What happened
Android Authority reported on October 21, 2026 that Gemini is receiving a new Customize menu for adding and managing apps and skills. The menu is rolling out from that date across Gemini on Android, iOS, Windows, macOS, and the web. Google said the rollout could take up to 15 days to reach all users; the report’s author said the option was already visible on the web.
The new entry appears in the left-hand menu of the Gemini interface. Its stated purpose is to put compatible apps and skills in one place rather than leaving people to locate each capability through separate surfaces. The report describes access to more than 40 capabilities that can be added to the assistant. That number is an inventory of what the menu exposes, not a statement that every user will receive the same set of connections or permissions.
Customize is divided into three tabs. Discover presents a selection of app and skill types. Connectors is the place for external applications that can be connected to Gemini. Skills presents preconfigured skills and includes a control to create a skill. The reported connector list includes Google-built services such as Gemini Notebook, Google Photos, and YouTube, alongside third-party services including GitHub, Shopify, and Dropbox.
The preconfigured skills are framed as reusable ways to shape Gemini’s work. Examples in the report include letting Gemini try to mimic a user’s writing style in text output, preparing for meetings, and workshopping ideas. The feature is therefore a management and discovery layer: it collects connected services and task-oriented presets behind one visible menu. The supplied reporting does not say that the rollout creates a new developer API, changes connector pricing, or grants new permissions to those services.
This distinction matters. An assistant can have integrations without users being able to find or reason about them easily. A menu that names connections separately from skills gives users a more legible model: one category links Gemini to services; another applies a configured behavior or task framing. That can affect adoption even when the underlying integrations are unchanged.
What the archive adds
This is a new fact on a known Gemini integration story. A prior Field Note reported that custom MCP support appeared to be reaching more Gemini desktop users. The Customize report adds a broader, user-facing control surface for named apps and skills, but it does not confirm that arbitrary custom MCP servers can be added through this menu.
Trend signal
The rollout modestly supports Agent-ready software exposes capabilities. Gemini is presenting connected services and reusable skills as explicit capabilities a person can select and manage. The evidence is about interface organization, however, not about the permission model, composability, or safety boundaries of those capabilities.
Why now
The source does not give a strategic rationale beyond the rollout itself. Its timing suggests Google is consolidating an expanding set of Gemini connections and presets into a discoverable interface, but it does not establish whether this follows an underlying platform change or merely a product-interface redesign.
Connections to other events
The earlier custom-MCP rollout report concerned a route beyond provider-named app connections. This change concerns a visible menu for apps, connectors, and skills. Together, they point to Gemini broadening both the set of potential integrations and the interface through which users encounter them, while leaving their relationship unproven.
Why it matters
For builders, capability discovery is part of the product surface. A connector is less useful when users cannot see what it does, whether it is enabled, or how it differs from a task preset. Gemini’s separation of Connectors and Skills makes that distinction more explicit: service access is not the same thing as a repeatable way of working.
Teams using Gemini should treat the menu as a governance and onboarding checkpoint. They can review which organizational services are surfaced, what a preconfigured skill asks the assistant to do, and whether users understand the boundary between an app connection and a custom behavior. The immediate change is not evidence that Gemini can perform more work than before. It is evidence that its existing capabilities are being packaged for easier selection.
What happens next
- More connected services and skill presets will appear in the menu. (3 months, 65%). Watch for: additions to the reported 40-plus capability catalog.
- Organizations will need clearer internal guidance on which Gemini connections to enable. (6 months, 60%). Watch for: administrative controls or documentation distinguishing approved connectors from personal ones.
Limits and uncertainty
The available material is a report about a rollout, not a technical specification. It does not enumerate all 40-plus capabilities, describe regional or plan eligibility, explain authentication and data-sharing behavior for each connector, or state whether administrators can control them. It also does not show that creating a skill is equivalent to creating a programmable workflow. Most importantly, it does not confirm support for arbitrary custom MCP servers; the earlier desktop observation and this Customize menu should not be treated as the same feature.
Skill to build now
Make a short capability inventory for the assistants your team already uses. Separate service connections from reusable task instructions, identify who can enable each one, and record what data each connection could expose to the assistant. Then test one meeting-preparation or writing-style skill against a real, low-risk task and define a human review step before its output is used externally.
Reader takeaway
Gemini is making apps and skills easier to locate and manage through one Customize menu. The useful response is not to assume a new agent platform has arrived, but to inspect the capabilities that become more visible, distinguish connections from task presets, and decide which ones belong in a team workflow.
Open question
Will Customize become merely a directory for Gemini’s existing integrations, or a governed surface where organizations can define, approve, and distribute their own capabilities?