the main chat as “Task.” Scheduled Actions, offered since 2025, will also be lined up in the same place. The Android Police article points out that the names alone do not allow users to infer each feature’s role.

The Turning Point From Gems to Skills

The clearest change in the shift to Skills is that the starting point for interaction moves from the sidebar to the prompt bar. Rather than switching among multiple bots, users layer instructions within a single conversation.

The ability to stack them means more than a simple rename. It becomes possible to use one Skill to enforce a company’s brand spelling and another to define the tone of the writing. Work that previously required passing output back and forth between bots is completed within a single thread.

Gems were standalone custom bots. You picked one from a sidebar and talked to it in its own walled-off window. Skills pull your custom instructions into the main chat.

Android Police puts it this way, positioning Skills not as a “consolidation of instructions” but as a “change in the structure of interaction.”

A New Design That Stays Within the Chat

The value of Skills lies in the separation of responsibilities through multiple applications. The writing process can be divided into units such as style rules, structure, and proofreading, each managed as an independent Skill. This increases prompt reusability and makes it easier to share instructions within a team.

However, at present, there is no feature for sharing Skills themselves. It still cannot be used to distribute a unified set of instructions across an organization. Google says it will soon support sharing and Drive file attachments, but it has not specified when.

Unsupported Features to Check Before Migration

One thing easily overlooked in the migration is the handling of built-in tools that could be enabled within a Gem. Canvas, Deep Research, Guided Learning, and video/music generation all do not work in Skills. Android Police urges users who have Gems that depend on these features to verify them before November 17.

The impact is significant for business use. Research Gems with Deep Research built in, or production workflows that use Canvas to build structure, will not be able to reproduce the same output after migration. Alternative procedures need to be prepared in advance.

The Reality of Scheduled Actions

Scheduled Actions is a feature that ties a prompt to an execution time and frequency. Users write an instruction and specify when and at what interval it should run. Results are generated in the background. Up to 10 can run at the same time.

On the web version, the relevant chat appears as unread. On mobile, a push notification arrives and, depending on settings, is also shown on the lock screen. It is suited to creating routine briefs but not to processing that involves complex conditional branching. Android Police also rates it as effective for simple summarization but lacking reliability beyond that.

The Role of Spark, Now Task

Spark was offered as an agent that runs in the background, but it will be integrated into the main chat under the name Task. Whereas Scheduled Actions triggers prompts based on time, Task is positioned as the entity that executes work. The names are similar, making the boundary between their roles hard for users to discern.

The Burden of Successive Renames on

Development Teams

For users, name changes manifest as increased learning costs. Feature names embedded in internal manuals, prompt collections, and training materials must be updated with each migration. External explainer articles and videos also assume the names as of publication, so they quickly become outdated.

As previously reported by this site in Google Develops “Frozen v2,” a New AI Chip for More Efficient Gemini, announcements around Gemini continue on both the infrastructure and feature sides. If name reorganizations occur frequently, simply tracking which is the current official name will require considerable effort.

The consolidation and elimination of features itself can be understood as a direction toward concentrating Gemini into a single conversational interface. Given the push for stronger reasoning shown in Google Announces Gemini 3.8 Flash: Stronger Reasoning Comes With Higher Real Costs, there is strong pressure to reorganize the interaction model as the model evolves.

On the other hand, the fact that the mandatory conversion will be carried out while some features remain unavailable at the destination is a clear step backward for users. Google promises additional support for unsupported features, but no schedule has been announced. Organizations that have incorporated Gems into their workflows need to verify alternatives with November 17 as the deadline.

Editorial Opinion

The mandatory migration on November 17 will force individual users to re-examine the Gems they have built into their workflows. Corporate accounts have until 2027, but plans for internal deployment will need to be revisited. We believe an increase in inquiries after the migration is unavoidable.

In the long term, the reorganization of feature names can be seen as groundwork for concentrating Gemini into a single conversational interface. It is a design intended not to repeat the past failure of letting peripheral tools proliferate. However, overlapping names raise learning costs and accelerate the obsolescence of external documentation.

What is at issue is less whether to migrate than how to close the feature gaps after migration. If support for Canvas and Deep Research is delayed for a long time, users will migrate to alternatives. We need to examine how much divergence between announcements and implementation should be tolerated.

References

• “Google is renaming Gemini again; here’s what each feature does now”, by Ben Khalesi — Android Police, 2026-10-18T21:00:13.000Z (ARR)
• Source URL: https://www.androidpolice.com/googles-renamed-gemini-features-field-guide/