Gemini Map Tool Arrives as Google Replaces Slash Commands With @

Google has added a Gemini Map tool on Android and iOS, while a second interface change replaces the slash command with an @ menu.

The map lets users place a geographic area directly into a prompt instead of describing that area in words. The @ change brings Skills, connected services, and other tools into one selection layer.

Together, these updates point toward a different kind of Gemini app. Google is turning the prompt box from a blank text field into a control surface for choosing context and capabilities.

The immediate change is modest. Users get a visual location picker and a consolidated menu. The larger bet is that an assistant becomes more useful when people can show it exactly where to work.

That puts pressure on the traditional text-first chatbot interface. OpenAI, Anthropic, and other assistant developers have expanded tool use, but Google owns a distinctive combination of mobile distribution and mapping infrastructure.

The Gemini Map tool tests whether that advantage can translate into a better prompting experience. It also raises familiar questions about availability, location accuracy, data handling, and whether Gemini can reliably act on selected context.

The Gemini Map Tool Turns an Area Into Prompt Context

The new map changes location from something users describe into something they can select.

The mobile Map tool appears in the Gemini app’s attachment carousel on Android and iOS. It sits alongside options such as Photos, Camera, Files, Drive, Avatar, and Notebooks.

Selecting Map opens a live view centered on the user’s current area. The interface includes a circular focus region that identifies the location Gemini should consider.

Users can move the map, zoom in or out, or search for another destination. Tapping “Explore this area” then adds a “Map Area” attachment to the prompt box.

That attachment can be combined with a written request. A traveler might select several blocks around a hotel and ask for quiet restaurants suitable for a meeting.

A user could also choose an unfamiliar neighborhood and request coffee shops near a transit stop. Someone planning errands might ask Gemini to identify useful stops within the selected area.

The important distinction is precision. A phrase such as “near downtown” can cover an unclear geographic boundary. A map selection gives the system a defined visual reference.

That does not guarantee a correct answer. It does reduce one source of ambiguity before Gemini begins reasoning about the request.

The workflow also separates area selection from destination search. Users do not need to start with the exact name or address of a business.

They can begin with geography, then let the prompt describe their intent. That structure fits exploratory questions better than a conventional destination field.

The reported rollout is mobile-only. The Map option is not currently available through Gemini’s web interface, even though the web app can use location for relevant responses.

That difference matters because the interaction depends on touch. Panning across a map and adjusting a target area feels natural on a phone used during travel.

It also makes the feature more immediate than copying coordinates, street names, or links between apps. The map becomes an input object like an image or document.

Google has already connected Gemini with geographic information in other settings. Its earlier Maps grounding release gave developers access to information covering more than 250 million places.

The consumer Map tool applies the same broad idea at the interface level. It lets users define geographic context before asking Gemini to interpret it.

However, the new picker should not be confused with full navigation. The reported interface prepares a location-based prompt rather than replacing turn-by-turn directions.

It also does not establish that every request will use live conditions, private saved places, or complete business information. Those capabilities depend on the services Gemini can access.

The most defensible reading is narrower. Google has created a direct way to attach a chosen area to a Gemini conversation.

That addition matters because geographic prompts are often difficult to express precisely. A selected region gives Gemini a clearer boundary for answering them.

It also creates the article’s central tension. Google can offer unusually rich location context, but users still need to understand what data enters the conversation.

Google Is Rebuilding the Prompt Box Around Tools

The prompt box is becoming a launcher for specialized inputs, reusable instructions, and connected services.

The Map option is one part of that transition. The second part appears in Google app beta version 17.63, according to the initial report.

In that beta, typing a forward slash produces a message that the slash is now @. The interface says users can access Skills, Connectors, and more from one place.

Google had only recently introduced the slash as a way to invoke Skills. The rapid switch suggests that Google views a unified tool menu as more important than preserving that command convention.

The @ symbol already has a clear meaning in many digital products. It usually identifies a person, service, agent, or resource that should participate in the current task.

That mental model is useful for Gemini. A user can treat a Skill or connected application as an explicit source of capability rather than an invisible system setting.

The broader interface is also changing its vocabulary. “Connected Apps” are expected to become “Connectors,” while Gems are being replaced by Skills.

Google defines Gemini Skills as reusable custom instructions for recurring tasks and workflows. Gemini can apply a relevant Skill automatically, or the user can specify one.

That model differs from a one-off prompt. A Skill preserves a repeatable method that can be called again when a similar task appears.

Google also says multiple Skills can work together. One could establish an organization’s writing rules while another defines the steps for producing a weekly report.

The @ menu gives those reusable instructions a shared home with external services. It turns the prompt box into a routing layer between the model, saved workflows, and connected data.

Google’s September connected app expansion illustrates the scale of that ambition. The company announced integrations across productivity, creative, and lifestyle categories.

Its examples included project management, database organization, design, website building, exercise planning, apartment search, credit monitoring, and event discovery.

Users can connect supported services in settings. They can then bring one into a conversation with an @ mention or a direct request.

The new interface reduces the distinction between those services and Google’s own reusable Skills. Both become callable resources inside the same conversation.

Map extends that resource model to physical space. A selected area becomes another object that Gemini can use alongside files, photos, notebooks, and connected applications.

This consolidation can lower interface friction. Users no longer need to remember that one resource uses @ while another uses a slash.

It can also make the menu crowded. A single chooser may eventually contain personal Skills, workplace processes, Google services, third-party tools, and media inputs.

Discovery will therefore matter as much as availability. If users cannot predict which resource Gemini selected, consolidation may hide complexity instead of removing it.

Automatic routing creates a related concern. Gemini’s ability to choose an appropriate Skill can save time, but an incorrect choice can silently change the result.

Explicit @ mentions provide a useful counterweight. They let the user identify the intended tool before the model interprets the task.

The interface is still in transition. The Map tool is described as widely available, while the replacement of the slash command is not yet broadly rolled out.

Users may therefore see different controls across accounts, platforms, and app versions. A beta interface should not be treated as a completed global migration.

Google’s direction is clearer than its schedule. The company wants the Gemini prompt box to coordinate a collection of tools, not merely accept prose.

The Gemini Map Tool Highlights Google’s Geographic Advantage

Google’s advantage is not the map picker alone, but the mapping system waiting behind it.

Competing assistants can accept addresses, coordinates, screenshots, and location-related questions. Some can also call web search or external services for current place information.

Google approaches the problem from a different position. It operates Google Maps, distributes Gemini across Android, and manages services that hold useful personal context.

That combination makes location a natural area for product integration. It also gives Google an opportunity to shorten workflows that currently move between a chatbot and a mapping application.

Google began exposing this strategy to developers before the new consumer picker appeared. Its Maps tool for the Gemini API connects model responses with current geospatial information.

The company said that developer product draws from more than 250 million places. That scale gives Gemini a substantial base for restaurant, attraction, service, and itinerary questions.

Google has also been expanding Gemini inside Maps itself. One path starts with navigation and introduces conversational assistance during a journey.

The Gemini Map tool travels in the opposite direction. It starts inside the assistant and imports a chosen geographic area into the conversation.

Those two paths are beginning to meet. Maps is gaining more conversational behavior, while Gemini is gaining more explicit geographic inputs.

That convergence pressures the idea that an AI assistant should remain a separate destination. Google can place Gemini before, during, and after a location-based decision.

Consider a weekend planning task. A user might select an area, ask for several options, compare them against calendar constraints, and then continue into navigation.

The Map attachment handles the first step by defining the search area. Connectors could eventually supply availability, reservations, tickets, or other task-specific information.

Skills could preserve the user’s recurring preferences. A saved instruction might prioritize walking distance, quieter venues, accessibility, or a particular schedule format.

That combination would make the assistant more than a place search box. It would coordinate geographic context with personal rules and connected services.

The current release does not establish that the full workflow works reliably. It only provides several of the interface pieces needed to build one.

This distinction is important. A visually selected area does not eliminate hallucinations, stale listings, missing hours, or questionable recommendations.

Location-based recommendations have many hidden requirements. The system must interpret the boundary, retrieve relevant places, rank them, and explain why they fit.

A mistake at any stage can weaken the answer. A beautifully selected map area offers little value if Gemini overlooks an important constraint in the prompt.

Google’s own mapping depth still changes the competitive calculation. Rivals can integrate mapping providers, but Google controls both the assistant and a major geographic platform.

The company also reaches users on iOS through its app, despite lacking Android-level operating system control there. The Map tool’s arrival on both mobile platforms broadens the test.

Web availability remains a notable gap. Desktop users researching travel or business locations may prefer a larger display and keyboard.

Google could eventually extend the picker to the web, but the reported launch does not promise that. Its absence keeps the current feature focused on mobile use.

The pressure on competitors is therefore specific. They must make location context equally easy to provide, even when they do not own the underlying map.

The pressure on Google is equally real. It must show that integration produces better outcomes, not merely more pathways into existing services.

A Better Location Prompt Carries a Larger Privacy Burden

The Map tool removes ambiguity from the request, but it makes data boundaries more important.

Gemini already uses location in several ways. Google says its apps may use a general location or a device’s precise location when permission is available.

On mobile, those permissions depend partly on the Google app that hosts Gemini’s assistant functions. Platform settings can therefore affect the context Gemini receives.

The new Map interface introduces a more deliberate signal. Users actively choose an area and attach it to their prompt.

That action is easier to understand than invisible background inference. The selected map offers a visible representation of the location being shared.

However, visibility at the attachment stage does not answer every question. Users may still want to know how the prompt is stored, processed, and combined with other information.

Google’s Gemini privacy controls explain that Gemini information can include prompts, uploaded content, connected-app data, device information, and location information.

The documentation also distinguishes between general and precise location. Users should review app permissions and Gemini activity settings before sharing sensitive geographic context.

A map selection can reveal more than a destination. It might identify a home area, workplace, medical facility, school, or repeated travel pattern.

That risk increases when location joins information from email, calendars, files, photos, contacts, or third-party services. Each connection can make a response more useful and more personal.

Google says connecting supported apps remains subject to user controls. Yet control becomes harder to evaluate as the assistant automatically combines several sources.

The unified @ menu may help by making selected resources visible. A user who explicitly invokes a connector has a clearer indication that it is entering the task.

Automatic Skill selection is less obvious. A saved workflow can influence an answer even when the user does not call it directly.

Skills are instructions rather than data sources, but they can shape what Gemini requests or how it uses connected information. That makes transparency important.

A trustworthy interface should show which Skill, connector, or attachment affected the response. It should also make those elements easy to remove before submission.

The Map attachment appears to provide that kind of explicit object in the prompt box. Whether the final interface explains downstream data use remains a separate issue.

Accuracy also belongs in this risk discussion. A selected area may be precise on screen while the underlying interpretation remains imperfect.

Boundaries can cut across neighborhoods, districts, campuses, or business zones. Gemini might treat a rough circle as a strict search radius or only as a general hint.

The feature’s name could create expectations that it performs a complete map search. Users should verify critical details in Google Maps or directly with a venue.

That is especially important for opening hours, accessibility, reservations, road conditions, and travel time. These details can change after Gemini generates a response.

The assistant should also distinguish between recommendations grounded in current Maps information and suggestions based on broader web material. The interface may not always make that provenance obvious.

A second uncertainty concerns rollout consistency. The Map tool reportedly appears widely on Android and iOS, but feature availability can still vary by account.

The @ replacement is earlier in deployment. Users on stable builds may continue seeing the slash menu while beta users receive the consolidated selector.

This staggered state can complicate instructions shared online or within organizations. A workflow documented for one version may not match another person’s screen.

The Gems-to-Skills migration adds another transition. Existing custom assistants are becoming reusable Skills, creating questions about compatibility and preserved behavior.

Google must manage these changes without making familiar workflows feel unstable. Renaming features and changing invocation symbols can impose a small but recurring learning cost.

The company’s bet is that one consistent @ entry point will repay that cost. The success of that bet depends on predictable routing and clear feedback.

The central challenge is not whether Google can put more tools inside Gemini. It is whether users can understand which tool acted, which data it used, and how to correct it.

Three Signals Will Show Whether Google’s Interface Bet Works

The next test is whether a cleaner tool menu produces more reliable, understandable actions.

The first signal is the full rollout of the @ selector. Google needs to move it beyond limited beta availability without fragmenting behavior across Android, iOS, and the web.

A completed rollout would strengthen the view that @ is Gemini’s permanent tool invocation language. Continued inconsistency would weaken the unified-interface argument.

The relevant detail is not just the symbol. Users should see the same basic organization for Skills, Connectors, and other resources across supported devices.

Google must also explain how explicit selection interacts with automatic routing. People need to know when Gemini chose a Skill and when their @ mention overrode that choice.

The second signal is broader availability for the Gemini Map tool. Web support would show that Google sees geographic selection as a core input, not a mobile experiment.

Desktop use could serve travel research, real estate comparisons, logistics, event planning, and location-based business analysis. Those tasks often benefit from a larger map.

More important, Google should reveal what the selected boundary actually controls. Users need confidence that Gemini respects the area rather than treating it as a loose suggestion.

Results should also disclose when current Google Maps information supports a recommendation. Clear grounding would make the tool easier to trust and verify.

The third signal is deeper coordination between Skills and Connectors. Google’s interface becomes more valuable when these resources cooperate without creating hidden behavior.

A user might invoke a planning Skill, attach a Map Area, and call a booking connector. Gemini would then need to preserve every constraint across the workflow.

That is a demanding test. The system must maintain context, call the correct services, present recoverable errors, and avoid taking unintended actions.

Google’s previous expansion of connected services suggests that it wants Gemini to cover more of these workflows. The @ menu gives that growing catalog a common entrance.

The change also reflects a broader movement from chatbots toward tool-using assistants. The model remains central, but the useful work increasingly happens through selected context and external systems.

That shift can benefit knowledge workers beyond travel. A good assistant should let people identify the source, workflow, and boundary relevant to each request.

The same principle applies when working with documents and personal information. A personal knowledge base becomes more useful when users can control which context enters a task.

Google’s map picker offers a concrete version of that idea. Instead of hoping the assistant infers the right place, the user supplies the intended area.

The @ menu extends the principle to capabilities. Instead of relying entirely on automatic selection, the user can call a particular Skill or connector.

Neither change removes the need for judgment. Users still need to verify important recommendations, monitor permissions, and check which services participated.

The Gemini Map tool is therefore less significant as a standalone map widget than as evidence of a changing interface model.

Google is replacing the empty prompt with structured choices. Files, photos, places, saved instructions, and external services can all become visible ingredients.

If those ingredients remain understandable, Gemini can feel more precise without requiring longer prompts. If they become opaque, the interface will merely hide additional complexity.

Watch the @ rollout, web map support, and multi-tool transparency over the coming months. Together, they will show whether Google has built a coherent assistant interface.

For now, mobile users can test the central premise directly. Select a clearly bounded area, ask a constrained question, and compare the answer with Google Maps.

That comparison will reveal more than the presence of a new button. It will show whether Gemini can turn chosen geographic context into a useful, verifiable result.