Google Unveils Beam: Real-Time Translation and Next-Gen AI Break Language Barriers

As Google I/O 2025 gets underway, CEO Sundar Pichai has made three significant announcements that are focused on the needs of consumers.

These efforts, which are driven by Google's Gemini AI models, include Project Mariner, which updates agentic AI capabilities that can assist users in automating tasks.

It also includes Google Beam (formerly Project Starline) updates and real-time speech translation in Google Meet; and the 'better' Project Astra, which will be rebranded as Gemini Live camera conversation. Pichai revealed that HP will be bringing Google Beam, formerly known as Project Starline, to computers.

In essence, it is a technological project created to improve remote communication by making it more organic and engaging. This technology simulates in-person interactions during video chats by combining artificial intelligence, 3D imagery, and other cutting-edge technologies.

Key Features of Google Beam

Pichai emphasised that Google Beam's primary advantages include lifelike depiction through realistic 3D imaging and the ability to make authentic eye contact, which is frequently lacking in traditional video chats.

The project's second component is real-time speech translation in Google Meet, which enables communication between persons who do not speak the same language.

Additionally, Google said that the AI-generated translation will maintain user voices, tones, and facial expressions. Subscribers can presently access the functionality in both Spanish and English. Later this year, Google plans to test this capability with Workspace users in Portuguese, German, and Italian, among other languages.

Gemini Live for iPhones and Android

Project Astra, a fascinating research initiative that was initially presented at I/O, examines the potential of a universal AI assistant that can comprehend your surroundings.

The camera and screen-sharing features of Project Astra are now integrated into Gemini Live. From preparing for interviews to training for marathons, people are utilising technology in intriguing ways.

All Android users currently have access to this feature, and iOS users will begin to receive it from now on.

Project Mariner

Updates to Project Mariner, a Chrome plugin created by Google DeepMind that enables users to automate web browsing tasks using an AI agent, are the third major announcement.

An AI agent that can comprehend and interact with websites to do tasks like research, booking, and purchasing is created using the Gemini AI paradigm. According to Pichai, the AI tool can now handle up to ten jobs concurrently.

For instance, the Gemini app's new Agent Mode will enable users to accomplish even more. If they are looking for a flat, it will be helpful to locate properties on websites such as Zillow that fit their requirements, modify filters, and use MCP to view the listings and even arrange a tour.

Subscribers will soon be able to access an experimental version of Agent Mode via the Gemini app. Additionally, it helps businesses like Zillow by increasing conversion rates and attracting new clients.