Google is expanding the capabilities of its “vibe-coding” app, Opal, with the introduction of a new agent-based workflow system. Announced on , the update allows users to create mini-applications capable of planning and executing tasks through simple text prompts, effectively automating complex processes without requiring traditional coding expertise.
Opal, initially launched for U.S. Users in , and subsequently rolled out to 15 additional countries including Canada, India, and Brazil by , aims to democratize app creation. The platform allows users to build and remix web applications. In , Google integrated Opal into the Gemini web app, providing a visual editor for app development, further lowering the barrier to entry.
The core of this new functionality lies in an agent powered by the Gemini 3 Flash model. This agent doesn’t simply execute commands. it autonomously determines the necessary steps to achieve a user-defined objective. It selects and utilizes appropriate tools – including Google Sheets for data persistence – to manage tasks and maintain context across multiple interactions. For example, the agent can build and maintain a shopping list within an e-commerce application, remembering items and preferences across sessions.
Crucially, Google emphasizes the interactive nature of these agents. If the agent requires further information to proceed, it will proactively ask the user for clarification or present a set of choices, ensuring a collaborative workflow. This contrasts with purely automated systems that might fail silently or produce unexpected results when encountering ambiguous instructions. This interactive element is intended to make the process more intuitive and accessible to users without a technical background.
The implications of this development extend beyond simple task automation. Google demonstrated the agent’s potential with an interior design application. Previously, such an app would have offered a static process: upload an image of a room and receive a redesigned version based on a specified style. With the new agent, the application becomes a dynamic collaboration. A user can upload a photo of an empty living room and describe a “mid-century modern” vision. The agent then generates an initial design concept, and the user can provide feedback on specific elements, iteratively refining the design in a conversation-like manner.
This shift from static workflows to agentic intelligence represents a significant evolution in Opal’s capabilities. Instead of manually configuring each step of a process, users can now define a high-level goal and let the agent handle the implementation details. This approach promises to unlock new levels of creativity and efficiency, allowing users to build more complex and personalized applications with less effort.
Google’s move into agentic workflows isn’t happening in a vacuum. A growing number of startups are exploring similar approaches to no-code and low-code application development. Companies like Lovable and Replit are gaining traction, and others – including Wabi (founded by the former creator of Replika), Emergent (backed by SoftBank and Lightspeed), and Rocket.new (backed by Accel) – are also vying for a share of this emerging market. The competition underscores the increasing demand for tools that empower individuals and businesses to build custom applications without relying on traditional software development expertise.
The introduction of the agent step in Opal marks a move towards more dynamic and interactive experiences. By automating the planning and execution of tasks, Google aims to make app creation more accessible and empower users to bring their ideas to life with greater ease. The success of this approach will likely depend on the agent’s ability to accurately interpret user intent, effectively leverage available tools, and seamlessly integrate feedback into the workflow.
