Google is expanding the capabilities of Gemini 3.5 Flash by adding built in computer use tools that allow developers to create AI agents capable of taking actions across browsers, mobile devices, and desktop applications.
Quick Summary – TLDR:
- Google has added computer use as a built in capability in Gemini 3.5 Flash.
- Developers can now build AI agents that can see, reason, and take actions across web, mobile, and desktop environments.
- The feature is available through the Gemini API and Gemini Enterprise Agent Platform.
- Google has also introduced new security safeguards to reduce prompt injection risks and improve enterprise safety.
What Happened?
Google has announced a significant upgrade to Gemini 3.5 Flash, giving the model native computer use capabilities. The update allows developers and businesses to build custom AI agents that can interact with software and digital environments across multiple platforms.
The company says the enhancement makes Gemini 3.5 Flash better suited for enterprise automation, software testing, and professional workflows that require agents to perform tasks over extended periods.
— Google AI Studio (@GoogleAIStudio) June 24, 2026
Computer Use Is Now Built Into Gemini 3.5 Flash
Until now, computer use capabilities were only available through a separate Gemini 2.5 computer use model. Google has now integrated the feature directly into the main Gemini 3.5 Flash model, simplifying development and making advanced automation tools more accessible.
According to Google DeepMind product manager Mateo Quiros:
Google says Gemini already performs strongly when using built in tools such as Search and Maps grounding. By adding computer use directly into the model, developers can create agents that are capable of understanding what is happening on screen, making decisions, and carrying out actions across different operating environments.
New Opportunities for AI Agents
The addition of computer use significantly expands what developers can build with Gemini 3.5 Flash.
Google says developers can now create custom agents that can:
- Navigate websites and browser based applications.
- Interact with mobile apps.
- Work inside desktop software.
- Complete multi step workflows.
- Assist with enterprise productivity tasks.
- Support continuous software testing processes.
The company believes these capabilities will improve performance in long running automation scenarios where AI systems must maintain context and complete tasks across multiple applications.
Google described the update as a way to unlock stronger performance for enterprise automation, knowledge work, and other professional use cases that require agents to interact with digital tools in real time.
Available Through Gemini API and Enterprise Platform
Developers and organizations can start using the new functionality immediately through the Gemini API and the Gemini Enterprise Agent Platform.
Google also showcased how computer use can analyze content inside applications and generate organized outputs. In one example, Gemini used computer interaction capabilities to examine the Gemini app and return a categorized list of features.
The update reflects Google’s broader push to make AI agents more practical for real world business environments, where systems need to move beyond answering questions and actively perform tasks.
Google Adds New Enterprise Safeguards
As AI agents gain more autonomy, security becomes increasingly important. To address potential risks, Google said it has implemented targeted adversarial training designed specifically for computer use tasks.
The company is also introducing two optional enterprise safeguards.
These include:
- Requiring explicit user confirmation before sensitive or irreversible actions are performed.
- Automatically stopping tasks when an indirect prompt injection attempt is detected.
Google said enterprises should take a defense in depth approach by combining these safeguards with secure sandboxing, human oversight, and strict access controls.
The goal is to ensure that AI agents can operate effectively while reducing the risk of malicious instructions or unintended actions in live environments.
Why This Matters?
The latest update moves Gemini 3.5 Flash closer to becoming a complete platform for building action oriented AI agents. Instead of simply generating responses, agents powered by Gemini can now interact with software, navigate interfaces, and complete tasks across different devices.
As businesses increasingly look for ways to automate repetitive work and improve productivity, these capabilities could make Gemini a stronger competitor in the growing AI agent market.
SQ Magazine Takeaway
I think this is one of Google’s most important Gemini updates since the launch of the 2.5 series. The AI industry is rapidly moving from chatbots to agents that can actually perform work, and computer use is a critical piece of that transition.
By bringing these capabilities directly into Gemini 3.5 Flash and adding enterprise focused security controls, Google is making it easier for businesses to experiment with real world AI automation. If AI agents become mainstream in the workplace, updates like this will likely be remembered as key milestones in that shift.
