All news about Android XR, Gemini, Google search and more in less than 5 minutes

15:45

Google I/O 2025: The most important innovations & first assessment

This year’s keynote began with the anecdote that Gemini successfully mastered the game Pokémon Blue. Google has now held its developer conference in 2025 and presented some AI renovations and developments away from Pokemon, including Android XR.

You can find out what it is all about, among other things, in our compact summary of the keynote.

Google I/O: That was presented

Android XR

Google introduced Android XR in more detail, which will sooner or later find its way into smart glasses. According to Samsung’s Project Moohan, the search engine giant has officially presented the second device that will run under this operating system.

However, the glasses shown are still in an early stage of development. Basically, the glasses such as Project Astra should work: You can speak to the gadget while it scans the surroundings with Gemini. In the glasses, microphone, camera and controls are installed in the form of buttons. In addition, information is projected into the field of vision.

Apparently Google does not seem to produce the glasses itself and, like Wear OS. The company fetches manufacturers on board that are ready to handle the hardware for the software. Until Google will bring a product onto the market itself at some point.

Link to Twitter content

Google Beam

With »Google Beam« an innovation for video calls was also announced. With this new chat method, six cameras film each person from different perspectives.

A AI puts together these images into a 3D image that is shown to the chat partner. This creates the impression of a real meeting, according to the company.

This requires special hardware for 3D recording and presentation. HP wants to bring such a device onto the market, initially with a focus on companies.

Translations in Google Meet

Google’s conference tool Google Meet receives a live translation function. At first, however, only the languages ​​of English and Spanish are available. The person said is automatically translated for his counterpart to the selected language. According to Google, the voice, tone and expression of the speaker are adopted.

Gemini Live

As of today, both iOS and Android users can Share your camera or screen in Gemini live for free.

Personal context in Gemini

In the future, Gmail will offer optionally personalized, intelligent answers to emails. Gemini searches emails, notes and drive to find suitable answers based on the previous correspondence.

Project Astra (Konzept-Video)

Google has presented a concept video that shows how Gemini helps to repair a bike. For this purpose, the AI ​​model initially searched and read a suitable guide on the Internet. Then it picked out a YouTube video and finally sent an email to a bike shop to order spare parts. The big goal: Multitasking of the AI.

Project Mariner und Agents

Project Mariner is basically a AI agent who is looking for you on the Internet and can keep an eye on up to ten tasks at the same time, says Google. With the help of the new “Teach and Repeat” function, the program can also learn tasks by showing it a task so that similar todos do it itself in the future. Project Mariner is to be made available to a broader public this summer.

Link to Twitter content

The so-called Ki agents can do various tasks. An example was that the agent finds offers on websites when looking for an apartment and can also use Project Mariner for special filter adjustments. An experimental version of this agent mode should soon be available for subscribers.

AI fashion (Google search)

AI mode is currently limited to the USA, but Google gives us another foretaste of how this mode will change the Google search (optional). AI Mode should be able to answer extensively on long and complex questions.

AI fashion includes more detailed answers with graphicsDescriptions, pictures, web links and more. Basically, you get an all -round question to the question. According to the company, powerful AI models are used to prevent the AI ​​from “hallucinating” the AI.

Virtual wardrobe

With this tool, users can digitally try on clothing from online shops. For this purpose, as Google presented, a photo of the person is uploaded from which a AI derives the body shape and size and creates a digital image.

Based on the material properties of the garment, the AI ​​calculates its case and adapts it to the individual body shape – at least the idea.

Link to Twitter content

In the future, the AI ​​should also be able to shop independently. The users set a price limit for a product. If this limit is below, only authentication is required for payment.

Canvas

Canvas receives an update that offers more options for creating code. In the shortest possible time, “complete new applications” should be created from simple descriptions.

Gemini for Chrome browser

The AI ​​comes for the browser, initially in the USA. The AI ​​assistant helps to understand web content and to do tasks.

Improved BILD-KI

Google’s new AI model »Imagen 4« for generating images can obviously be much better dealt with letters and should also be faster when processing.

Ki-Videogenerator: I see 3

The improved video tool gives people, animals and more to life by giving them a suitable voice and also enables dialogues. Background noises such as forest noises can also be generated at night. The quality of the videos is also said to have been improved according to Google.

Link to Twitter content

With »Flow« Google also presents a new tool for video production. The tool is based on the “progressive AI models from VEO 3” and allows the seamless combination of several video clips to be created by artificial intelligence.

Google also works with the filmmakers of “Primordial Soup Labs”, “to explore the role of AI in filmmaking”.

Google you have ultra

In addition to Google AI Pro, Google offers a new subscription model: AI Ultra for an impressive $ 250 per month. Including are-not surprisingly-all AI applications, YouTube Premium and 30 TB online memory.

Did you follow the Google keynote? What do you think of developments in the field of AI? What was your personal highlight? Feel free to write it in the comments below!