Since its early days, Google intended for Gemini to replace Assistant. This led to criticism, as Gemini was a barebones app compared to the much more capable Google Assistant it aimed to replace. However, things have improved drastically since then, thanks to apps, previously known as extensions.
Table of Contents
What are apps?
Apps are extensions that you enable in Gemini to integrate it with other services and applications. They enhance Gemini’s functionality by allowing it to perform additional tasks. Through these apps, Google is bridging the gap between the old, trusted Google Assistant and Gemini—sometimes even offering a better experience with Gemini’s natural language processing and understanding capabilities as an AI LLM.
Although many apps are enabled by default, some are not. Users can choose to enable the disabled ones or disable those that are enabled by default.
How to Find and Enable/Disable Apps in the Gemini Android App
To see the list of available apps, follow these steps:
- Open the Gemini app.
- Activate the “Signed in as (your account name)” option, typically located near the top right of the screen.
- Find and select “Apps.”
Here, you will see a list of all available apps, previously known as extensions, along with an introduction to the feature. Apps are divided into categories, and you can tap on a category to explore the apps it contains. If you continue navigating, you will also find a list of all categories along with the apps included in each one.
Each app has a toggle to enable or disable it, along with a brief overview and example prompts you can use. Some apps also include a “Learn more” button for additional details. Some may require extra steps for authorization, such as providing consent or linking an account.
Available Apps
The list of apps is gradually expanding as Google adds more integrations and supports additional services and features over time. Below are the available apps at the time of writing. To learn more about them, you can check Google’s help articles or use the “Learn more” option when available.
- Google Workspace: Allows Gemini to perform tasks in Google services like Drive, Docs, and Gmail. It can, for example, summarize content or retrieve specific data from documents.
- Messages: Enables sending SMS messages using your phone’s default messaging app directly through Gemini.
- Phone: Allows initiating phone calls.
- WhatsApp: Enables sending WhatsApp messages and making calls.
- Google Home: Controls smart home appliances.
- Utilities: Provides better integration with your system, allowing you to open apps, change settings, set timers, take pictures, and more.
- Google Flights: Fetches flight information.
- Google Hotels: Helps find hotels.
- Google Maps: Enhances integration with Maps, providing useful information like addresses and directions.
- Spotify: Enables finding and playing music from Spotify through Gemini prompts.
- YouTube Music: Enables finding and playing music from YouTube Music through Gemini prompts.
- YouTube: Helps find and summarize YouTube content.
- OpenStax: Retrieves passages from openly licensed textbooks on OpenStax.
How to Use Apps
Gemini leverages its capabilities as an AI LLM to make it easy for users to access what apps offer without requiring strict, specific prompts. It attempts to understand what is needed based on natural language interpretation and conversational context. Tasks related to apps can be requested through the following methods:
- Calling the Assistant: You can invoke the assistant using your preferred way, such as the wake word or a gesture. Once the assistant is listening, you can speak or type your request.
- Using the Gemini App: While in a conversation within the Gemini app, you can type your prompt or use the microphone button to speak it.
If you want to specify an app, you can use the @ symbol before the app’s name. For example, typing @Spotify followed by your request tells Gemini to use the Spotify app. As you type the app name after @, suggestions should appear above the keyboard, allowing you to select the desired app.
While in a conversation, Gemini may also suggest enabling apps based on your prompts if they are not already enabled.
Typically, there’s no need to follow strict wording or specific terms in your prompts. However, making them clear helps prevent Gemini from misunderstanding your request.
Testing and Observations
The Gemini apps feature enhances its reliability and usefulness as an assistant. With the addition of more apps, Gemini continues to gain functionality, a significant improvement from its early days when its limitations outweighed its capabilities.
Despite being the conventional person who performs actions by hand, not through the voice assistant—which I primarily use to get quick answers or information—I’m finding myself getting used to the ease of relying on the assistant to perform certain tasks, like opening settings sections.
While I haven’t explored all available apps or tested every functionality, I’m sharing some observations based on conducted tests.
It’s important to note that my system and Gemini’s language are set to English and that I use the free Gemini tier.
Utilities
Through the Utilities app, I can increase and decrease volumes, open apps and settings sections, and open websites. Since I didn’t grant Gemini notification access, I can’t comment on reading or interacting with notifications. It’s also possible to take photos; after issuing the prompt, there will be a short timer before the photo is captured.
Despite improvements, Gemini still gets confused at times, and this is also true when using the Utilities extension. For example, I once asked it to turn on Bluetooth, which it did, but when I later asked it to turn it off, it told me it couldn’t and suggested I should do it through settings. I tried different prompts, but it kept refusing to execute the command. After continuing with unrelated prompts, I asked it to check whether Bluetooth was on or off. It told me Bluetooth was on, and when I asked it to turn it off again, it did so without issue. I was then able to enable and disable Bluetooth repeatedly without any further trouble.
Sending WhatsApp messages and starting calls seems to work reliably. One thing I like is the ability to spell out the name if Gemini cannot find the contact due to a different spelling. After Gemini tells me it can’t find the contact, I simply ask it using the correct spelling of the name.
It’s also possible to state what you want to convey in the message and let Gemini write it for you, then ask it to change the tone. When using voice prompts, Gemini reads out the message and then listens again for your response, so you can continue asking it to make edits or tell it to go ahead and send the message.
However, Gemini sometimes seems unsure of its actual abilities in WhatsApp. For example, when I told it to translate something from English to Arabic and send the translated message, it insisted that it couldn’t do this two-step task. It said it would translate the message, but I would need to manually copy and paste it into the chat. I asked it to go ahead with the translation, and once it did, I asked it to send it, and it successfully obeyed my order.
At the time of writing, the WhatsApp Gemini app cannot send voice messages, which was something the old Google Assistant could do — it would send voice messages and also provide their transcriptions. WhatsApp-related tasks in Gemini are currently limited to sending text messages and starting calls.
YouTube
When I asked Gemini about what it could do with YouTube videos, it pointed out that it could summarize and provide specific information from the videos. Feeling excited, I asked if I could give it the link to a specific video, and it gave a positive response. I went ahead and pasted the link in the chat and requested a summary. However, Gemini retreated and told me that it couldn’t summarize the video. When I reminded it of what it had said earlier, it explained that summaries are built on transcriptions available in the videos, not the actual video content. It suggested providing me with information about the video through a search, which didn’t yield much because I selected an unpopular video. I was disappointed, but knowing Gemini’s inconsistency with providing accurate information on what it can and cannot do, I found this expected.
However, it is possible to ask Gemini to find videos on specific topics, and you can also ask it to search for particular channels. But here, another inconsistency arose. After Gemini found the Blind Android Users YouTube channel, I asked it to give me the most recent video. What it did, however, was bring up an older video from the channel. After I pointed out that this wasn’t the most recent video, it apologized and admitted its mistake, stating that the channel is active, but then provided another incorrect statement by saying it publishes both regular and short videos (the channel rarely publishes shorts).
Another limitation in the YouTube app is the inability to play videos directly. No matter what prompt I used, I couldn’t get a video to play without using the screen to start it.
Google Workspace (specifically Gmail)
I only tested the Gmail part of the Workspace app.
Using a prompt to get the most recent message worked, as did retrieving the previous one. However, in one instance, when I asked “What is my last email?”, it asked me for notification permission, which meant it was using the Utilities app. Being more specific by stating that I wanted the Gmail inbox, it performed the task I asked for.
It is possible to retrieve data from a specific email. For example, I asked for a summary of the last received email and then requested data specific to a certain part, and Gemini provided answers.
Going through the entire inbox to extract certain information wasn’t as successful, though. I asked for my last app purchase, and I was surprised to receive a response indicating an app store purchase forwarded to me a few years ago during an email exchange. I told it to get Play Store orders, but it failed to get the last Play Store order. It presented some information from Play Store emails but didn’t give me what I requested.
The most significant limitation, however, is the inability to send and reply to emails. It can compose emails, but it cannot send them. Unlike the WhatsApp translated message, Gemini insisted that it is not able to send emails and that I must do it manually.
Spotify
The following Spotify observations are from Accessible Android teammate and founder, Salih Kunduz: After authorizing his account, he was able to play random songs from his library, stop playback, and move between songs. He was also able to get more information about the currently playing song. Additionally, he could specify songs from different countries, such as asking Gemini to play a Vietnamese or Turkish song. The only task Gemini failed to perform was creating a playlist.
Maps
Asking about locations usually works well. I can inquire about nearby places or places in a specific area. For instance, I can ask if a certain restaurant offers delivery or request a business phone number. During the conversation, it is also possible for Gemini to initiate a call.
No Apps While Using Gemini Live
Currently, app-related tasks can only be carried out through the Gemini app or the assistant directly. While live
conversations can run in the background, task performance is not yet supported. Generally, Gemini clearly states when a feature isn’t supported, such as when inquiring about actions that require apps like checking Gmail. However, there are times when it provides misleading responses. For example, when I asked it to open a specific section of the settings, it stated that it couldn’t but offered to work with general settings and asked if I wanted to try that capability. I agreed and asked for examples, but it couldn’t provide any that worked.
Adding to the frustration, Gemini sometimes pretends to perform tasks and even provides incorrect information. I asked it to turn Bluetooth on, and it confirmed that it had, yet in reality, Bluetooth remained off. I then asked it to turn it off, and it gave another confirmation. Despite repeatedly trying, it kept assuring me that Bluetooth was being toggled, with no actual change to its state. The Bluetooth remained off throughout.
Battery percentage is another area where Gemini Live failed. It consistently gave incorrect battery readings. The only thing it got right was whether the battery was charging or not. Despite the battery being around 20%, Gemini told me it was at 90%. I repeated the question hoping for an apology, but it continued to give the wrong answer. This issue didn’t occur outside of the Live conversation.
Final Thoughts
There is no doubt that Google is determined to improve the capabilities of Gemini as an assistant. The progress from the early versions of Gemini to the current one, especially with the integration of apps, is noticeable. Apps have pushed the service forward, making it finally a serious contender as a reliable default Google Assistant. As the AI language model itself develops, the app features of Gemini will continue to improve. Being able to use conversational prompts to get information or perform tasks takes the assistant to the next level, where the user no longer needs to memorize prompts or worry about the specific terms for a successful request.
While Gemini excels in certain areas, it still shows weaknesses in basic conversations. A pronounced issue is its inability to accurately communicate what it can and cannot do, and it can also be easily confused. These weaknesses are evident in its use of additional apps. Despite these challenges, apps still provide a major productivity boost. Even with some limitations and issues, it’s easy to find yourself relying on them rather than manually completing tasks.
Gemini, as both an AI model and an app, is progressing rapidly, meaning the number of apps and their functionalities will inevitably increase and improve. What is observed in testing today may not reflect the same experience in the near future.
The main goal of this article is to introduce the apps feature in general and explain how to enable and use apps. Real-life results may vary between phone brands and Gemini tiers, and the issues shared in this article should not deter interested users from trying the features themselves. There is a strong possibility that certain issues are temporary or will be resolved as Gemini gains more knowledge and refinements. It’s also important to remember that hiccups and inconsistencies are common at this stage with AI services like Gemini, even though these issues may be more noticeable compared to other popular AI chatbots, which impacts Gemini’s role as an assistant utilizing app integration.
Keep visiting the Gemini apps section, learn more about the apps it offers, and explore the possibilities along with their limitations and restrictions.
Comments