Google just turned your phone into an AI coworker

What happened

Google launched Gemini 3.5 Pro, described as its newest and most powerful AI model to date, at the World Artificial Intelligence Conference in Shanghai. This announcement was not presented as another improved chatbot but as an "agent" designed to act as a permanent AI coworker. Unlike previous AI models that operated like a vending machine, requiring users to input a question for a one-off answer, Gemini 3.5 Pro is designed to "do things" beyond answering.

The company demonstrated Gemini 3.5 Pro running as a side panel within Android and Chrome. In this configuration, the AI observes the user's current activity, with permission, and interacts with the content on the screen. Examples included summarizing a long email thread and drafting a response, filling out a form using information from email and calendar, and extracting action items from a meeting transcript into a Google Doc. This positions Gemini 3.5 Pro as an AI layer that lives on top of a user's digital life to handle repetitive tasks.

Why it matters

The introduction of Gemini 3.5 Pro as an embedded agent marks a strategic pivot from reactive chatbots to proactive AI systems that take action. By integrating AI directly into operating systems and browsers, Google is aiming to make AI an always-on, contextual layer rather than a separate application users must actively visit. This move suggests a future where AI handles routine digital tasks, freeing up human operators for more complex work.

Google's choice to launch this at the World Artificial Intelligence Conference in Shanghai, a global event, signals a competitive stance in the international AI race. With Chinese leadership declaring AI a national priority and other major AI labs like OpenAI, Anthropic, and XAI releasing new models, Google's announcement underscores a broader industry shift towards AI that not only understands but also executes. This competitive environment accelerates the development and deployment of agentic AI.

This development positions AI as a core component of daily digital workflows, moving beyond mere information retrieval. The ability of Gemini 3.5 Pro to observe and interact with user screens, process information across applications, and perform multi-step tasks in plain English fundamentally changes the nature of human-computer interaction. It transforms AI from a conversational tool into a functional assistant capable of autonomous operation within defined parameters.

What to watch next

• Will other major tech companies like Apple or Microsoft launch similar embedded AI agents that operate across their operating systems and browsers?
• How will user adoption rates for persistent, screen-watching AI agents evolve, given the privacy implications discussed?
• What new industry standards or regulations will emerge to address data privacy and security concerns for AI agents with broad system access?
• Will the demonstrated multi-step task completion capabilities of Gemini 3.5 Pro prove robust and reliable in real-world, varied user scenarios?
• How quickly will businesses and individual users integrate these agentic AIs into their core workflows and daily routines?

What this means for you

Business leaders and operators should recognize that AI is moving beyond simple conversational interfaces to become an embedded operational layer. This shift means that tasks involving digital drafting, organization, scheduling, and data management are increasingly automatable through plain English instructions. Evaluate current workflows for repetitive, time-consuming digital tasks that could be offloaded to an AI agent, allowing human resources to focus on higher-value activities.

The privacy implications of persistent AI agents require careful consideration. Before adopting these tools at scale, understand the data access permissions being granted, review platform privacy settings, and establish clear internal usage policies. For employees, clarify company guidelines on AI agent use, especially concerning sensitive business information. Strategic implementation of these tools, combined with a robust understanding of their data handling practices, will be crucial for leveraging their benefits responsibly.

Key takeaways

• Google's Gemini 3.5 Pro is an AI agent, not just a chatbot, embedded in Android and Chrome.
• The agent observes user activity and performs multi-step tasks from plain English instructions.
• This represents a shift from AI answering questions to AI planning and taking action.
• The launch in Shanghai signals global competition in AI agent development.
• Privacy concerns arise from AI agents seeing screens and accessing data, requiring careful management.

FAQ

What is Google's Gemini 3.5 Pro?

Google's Gemini 3.5 Pro is their newest and most powerful AI model, launched at the World Artificial Intelligence Conference in Shanghai. It is designed to function as an AI agent rather than a typical chatbot. Gemini 3.5 Pro lives on devices like phones and laptops, within Android and Chrome, to observe user activity and perform multi-step tasks, such as summarizing emails or filling forms, based on plain English commands.

How does Gemini 3.5 Pro differ from a regular chatbot?

Gemini 3.5 Pro differs from a regular chatbot because it is an AI agent that "does things" beyond just answering questions. While chatbots typically require new input for each interaction and don't retain context, Gemini 3.5 Pro is designed to persist, watch what a user is doing on screen, and autonomously complete tasks. It can plan, take steps, use tools, open apps, and interact with digital content on behalf of the user.

Where does Gemini 3.5 Pro operate?

Gemini 3.5 Pro operates as a side panel within Android and Chrome. This allows it to be present on the edge of the user's screen, observing the current activity, with permission. This integration enables it to interact with ongoing tasks, such as email threads, forms, or meeting transcripts, directly within the applications the user is already employing.

What kinds of tasks can Gemini 3.5 Pro handle?

Gemini 3.5 Pro is designed to handle various digital tasks by taking plain English instructions. Examples provided include summarizing long email threads and drafting responses, filling out forms using information from a user's email and calendar, and extracting action items from meeting transcripts to place them in a Google Doc. It targets repetitive digital tasks like drafting, organizing, and managing documents and schedules.

What are the privacy implications of using an AI like Gemini 3.5 Pro?

Using an AI like Gemini 3.5 Pro, which can see a user's screen, read emails, and interact with apps, raises legitimate privacy questions. The AI will know a lot about the user. It is advised to understand what permissions are being granted when enabling these features, check privacy settings, and understand how data is stored. Users should vet tools for sensitive information and be aware of company AI usage policies.