Google is renaming Gemini again; here’s what each feature does now

The latest round of changes kicks off on November 17, when Google begins force-migrating personal Gemini accounts away from custom bots known as Gems and into a newly branded format called Skills. Enterprise and educational accounts will follow suit with their own migrations slated for 2027. Meanwhile, Google is quietly transitioning Spark, its asynchronous background agent, into the main chat interface under the designation of "Task," positioning it right alongside Scheduled Actions, a recurring prompt and timer feature introduced in 2025.

For everyday users trying to navigate the rapidly shifting AI landscape, telling these capabilities apart based on their names alone has become an exercise in frustration. Understanding how each tool functions and securing your saved workflows before the legacy interfaces vanish requires a close look at what Google is changing and why.

Gems are turning into Skills, and they’re moving into your chat

Starting November 17, Google will begin the automated conversion of personal Gems into Skills. Until that date, users can continue utilizing their existing Gems as standalone custom bots accessed from a dedicated sidebar. However, once the transition takes effect, Gems will disappear entirely from personal accounts, replaced by a much more integrated approach to custom instructions.

Previously, Gems operated within their own walled-off windows, requiring users to switch between separate bot interfaces to maintain specific conversation contexts. Skills, by contrast, pull custom instructions directly into the primary chat window. Users can summon a specific Skill simply by typing a forward slash in the prompt bar, or allow Gemini to apply a relevant Skill automatically when it detects a match with the ongoing conversation.

The most substantial upgrade in this transition is the ability to stack multiple Skills in a single conversation. For instance, a user might apply one Skill to enforce strict corporate branding guidelines while simultaneously deploying another to dictate a specific writing tone. This eliminates the tedious process of shuttling text outputs back and forth between different bots to achieve a desired result.

Despite the added flexibility, the transition comes with notable limitations. Most of the advanced built-in tools previously accessible within individual Gems—such as Canvas, Deep Research, Guided Learning, and various video and music generation capabilities—do not currently function with Skills. Furthermore, users cannot yet share Skills or attach files directly from Google Drive, though Google has indicated that these features are in development. Users whose custom workflows rely heavily on these integrated tools are advised to test their setups before the November 17 cutoff date.

Scheduled Actions are prompts with an alarm clock

Another pillar of the Gemini feature set is Scheduled Actions, which allow users to write a prompt, specify a timeline and frequency, and have the system prepare the results in the background. Up to ten of these automated prompts can run simultaneously. On the web interface, completed actions mark the chat as unread, while mobile users receive push notifications depending on their device settings.

This functionality is expanding to free accounts as Google rolls it out more broadly, provided users keep their chat activity history enabled. Gemini typically prepares each requested result during the hour leading up to its scheduled delivery. This preparatory window introduces a distinct limitation, making the feature best suited for static information rather than rapidly changing data. Google’s primary use cases include morning digests compiling calendar entries, to-do lists, and unread emails, alongside daily weather forecasts paired with outfit suggestions.

A notable quirk involves location-dependent prompts, such as daily weather reports. Gemini relies strictly on the geographical location where the action was originally configured rather than tracking the user’s current whereabouts. Individuals who travel frequently will continue receiving forecasts tailored to their home location unless they manually edit the scheduled action to reflect their new city. For dynamic tasks that require real-time adaptation or event-driven triggers, users must look to Google’s background agent architecture.

Spark is the agent that keeps working after you log off

Reflecting Google’s penchant for shifting nomenclature, the background agent previously known as Spark is undergoing a rebranding. In a redesign currently rolling out to users, the feature appears as "Task" within a newly added Chat and Task toggle situated directly next to the prompt box, with its corresponding settings page similarly renamed to Task settings. Despite the interface updates, Google’s documentation continues to reference the system as Spark.

Operating on Google Cloud virtual machines, Spark functions as an always-on background agent capable of running even after a user closes their laptop or locks their smartphone. Users can assign broad, high-level objectives—such as comprehensive web research regarding competitive market trends—and allow the agent to execute the necessary steps independently.

To maintain security and user oversight, Google has embedded specific checkpoints into the agent’s framework. Spark is designed to pause and request explicit user confirmation before transmitting messages, altering personal data, executing financial transactions, or submitting online web forms. It is also programmed to hand control back to the human user whenever a targeted website demands manual password entry or payment details. Google emphasizes that these protective measures are not entirely foolproof, advising users to monitor sensitive operations closely.

Running up to fifteen concurrent tasks requires a paid Google AI Pro or Ultra subscription alongside active chat history settings.

Why scheduling a Spark task isn’t a Scheduled Action

The potential for user confusion deepens when examining the overlapping scheduling capabilities of Gemini’s tools. Both Scheduled Actions and Spark tasks can be set to run on a predetermined timeline, leading to understandable ambiguity about their differences.

Unlike standard chats or Scheduled Actions, which are strictly bound to a clock, Spark possesses the ability to initiate actions dynamically based on external events. It can also execute complex, multi-step workflows across various web applications and services while the user is entirely offline. For example, a user can instruct Spark to monitor airline ticket prices continuously and deliver a notification only when a fare drops below a specified budget threshold, or have it audit an email inbox to systematically unsubscribe from unwanted newsletters.

Back up your prompts while you can

As Google continues to reshape its artificial intelligence portfolio, managing the transition requires a proactive approach. For users seeking consistent behavioral guidelines from Gemini across sessions, Skills provide a free and flexible solution. Those interested in automated morning briefings can utilize Scheduled Actions without requiring a paid subscription. Meanwhile, Spark—or Task, depending on the current interface iteration—remains an option reserved for subscribers who need autonomous, event-driven agents operating across multiple applications in the background.

With the retirement of personal Gems approaching on November 17, users are strongly encouraged to secure their custom prompts by copying individual instructions into Markdown files and downloading any associated knowledge assets. Given Google’s track record of renaming, merging, and sunsetting its software features, maintaining independent backups ensures that valuable custom workflows remain intact regardless of future corporate rebrandings.