Google DeepMind has unveiled a new AI system that could change how Deaf and hard-of-hearing people interact with their phones. SL2T, short for sign-language-to-text, can convert sign language into text and is coming first to Pixel 11. The model will power new features in Gboard and Live Transcribe, allowing users to sign in places where they would normally type. Google says the first supported language pair will be American Sign Language (ASL) to English, with more languages and devices planned.The announcement is significant because Google is moving sign-language AI beyond a research demonstration and into a consumer product. Google DeepMind describes SL2T as a breakthrough model designed specifically for Deaf and hard-of-hearing users.&nbsp;

<blockquote class="twitter-tweet">SL2T is our breakthrough sign language-to-text model powering new features for Deaf and hard of hearing users on <a href="https://x.com/Android?ref_src=twsrc%5Etfw">@Android</a>. Starting with American Sign Language-to-English on Pixel 11, people can sign directly into Gboard and Live Transcribe instead of typing. <a href="https://t.co/p9Vx7tJtLT">pic.twitter.com/p9Vx7tJtLT</a>&mdash; Google DeepMind (@GoogleDeepMind) <a href="https://x.com/GoogleDeepMind/status/2087541213284946191?ref_src=twsrc%5Etfw">August 12, 2026</a></blockquote> <script async src="https://platform.x.com/widgets.js" charset="utf-8"></script>

<h2>How Google's sign-language AI works</h2>SL2T is designed to understand more than individual hand gestures. Sign language also relies on movement, facial expressions and other parts of the body, making it much harder for AI to interpret than a simple gesture-recognition system. The model can turn signing into text that users can then use for everyday tasks. That includes searching the web, writing messages or documents, and interacting with Gemini. In Live Transcribe, users can also sign responses during conversations rather than repeatedly typing them out. Google says the system was developed with input from members of the Deaf community, helping the technology address real-world signing rather than relying only on laboratory demonstrations.&nbsp;<h2>Designed for real-world signing</h2>One challenge is that people do not always sign in perfect conditions. A user may be holding a phone, change their position or sign with one hand.Google's model has been designed to handle such situations, according to details shared around the launch. It also considers facial and body movements that carry meaning in sign language. The system's approach is particularly important because sign languages are full languages, not simply hand gestures translated word-for-word into English.&nbsp;<h2>Pixel 11 is only the beginning</h2>SL2T will initially support ASL-to-English on Pixel 11, but Google says the technology will expand to additional devices and languages.That could eventually make sign-language input a more common part of Android, rather than a specialist accessibility feature. Google's broader AI strategy is increasingly focused on making models interact with the physical world and different forms of human communication. Its Gemini family already supports multimodal inputs such as text, images, video and audio. With SL2T, Google is adding another form of communication to that list giving users a way to communicate with a phone through sign language rather than speech or typing.&nbsp;

Google DeepMind has introduced SL2T, a sign-language-to-text AI model coming first to Pixel 11. It will let Deaf and hard-of-hearing users sign directly into Gboard and Live Transcribe for tasks such as searching, messaging and interacting with Gemini.


Google DeepMind has unveiled a new AI system that could change how Deaf and hard-of-hearing people interact with their phones. SL2T, short for sign-language-to-text, can convert sign language into text and is coming first to Pixel 11. The model will power new features in Gboard and Live Transcribe, allowing users to sign in places where they would normally type. Google says the first supported language pair will be American Sign Language (ASL) to English, with more languages and devices planned. The announcement is significant because Google is moving sign-language AI beyond a research demonstration and into a consumer product. Google DeepMind describes SL2T as a breakthrough model designed specifically for Deaf and hard-of-hearing users.

Google DeepMind has introduced SL2T, a sign-language-to-text AI model coming first to Pixel 11. It will let Deaf and hard-of-hearing users sign directly into Gboard and Live Transcribe for tasks such as searching, messaging and interacting with Gemini.

Google DeepMind unveils AI that can understand sign language

Google is taking another major step towards building more capable humanoid robots. The company has unveiled Gemini Robotics 2, a new artificial intelligence model that acts as a high-level brain for robots. Instead of simply following pre-programmed instructions, robots powered by the new system can understand spoken commands, reason about their surroundings and perform complex physical tasks while adapting to changing environments. The launch highlights Google's growing ambition to bring Gemini beyond smartphones and computers into the physical world, where AI-powered robots could eventually work in warehouses, factories, hospitals and even homes.

<blockquote class="twitter-tweet" data-media-max-width="560">One brain. For any robot. 🤖 We’re launching Gemini Robotics 2: our next-generation physical AI bringing full body intelligence to humanoids, advanced dexterity, multi-robot teamwork and more. <a href="https://t.co/1rFw3NUytb">pic.twitter.com/1rFw3NUytb</a>&mdash; Google DeepMind (@GoogleDeepMind) <a href="https://x.com/GoogleDeepMind/status/2082844162928381956?ref_src=twsrc%5Etfw">July 30, 2026</a></blockquote> <script async src="https://platform.x.com/widgets.js" charset="utf-8"></script>

<h2>A smarter brain for humanoid robots</h2>Traditional robots are excellent at repeating the same task but often fail when something unexpected happens. Google says Gemini Robotics 2 is designed to solve that problem by giving robots the ability to understand context, plan actions and adjust their behaviour in real time. Rather than programming every movement individually, developers can teach robots to reason through tasks much like humans do.One of the biggest upgrades is full-body control. Unlike the previous generation, which mainly controlled a robot's upper body, Gemini Robotics 2 allows humanoids to walk, crouch, balance, bend and use both hands simultaneously. Google also says the model improves dexterity, enabling five-finger robotic hands to perform delicate actions such as tying knots, sealing zip-lock bags and handling small objects, while also supporting industrial grippers for warehouse work.&nbsp;<h2>Gemini Robotics ER 2 plans the work</h2> Alongside the main model, Google introduced Gemini Robotics ER 2, which functions like a project manager for robots. Instead of controlling motors directly, ER 2 understands spoken instructions, breaks large jobs into smaller steps, monitors progress and replans tasks if something changes. Google says this enables robots to complete long workflows involving hundreds of decisions and even coordinate with other robots. The company also launched Gemini Robotics On-Device 2, allowing robots to run AI locally without relying on cloud connectivity. Developers can reportedly customise the model for new robotic platforms in just a few hours using fewer than 200 training examples.&nbsp;<h2>Safety remains a priority</h2>Google says safety has been built into every layer of the platform. The company introduced a new benchmark called ASIMOV-Agentic to evaluate whether robots can refuse unsafe instructions, recognise when human intervention is required and avoid dangerous situations. Robots can also detect nearby people, pause automatically and resume work only when it is safe. As competition in humanoid robotics accelerates with companies including Tesla, Figure AI, Agility Robotics and Unitree all developing increasingly capable machines Google is betting that the future will depend not just on better hardware, but on smarter AI that allows robots to think, adapt and collaborate in the real world.

Google DeepMind has launched Gemini Robotics 2, an AI model designed to give humanoid robots advanced reasoning, full-body movement and task-planning abilities. 

Google is taking another major step towards building more capable humanoid robots. The company has unveiled Gemini Robotics 2, a new artificial intelligence model that acts as a high-level brain for robots. Instead of simply following pre-programmed instructions, robots powered by the new system can understand spoken commands, reason about their surroundings and perform complex physical tasks while adapting to changing environments. The launch highlights Google's growing ambition to bring Gemini beyond smartphones and computers into the physical world, where AI-powered robots could eventually work in warehouses, factories, hospitals and even homes.

Google DeepMind has launched Gemini Robotics 2, an AI model designed to give humanoid robots advanced reasoning, full-body movement and task-planning abilities. The company also introduced Gemini Robotics ER 2 and On-Device 2, expanding its vision for smarter, safer robots that can work both online and offline.

Google unveils Gemini Robotics 2: New AI brain lets humanoid robots walk, think and complete complex tasks

Google has entered the AI smart glasses market with a new product powered by Gemini, marking a direct challenge to Meta, which currently leads the segment.The announcement was made at Google I/O 2026, where the company introduced its “intelligent eyewear” built in partnership with Samsung, Warby Parker and Gentle Monster. The product is expected to launch in fall 2026, though pricing details have not yet been disclosed.<h2>What Google’s AI smart glasses can do</h2>The glasses are designed around an audio-first experience and do not include a display inside the lens. Instead, they rely on voice interaction powered by Gemini AI.Key features include:Real-time translation across languagesNavigation with voice-based directionsReading signs and identifying surroundings using camera inputManaging calls, messages and notificationsTaking photos and editing them using AICompleting tasks like ordering food or booking ridesUsers can activate the assistant by saying “Hey Google” or tapping the frame.iPhone support expands market reachOne of the major highlights is that the glasses will support both Android and iPhone devices. This is a strategic move by Google, as it allows the company to reach a wider audience instead of limiting the product to Android users. Industry reports suggest that cross-platform support is important in this category, especially as Meta’s smart glasses already work across both ecosystems.<h2>Partnership-led hardware strategy</h2>Unlike its earlier Google Glass project, the company is now working with established eyewear brands to improve design and usability.&nbsp;Warby Parker will focus on practical and prescription-friendly frames Gentle Monster will bring fashion-focused designs Samsung is handling hardware engineering and productionGoogle has also confirmed future collaborations, including a luxury segment with Gucci planned for 2027.<h2>Competition with Meta’s smart glasses</h2>Meta currently leads the smart glasses market, with around 85 per cent global share and over 7 million units sold in 2025, according to industry estimates. Google’s main advantage lies in Gemini AI, which can handle more complex, multi-step tasks compared to existing systems. However, Meta already has an established user base and hardware presence, giving it a strong position in the market.<h2>What comes next in AI wearables</h2>Google has indicated that this is just the first step. A future version with a built-in display is expected around 2027, which could show visual information directly in the user’s field of view. The company is also working on extended reality (XR) devices under its Android XR platform, aiming to build a broader ecosystem around AI-powered wearables.<h2>What this means for the tech industry</h2>The launch signals growing competition in the AI wearable space, with companies focusing on integrating AI into everyday devices.Instead of traditional screens, the focus is shifting towards voice-based interaction and real-time assistance.<h2>What happens next</h2>The glasses will roll out gradually after launch, with more details on pricing and availability expected closer to release. As companies like Google, Meta and others continue to invest in AI wearables, the market is likely to expand in the coming years, shaping how users interact with technology in daily life

Google has entered the AI smart glasses market with a new product powered by Gemini, marking a direct challenge to Meta, which currently leads the segment.


Google has entered the AI smart glasses market with a new product powered by Gemini, marking a direct challenge to Meta, which currently leads the segment. The announcement was made at Google I/O 2026, where the company introduced its “intelligent eyewear” built in partnership with Samsung, Warby Parker and Gentle Monster. The product is expected to launch in fall 2026, though pricing details have not yet been disclosed. What Google’s AI smart glasses can do The glasses are designed around an audio-first experience and do not include a display inside the lens. Instead, they rely on voice interaction powered by Gemini AI. Key features include: Real-time translation across languages Navigation with voice-based directions Reading signs and identifying surroundings using camera input Managing calls, messages and notifications Taking photos and editing them using AI Completing tasks like ordering food or booking rides Users can activate the assistant by saying “Hey Google” or tapping the frame. iPhone support expands market reach One of the major highlights is that the glasses will support both Android and iPhone devices. This is a strategic move by Google, as it allows the company to reach a wider audience instead of limiting the product to Android users. Industry reports suggest that cross-platform support is important in this category, especially as Meta’s smart glasses already work across both ecosystems. Partnership-led hardware strategy Unlike its earlier Google Glass project, the company is now working with established eyewear brands to improve design and usability. Warby Parker will focus on practical and prescription-friendly frames Gentle Monster will bring fashion-focused designs Samsung is handling hardware engineering and production Google has also confirmed future collaborations, including a luxury segment with Gucci planned for 2027. Competition with Meta’s smart glasses Meta currently leads the smart glasses market, with around 85 per cent global share and over 7 million units sold in 2025, according to industry estimates. Google’s main advantage lies in Gemini AI, which can handle more complex, multi-step tasks compared to existing systems. However, Meta already has an established user base and hardware presence, giving it a strong position in the market. What comes next in AI wearables Google has indicated that this is just the first step. A future version with a built-in display is expected around 2027, which could show visual information directly in the user’s field of view. The company is also working on extended reality (XR) devices under its Android XR platform, aiming to build a broader ecosystem around AI-powered wearables. What this means for the tech industry The launch signals growing competition in the AI wearable space, with companies focusing on integrating AI into everyday devices. Instead of traditional screens, the focus is shifting towards voice-based interaction and real-time assistance. What happens next The glasses will roll out gradually after launch, with more details on pricing and availability expected closer to release. As companies like Google, Meta and others continue to invest in AI wearables, the market is likely to expand in the coming years, shaping how users interact with technology in daily life

Google has entered the AI smart glasses market with a new product powered by Gemini, marking a direct challenge to Meta, which currently leads the segment.

Google unveils AI smart glasses with Gemini, targets Meta’s dominance

Google has accidentally revealed details of its upcoming AI assistant, COSMO, after an experimental app briefly appeared on the Play Store. The app, which was around 1.13 GB in size, was uploaded on May 1, 2026, and quickly downloaded and analysed before being removed. The leak comes just weeks before Google I/O 2026, where the company is expected to showcase its latest AI developments. According to reports, the listing was not meant for public release, but it has already given a clear idea of Google’s next direction in AI assistants.&nbsp;<h2>What is Google COSMO and how is it different?</h2>COSMO appears to be very different from AI assistants. Instead of waiting for user commands, it is designed to act as a system that can understand context and assist users automatically. In simple terms, it works like an AI layer inside the device that can observe what the user is doing and offer help when needed. This marks a shift from reactive assistants, which respond only when asked, to proactive systems that can anticipate actions. The goal is to reduce the need for manual inputs and make everyday tasks easier.<h2>Powered by on-device and hybrid AI system</h2>One of the key aspects of COSMO is its ability to run directly on the device. It uses Gemini Nano, which allows it to function locally on Android devices without depending fully on the internet. This means faster responses and better privacy in certain cases. At the same time, COSMO also uses a hybrid system that can switch to cloud-based processing for more complex tasks. This combination of local and cloud AI helps balance speed, accuracy and performance, making the assistant more flexible in real-world use.&nbsp;<h2>Real-time screen awareness and smart assistance</h2>Another important feature of COSMO is its ability to understand what is happening on the user’s screen. By using Android system tools, the assistant can analyse content in real time and provide relevant suggestions. For example, it can recognise conversations, detect tasks, and offer help without the user needing to ask. This level of awareness allows COSMO to assist in a more natural way, making interactions smoother compared to earlier AI tools.<h2>Key features revealed in the leak</h2>The leaked version of COSMO shows several practical features that highlight its capabilities. It can suggest calendar events when it detects plans being discussed, find photos based on user references, and help write or summarise documents. It can also provide summaries of conversations and compile information from multiple sources for research tasks. Another notable feature is browser automation, where the assistant can perform actions on websites instead of just providing answers. These features suggest that COSMO is designed to complete tasks, not just respond to queries.<h2>What the leak suggests about Google’s AI strategy</h2>The accidental release of COSMO gives insight into Google’s broader AI strategy. The company appears to be focusing on assistants that are more deeply integrated into devices and capable of taking action. This includes combining on-device intelligence with cloud systems and building assistants that understand context better. Industry trends also show a similar direction, where AI tools are becoming more proactive and useful in daily workflows rather than just being conversational tools.&nbsp;It is still unclear how Google will officially introduce COSMO. The assistant may launch as a standalone product or become part of existing services like Gemini. Since the leaked version appears unfinished, changes are likely before any public release. However, one thing is clear: the future of AI assistants is moving towards systems that can understand users better, act independently, and help complete tasks efficiently. With Google I/O 2026 approaching, COSMO has already set expectations for what could be announced next.&nbsp;&nbsp;

What is Google COSMO and how is it different?COSMO appears to be very different from AI assistants. Instead of waiting for user commands, it is designed to act as a system that can understand context and assist users automatically. In simple terms, it works like an AI layer inside the device that can observe what the user is doing and offer help when needed. This marks a shift from reactive assistants, which respond only when asked, to proactive systems that can anticipate actions. The goal is to reduce the need for manual inputs and make everyday tasks easier.<h2>Powered by on-device and hybrid AI system</h2>One of the key aspects of COSMO is its ability to run directly on the device. It uses Gemini Nano, which allows it to function locally on Android devices without depending fully on the internet. This means faster responses and better privacy in certain cases. At the same time, COSMO also uses a hybrid system that can switch to cloud-based processing for more complex tasks. This combination of local and cloud AI helps balance speed, accuracy and performance, making the assistant more flexible in real-world use. <h2>Real-time screen awareness and smart assistance</h2>Another important feature of COSMO is its ability to understand what is happening on the user’s screen. By using Android system tools, the assistant can analyse content in real time and provide relevant suggestions. For example, it can recognise conversations, detect tasks, and offer help without the user needing to ask. This level of awareness allows COSMO to assist in a more natural way, making interactions smoother compared to earlier AI tools.<h2>Key features revealed in the leak</h2>The leaked version of COSMO shows several practical features that highlight its capabilities. It can suggest calendar events when it detects plans being discussed, find photos based on user references, and help write or summarise documents. It can also provide summaries of conversations and compile information from multiple sources for research tasks. Another notable feature is browser automation, where the assistant can perform actions on websites instead of just providing answers. These features suggest that COSMO is designed to complete tasks, not just respond to queries.<h2>What the leak suggests about Google’s AI strategy</h2>The accidental release of COSMO gives insight into Google’s broader AI strategy. The company appears to be focusing on assistants that are more deeply integrated into devices and capable of taking action. This includes combining on-device intelligence with cloud systems and building assistants that understand context better. Industry trends also show a similar direction, where AI tools are becoming more proactive and useful in daily workflows rather than just being conversational tools. It is still unclear how Google will officially introduce COSMO. The assistant may launch as a standalone product or become part of existing services like Gemini. Since the leaked version appears unfinished, changes are likely before any public release. However, one thing is clear: the future of AI assistants is moving towards systems that can understand users better, act independently, and help complete tasks efficiently. With Google I/O 2026 approaching, COSMO has already set expectations for what could be announced next.

Google has accidentally revealed its upcoming AI assistant COSMO after a test app briefly appeared on the Play Store ahead of Google I/O 2026. The leak suggests a major shift from traditional AI to proactive, context-aware AI that can understand user activity and act automatically. 

Google has accidentally revealed details of its upcoming AI assistant, COSMO, after an experimental app briefly appeared on the Play Store. The app, which was around 1.13 GB in size, was uploaded on May 1, 2026, and quickly downloaded and analysed before being removed. The leak comes just weeks before Google I/O 2026, where the company is expected to showcase its latest AI developments. According to reports, the listing was not meant for public release, but it has already given a clear idea of Google’s next direction in AI assistants. What is Google COSMO and how is it different? COSMO appears to be very different from AI assistants. Instead of waiting for user commands, it is designed to act as a system that can understand context and assist users automatically. In simple terms, it works like an AI layer inside the device that can observe what the user is doing and offer help when needed. This marks a shift from reactive assistants, which respond only when asked, to proactive systems that can anticipate actions. The goal is to reduce the need for manual inputs and make everyday tasks easier. Powered by on-device and hybrid AI system One of the key aspects of COSMO is its ability to run directly on the device. It uses Gemini Nano, which allows it to function locally on Android devices without depending fully on the internet. This means faster responses and better privacy in certain cases. At the same time, COSMO also uses a hybrid system that can switch to cloud-based processing for more complex tasks. This combination of local and cloud AI helps balance speed, accuracy and performance, making the assistant more flexible in real-world use. Real-time screen awareness and smart assistance Another important feature of COSMO is its ability to understand what is happening on the user’s screen. By using Android system tools, the assistant can analyse content in real time and provide relevant suggestions. For example, it can recognise conversations, detect tasks, and offer help without the user needing to ask. This level of awareness allows COSMO to assist in a more natural way, making interactions smoother compared to earlier AI tools. Key features revealed in the leak The leaked version of COSMO shows several practical features that highlight its capabilities. It can suggest calendar events when it detects plans being discussed, find photos based on user references, and help write or summarise documents. It can also provide summaries of conversations and compile information from multiple sources for research tasks. Another notable feature is browser automation, where the assistant can perform actions on websites instead of just providing answers. These features suggest that COSMO is designed to complete tasks, not just respond to queries. What the leak suggests about Google’s AI strategy The accidental release of COSMO gives insight into Google’s broader AI strategy. The company appears to be focusing on assistants that are more deeply integrated into devices and capable of taking action. This includes combining on-device intelligence with cloud systems and building assistants that understand context better. Industry trends also show a similar direction, where AI tools are becoming more proactive and useful in daily workflows rather than just being conversational tools. It is still unclear how Google will officially introduce COSMO. The assistant may launch as a standalone product or become part of existing services like Gemini. Since the leaked version appears unfinished, changes are likely before any public release. However, one thing is clear: the future of AI assistants is moving towards systems that can understand users better, act independently, and help complete tasks efficiently. With Google I/O 2026 approaching, COSMO has already set expectations for what could be announced next.

Google has accidentally revealed its upcoming AI assistant COSMO after a test app briefly appeared on the Play Store ahead of Google I/O 2026. The leak suggests a major shift from traditional AI to proactive, context-aware AI that can understand user activity and act automatically.&nbsp;

Google accidentally reveals COSMO, an AI assistant that can think and act for users without commands: Here’s what it can do

Google DeepMind has introduced Gemma 4, a new generation of open AI models designed for advanced reasoning and autonomous AI workflows. The release represents Google’s latest push in the open-model ecosystem, offering developers more powerful AI tools while allowing broad use under the Apache 2.0 open-source licence.According to Google DeepMind’s official blog, Gemma 4 is designed to deliver higher intelligence with fewer parameters, making it more efficient for real-world applications.

<blockquote class="twitter-tweet">Today, we’re launching Gemma 4, our most intelligent open models to date. Built with the same breakthrough technology as Gemini 3, Gemma 4 brings advanced reasoning to your personal hardware and devices. Here’s what Gemma 4 unlocks for developers: — Intelligence-per-parameter:… <a href="https://t.co/JgwRZvQHgF">pic.twitter.com/JgwRZvQHgF</a>&mdash; Google AI (@GoogleAI) <a href="https://twitter.com/GoogleAI/status/2039735543068504476?ref_src=twsrc%5Etfw">April 2, 2026</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

<h2>Four model sizes designed for different workloads</h2>Google released Gemma 4 in four model configurations, each designed for specific computing needs.The models include:E2B (effective 2 billion parameters) - lightweight model for edge devicesE4B (effective 4 billion parameters) - improved performance for mobile environments26B Mixture-of-Experts (MoE) - optimised for complex reasoning tasks31B Dense model - the most powerful version in the Gemma 4 seriesAccording to benchmark rankings, the 31B model currently holds the third position among open AI models on the Arena AI leaderboard, while the 26B model ranks sixth, outperforming models estimated to be 20 times larger.A Google DeepMind research leader wrote in the official announcement that “Gemma 4 delivers an unprecedented level of intelligence-per-parameter.”<h2>Built for agentic AI systems</h2>One of the key goals of Gemma 4 is to support agentic AI systems, which can perform tasks autonomously rather than simply generating text responses.The models include several capabilities designed for AI agents, such as:Native function calling Structured JSON output Support for system commands Integration with external tools and APIsThese features allow developers to build AI systems that can analyse data, trigger workflows and interact with digital tools.<h2>Multimodal AI with support for 140 languages</h2>Gemma 4 is also designed as a multimodal AI system, meaning it can process different types of data.The models support:Text Images VideoAdditionally, the smaller E2B and E4B versions include native audio input for speech recognition.Google said the models support more than 140 languages, allowing developers to build AI applications for global users.Another notable upgrade is the 256,000-token context window for larger models. This enables developers to process long documents, full research papers or even entire code repositories in a single prompt.<h2>AI models designed for mobile and edge devices</h2>Unlike many large AI systems that require powerful servers, some Gemma 4 models are designed to run directly on devices. The E2B and E4B models are optimised for mobile phones, Internet-of-Things devices and edge computing systems.Google said these models can run offline on devices such as smartphones, Raspberry Pi computers and NVIDIA Jetson Orin Nano, offering near-zero latency.The company also collaborated with Qualcomm and MediaTek to improve mobile performance.<h2>A growing ecosystem around Gemma models</h2>Since the first Gemma models were introduced, adoption has grown rapidly. Google reports that Gemma models have been downloaded more than 400 million times, leading to the creation of over 100,000 community-built variants, often referred to as the “Gemmaverse”.Developers can access Gemma 4 through platforms such as:Hugging FaceKaggleOllamaGoogle AI Studio<h2>What this means for developers and the AI industry</h2>Gemma 4 reflects a broader shift in the AI industry toward open models that developers can customise and deploy across different environments.By offering multiple model sizes and device compatibility, Google DeepMind aims to support both large-scale enterprise systems and smaller applications running on personal devices. As AI development moves toward autonomous systems capable of executing tasks, models like Gemma 4 may become an important foundation for building next-generation AI agents.&nbsp;

Google DeepMind has launched Gemma 4, a new generation of open AI models designed for advanced reasoning and agentic AI workflows. Released under the Apache 2.0 open-source licence, the models aim to deliver higher intelligence with fewer parameters, improving efficiency. 

Google DeepMind has introduced Gemma 4, a new generation of open AI models designed for advanced reasoning and autonomous AI workflows. The release represents Google’s latest push in the open-model ecosystem, offering developers more powerful AI tools while allowing broad use under the Apache 2.0 open-source licence. According to Google DeepMind’s official blog, Gemma 4 is designed to deliver higher intelligence with fewer parameters, making it more efficient for real-world applications.

Google DeepMind has launched Gemma 4, a new generation of open AI models designed for advanced reasoning and agentic AI workflows. Released under the Apache 2.0 open-source licence, the models aim to deliver higher intelligence with fewer parameters, improving efficiency.&nbsp;

Google DeepMind launches Gemma 4: AI models for advanced reasoning and AI agents

Artificial intelligence agents are increasingly becoming part of workplace tools. Now, Google is experimenting with an internal AI system called Agent Smith, designed to automate tasks and allow employees to monitor work directly from their phones.&nbsp;The internal tool allows engineers and other employees to assign tasks that run automatically in the background. Workers can then check progress, review updates or send new instructions using their mobile devices. Reports indicate the system is already helping engineers complete routine tasks more quickly while reducing manual workload.&nbsp;<h2>What is Google’s ‘Agent Smith’?</h2>Agent Smith is an AI-powered system developed internally at Google. Unlike traditional coding assistants, the tool can plan and execute multiple steps of a workflow on its own. The AI system builds on Google’s existing agentic coding platform known as Antigravity. It can interact with various internal tools and services used inside the company.&nbsp;Employees can assign tasks to the system, which then works independently. The agent operates asynchronously, meaning it continues running even when the employee is offline or their laptop is closed.&nbsp;Workers can later check progress from their phones, review results or give further instructions.&nbsp;<h2>How the AI agent works inside Google</h2>According to people familiar with the system, Agent Smith can perform several functions that normally require manual work.&nbsp;These include:Automating parts of coding workflowsAccessing internal documents and toolsManaging multi-step tasks in the backgroundSending updates through internal chat platforms&nbsp;Because the system can access employee profiles and internal documentation, it can retrieve information that workers would otherwise need to search manually.&nbsp;Some engineers say the tool helps reduce time spent on repetitive tasks and allows them to focus on more complex work.&nbsp;<h2>Why Google is investing heavily in AI agents</h2>Agentic AI systems that perform actions autonomously is becoming a major focus across the technology industry.&nbsp;Google co-founder Sergey Brin reportedly told employees during a recent town hall meeting that AI agents will play an important role inside the company this year. At the same time, Google CEO Sundar Pichai has encouraged employees to adopt AI tools internally. According to reports, some teams have even been told that the use of AI may be considered in performance evaluations. The company sees AI automation as a way to improve productivity as competition among technology firms intensifies.&nbsp;<h2>Part of a broader industry trend</h2>Google is not the only company exploring autonomous AI agents. Several major technology firms are investing in systems that go beyond simple chatbots.&nbsp;Instead of just generating answers, agentic AI tools can:&nbsp;Run commandsanalyse documentscoordinate workflowsinteract with software systems&nbsp;This approach aims to turn AI into a digital worker that can handle routine tasks independently. For companies with large engineering teams, such tools could significantly change how software development and internal operations are managed.&nbsp;<h2>What the future of AI agents may look like</h2>Google has not officially confirmed detailed plans for Agent Smith. A company spokesperson told reporters that Google is constantly experimenting with new ways to build AI agents that solve real-world problems. If systems like Agent Smith continue to develop, they could change how employees interact with computers. Instead of working primarily from laptops, workers may increasingly assign tasks to AI systems and monitor progress remotely. The concept of “work from phone” may become more common as AI agents handle complex workflows in the background. For now, Agent Smith remains an internal experiment. However, it highlights how technology companies are exploring new ways to integrate AI into everyday work.

Google is testing an internal AI agent called Agent Smith, designed to automate workplace tasks. Built on Antigravity, the system can run workflows independently, access internal tools and documents, and allow employees to assign tasks and monitor progress from their phones. 

Artificial intelligence agents are increasingly becoming part of workplace tools. Now, Google is experimenting with an internal AI system called Agent Smith, designed to automate tasks and allow employees to monitor work directly from their phones. The internal tool allows engineers and other employees to assign tasks that run automatically in the background. Workers can then check progress, review updates or send new instructions using their mobile devices. Reports indicate the system is already helping engineers complete routine tasks more quickly while reducing manual workload. What is Google’s ‘Agent Smith’? Agent Smith is an AI-powered system developed internally at Google. Unlike traditional coding assistants, the tool can plan and execute multiple steps of a workflow on its own. The AI system builds on Google’s existing agentic coding platform known as Antigravity. It can interact with various internal tools and services used inside the company. Employees can assign tasks to the system, which then works independently. The agent operates asynchronously, meaning it continues running even when the employee is offline or their laptop is closed. Workers can later check progress from their phones, review results or give further instructions. How the AI agent works inside Google According to people familiar with the system, Agent Smith can perform several functions that normally require manual work. These include: Automating parts of coding workflows Accessing internal documents and tools Managing multi-step tasks in the background Sending updates through internal chat platforms Because the system can access employee profiles and internal documentation, it can retrieve information that workers would otherwise need to search manually. Some engineers say the tool helps reduce time spent on repetitive tasks and allows them to focus on more complex work. Why Google is investing heavily in AI agents Agentic AI systems that perform actions autonomously is becoming a major focus across the technology industry. Google co-founder Sergey Brin reportedly told employees during a recent town hall meeting that AI agents will play an important role inside the company this year. At the same time, Google CEO Sundar Pichai has encouraged employees to adopt AI tools internally. According to reports, some teams have even been told that the use of AI may be considered in performance evaluations. The company sees AI automation as a way to improve productivity as competition among technology firms intensifies. Part of a broader industry trend Google is not the only company exploring autonomous AI agents. Several major technology firms are investing in systems that go beyond simple chatbots. Instead of just generating answers, agentic AI tools can: Run commands analyse documents coordinate workflows interact with software systems This approach aims to turn AI into a digital worker that can handle routine tasks independently. For companies with large engineering teams, such tools could significantly change how software development and internal operations are managed. What the future of AI agents may look like Google has not officially confirmed detailed plans for Agent Smith. A company spokesperson told reporters that Google is constantly experimenting with new ways to build AI agents that solve real-world problems. If systems like Agent Smith continue to develop, they could change how employees interact with computers. Instead of working primarily from laptops, workers may increasingly assign tasks to AI systems and monitor progress remotely. The concept of “work from phone” may become more common as AI agents handle complex workflows in the background. For now, Agent Smith remains an internal experiment. However, it highlights how technology companies are exploring new ways to integrate AI into everyday work.

Google is testing an internal AI agent called Agent Smith, designed to automate workplace tasks. Built on Antigravity, the system can run workflows independently, access internal tools and documents, and allow employees to assign tasks and monitor progress from their phones.&nbsp;

Google’s ‘Agent Smith’: AI agent lets employees run tasks directly from their phones

Artificial intelligence is slowly entering sectors once considered purely human-driven. The latest example comes from Google DeepMind, which has demonstrated an AI system capable of planning and designing parts of city infrastructure.

<blockquote class="twitter-tweet" data-media-max-width="560">We used Gemini 3.1 Pro to build a realistic city planner app. 🏙️ Watch how the model tackles complex terrain, maps out infrastructure, and simulates traffic to generate a high-quality visualization. <a href="https://t.co/SKoVzwtBy8">pic.twitter.com/SKoVzwtBy8</a>&mdash; Google DeepMind (@GoogleDeepMind) <a href="https://twitter.com/GoogleDeepMind/status/2024570291767181557?ref_src=twsrc%5Etfw">February 19, 2026</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

The development has triggered debate online, with some questioning whether careers such as civil engineering or urban planning could face disruption in the future.However, experts say the reality is more complex.<h2>What did Google DeepMind announce?</h2>Google DeepMind shared a post on X, in which they demonstrated that Gemini 3.1 Pro can help design and optimise city infrastructure layouts.&nbsp;The system can simulate urban environments and suggest improvements in road networks, traffic systems and infrastructure planning. The focus is on efficiency, sustainability and better use of land and resources.DeepMind has been working on applying artificial intelligence to real-world challenges, including energy optimisation and traffic management. Earlier, DeepMind collaborated with Google to reduce data centre energy use by up to 40 per cent using AI-driven cooling systems.

Now, it appears urban planning is also becoming an area of experimentation.<h2>How does AI help in city planning?</h2>Urban planning involves analysing large datasets traffic flow, population density, land use, energy demand and environmental impact. AI systems are particularly strong at handling complex data patterns.Using machine learning models, AI can:Simulate different infrastructure layoutsPredict traffic congestionSuggest optimal road or building placementsAnalyse environmental impactSuch tools do not replace engineers. Instead, they provide simulations and data-backed suggestions to support decision-making.City infrastructure projects often cost billions of dollars and take years to complete. Even small efficiency improvements can reduce costs and improve sustainability.<h2>Does this make civil engineering or UPSC roles useless?</h2>The answer is no.AI systems can generate models and suggestions, but final decisions still require human expertise, regulatory understanding and local governance knowledge.Civil engineers deal with on-ground realities, safety standards, soil conditions, legal approvals and community concerns. Similarly, administrative officers under services like the UPSC manage policy execution, budgeting, public welfare and coordination between departments.AI tools may assist in planning, but they cannot replace field inspections, public consultations or political accountability. Experts believe that the nature of these jobs may evolve rather than disappear.<h2>AI in infrastructure: A growing trend</h2>Globally, governments and technology firms are investing in “smart city” projects. AI is being used for:Traffic signal optimisationWater distribution monitoringPredictive maintenance of bridges and roadsUrban heat mappingAccording to industry reports, the global smart city market is expected to grow steadily over the next decade as cities adopt digital infrastructure.Google DeepMind’s experiment fits into this broader shift, where AI becomes a tool for planning and analysis rather than direct replacement.<h2>What this means for students and professionals</h2>The rise of AI in urban planning suggests that future civil engineers and administrators may need new skills.Data literacy, AI understanding and technology integration are becoming important alongside traditional engineering knowledge.Instead of making degrees useless, AI may change what is expected from professionals.The key question is not whether AI will replace infrastructure experts, but how professionals will work alongside AI systems.

Google DeepMind has demonstrated Gemini 3.1 Pro, which can help simulate and optimise city infrastructure, including roads and traffic systems. While AI can analyse complex urban data and suggest improvements, it will support, not replace, civil engineers and administrators.

Artificial intelligence is slowly entering sectors once considered purely human-driven. The latest example comes from Google DeepMind, which has demonstrated an AI system capable of planning and designing parts of city infrastructure.

Google DeepMind has demonstrated Gemini 3.1 Pro, which can help simulate and optimise city infrastructure, including roads and traffic systems. While AI can analyse complex urban data and suggest improvements, it will support, not replace, civil engineers and administrators.

UPSC and Civil Engineering now useless? Google’s DeepMind now plans and designs city infrastructure

google-deepmind

Google Deepmind, News on Google Deepmind, stories on Google Deepmind, articles on Google Deepmind, Google Deepmind stories at Wion, Google Deepmind news, Google Deepmind latest news

google deepmind

Google DeepMind unveils AI that can understand sign language

Google's Gemini Robotics 2 gives humanoid robots a smarter AI brain

Google unveils AI smart glasses with Gemini, targets Meta’s dominance

Google accidentally reveals COSMO AI assistant: Here’s what it can do

Gemma 4 launch signals Google’s push into open, agent-ready AI systems

‘Agent Smith’: Google’s internal AI tool automates coding and workflows

Is civil engineering at risk? Google DeepMind unveils AI city planning tool