- Google launches Personal Intelligence in Gemini, a feature that connects Gmail, Photos, YouTube, and other apps to provide personalized responses.
- The feature is disabled by default; the user chooses which apps to link, and Google insists that it does not train its models with emails or photos.
- For now, it's only available in beta for Google AI Pro and AI Ultra subscribers in the United States, with plans to expand to more countries and the free plan.
- Gemini can reason between multiple sources (text, images, and video) for tasks such as planning trips, finding license plates, or suggesting purchases, although Google admits that it can still make mistakes.
Google has taken another step in its commitment to artificial intelligence applied to everyday life with the launch of a new feature called "Personal Intelligence" within Gemini:With this move, the assistant ceases to be a simple generic chatbot, instead relying on the information we already have in the company's services to offer answers much more tailored to each person.
The idea is that Gemini can connect in a controlled manner to gmailGoogle Photos, YouTube, or Search itself to better understand the context of each query. All of this is done under a system of explicit permissions, disabled by default, and with a strong message from Google regarding privacy and the limited use of personal data.
What is Google's "Personal Intelligence" and what is its purpose?
With this feature, the company wants Gemini to go from knowing things about the world to also understand fragments of each user's digital lifeInstead of simply offering general information, the assistant can cross-reference what's already in your email, your photos, or your video history to give you a tailored answer.
According to Google's explanation, the feature relies on two pillars: firstly, the ability to reasoning between complex sourcesOn the other hand, there is the ability to extract very specific details from an email, a photo, or a video when needed. Gemini combines both to work simultaneously with text, images, and audiovisual content.
The company presents this Personal Intelligence as a system of Gemini advanced customizationIt can connect to applications within the Google ecosystem to better understand each user's needs. This aligns with a general trend in the sector: major tech companies like Google, Microsoft, Apple, and Anthropic agree that the future of AI lies in assistants that deeply understand the person in front of them.
In practice, Google wants its assistant to become a kind of organized digital memorycapable of retrieving data that the user no longer remembers, reorganizing scattered information, and proposing proactive actions based on previous habits and interests.
How Personal Intelligence Works and How to Activate It
Personal Intelligence is not switched on by default: It is an optional function that must be explicitly activated.When the user enables it, they can decide quite precisely which services they want to connect to Gemini and which they prefer to keep completely separate.
Google explains that, when you turn it on, Gemini only accesses app data to respond to specific requests. or to carry out a requested task, such as searching for an old email, locating a license plate among photos, or reconstructing travel information. This is not permanent, "anything goes" access, but rather limited use depending on each query.
In the settings, the user can choose connected apps (for example, Gmail, Photos, or YouTube) and change their mind at any time. The entire system is designed so that the connection is reversible, allowing the user to completely turn off the feature if they lose trust in it or simply no longer need it.
For those who already have access in the United States, the process is quite simple: just open Gemini, enter the section of Configurationtap on "Personal Intelligence" and, from there, select the apps that want to be linkedFor now, it only works with personal Google accounts, excluding Workspace business or education accounts.
Once set up, Google promises that the Gemini experience will become more proactive and contextual, with an assistant that anticipates certain needs by better understanding the person's situation and the environment of the consultation.
Practical examples: from car registration to family holidays
To illustrate the scope of Personal Intelligence, Google has shared several use cases that go beyond the typical generic response. The company wants to make it clear that Gemini doesn't just search for data, but that reason about them and combines them in ways that were not possible before.
One of the most repeated examples is that of a user who needs to change the your car's tiresBut it doesn't remember the measurements or the exact model. Instead of offering a standard table, Gemini can check old emails in Gmail where invoices or vehicle documentation appear. Identify the license plate in Google Photos and, with that information, suggest suitable tires, approximate prices and recommendations adapted to the type of journeys seen in the travel photos.
In the same scenario, if the user is unsure when they last took the car to the garage, Gemini could search for the date in the email history or in photos showing previous visits to the mechanic, and use that data to suggest whether or not it's a good time for a check-up.
Another case that Google highlights is that of the travel planningInstead of simply suggesting the usual destinations, the assistant can analyze emails related to past bookings, vacation photo albums, and preferences revealed by YouTube activity. From there, Gemini can design itineraries that avoid overcrowded places and better suit what the family typically enjoys, even including suggestions for board games or quiet activities if it detects that these are part of their regular routines.
Beyond travel, Personal Intelligence is also useful for more everyday tasks, such as quickly find a license plate among thousands of images, locate an attached document buried in an email chain or recommend readings, series and clothing based on previous purchases and content saved by the user.
Privacy and security: disabled by default and not trained on your emails
The rollout of a feature that accesses such sensitive data as private emails or family photo albums is, obviously, delicate. That's why Google insists time and again that the Privacy is the central issue of this Personal Intelligence and which has been designed with several layers of protection.
The first thing the company emphasizes is that The function is turned off by default.The user must intentionally activate it and grant specific permissions. Nothing connects without consent, and access can be revoked or data associated with a conversation deleted at any time if the user prefers that it not influence the assistant's future behavior.
Furthermore, Google claims that Gemini does not train directly with the raw content of the Gmail inbox or with Google Photos librariesAccording to the official explanation, the model is improved with limited information, such as the instructions written in Gemini and the generated responses, after applying filtering and obfuscation processes to prevent personal data from becoming identifiable training material.
In other words, the system learns to find the license plate when askedBut not to memorize and reproduce it without reason. The data is used as a reference at the time of consultation, not as a permanent basis for model formation.
Another key element is transparency: in its responses, Gemini tries to Please indicate where you obtained the information from.The user can see if a recommendation is based on a specific email, photo, or YouTube video, allowing them to review the source and correct any misunderstandings about how the assistant is interpreting their data.
Errors, limitations, and the role of user feedback
Despite the project's ambition, Google acknowledges that Personal Intelligence It is still in beta phase And that, at this point, it is not infallible. The team responsible, including Google Labs Vice President Josh Woodward, admits that there will be inaccurate answers and cases where the AI "overinterprets" personal information.
One example the company mentions is that of a user with hundreds of photos on a golf courseGemini might infer that the person is a big fan of that sport, when in reality they may have simply accompanied a family member or attended a specific event. In other words, the observer could form an inaccurate impression of the person's true tastes based on misinterpreted signals.
To mitigate these kinds of errors, Google encourages users to participate directly in debugging the system: if Gemini makes a mistake, it can be fixed. correct in the conversation itself with phrases like "I don't like golf" or using the negative rating button. This feedback will help adjust the understanding of interests, personal relationships, and other nuances.
Clear limitations are also contemplated on sensitive issues, especially those related to health, finances, or other sensitive areasGoogle states that it has introduced security barriers to prevent Gemini from drawing conclusions from private data in these areas without the user's explicit permission.
In any case, the company admits that getting AI to understand correctly remains a significant challenge. complex human relationships, such as distinguishing between friends, family members or ex-partners in a photo album, and that this will be one of the great technical challenges for Personal Intelligence to be truly reliable in the long term.
Where is it available and what plans does Google have for Europe?
For now, Personal Intelligence is taking its first steps in a rather limited way: It is only offered in beta version to subscribers of Google AI Pro and AI Ultra in the United StatesIt works both on the web and in the Gemini apps for Android and iOS, and is being rolled out gradually to eligible users. More details on how to try the Pro versions can be found in Gemini Pro testing guides.
Google itself has indicated that the function It is not yet available in Spain or the rest of EuropeAnd for now, it's also excluded from Workspace's business and education accounts. The company prefers to start with personal accounts, where control over data rests directly with the end user.
In its roadmap, the technology company indicates that, over time, Personal Intelligence will be expanded to more countries and the free plan from Gemini. Furthermore, the intention is to integrate it into Google Search's "AI Mode," which would allow for even more context-sensitive responses when queries are made through the search engine. This integration is linked to the changes the search engine is introducing with Gemini and advanced AI features.
The arrival in Europe will not be trivial, as Google will have to deal with stricter data protection regulationssuch as the GDPR or the new regulations on artificial intelligence that are being developed in the European Union. It is expected that, before activating the feature in European territory, the company will adapt some aspects of the collection, storage, and processing of personal information.
For now, users in Spain and other European countries will have to wait until the initial testing phase is complete in the United States, as Google details how it plans to comply with local regulatory requirements for such a sensitive function.
A move amid the race for AI personal assistants
The launch of Personal Intelligence in Gemini occurs in a context of intense competition in the generative AI sectorGoogle is not the only one trying to turn its assistant into a truly personal tool connected to each user's digital ecosystem.
Microsoft, for example, has been equipping its Copilot platform with long-term memory and integration with third-party servicesincluding the ability to work with Google Drive files or Gmail emails. Anthropic, for its part, has presented solutions like Claude aimed at managing documents and data for non-technical users.
In the case of Apple, the company has announced its Apple Intelligence system and is preparing a Siri's profound evolution, partly supported by an alliance with Google itself to use models like Gemini in some scenarios. In this framework, Google's Personal Intelligence is also interpreted as a response to the expectations surrounding the future of Siri and other integrated device assistants.
What all these proposals have in common is the intention to move from an AI that responds generically to one that will better understand each person's private contextThe big question is to what extent users will be willing to grant access to so much personal information in exchange for that convenience.
Google is trying to leverage the advantage of its cloud ecosystem, where services like Gmail or Google Photos are concentrated, in many cases, years of mail, family memories, and important documentsIf it manages the issues of privacy and accuracy well, the company could position itself advantageously over its rivals in the race for the ultimate personal assistant.
With this Personal Intelligence, Gemini: It's entering a new phase that combines a high degree of personalization with a strong focus on privacy and a phased rollout. It remains to be seen how all this will translate when it arrives in Spain and the rest of Europe, but if Google delivers on its promises, we're heading towards a generation of assistants that no longer just answer individual questions, but are capable of... accompany and structure a good part of our digital life.
