Google adds automated workflow agents to Opal app

Google announced on Tuesday that it is adding a new way to create automated workflows to Opal, its “vibe-coding” app—a term used to describe the process of building applications using natural language. A new agent introduced within Opal allows users to create mini-apps that can plan and execute tasks based on text prompts.

The feature is powered by the Gemini 3 Flash model and automatically chooses tools to execute tasks. The new agent plans and creates the next steps for tasks on its own. Google claims that with this addition, users without technical knowledge could build complex workflows within their apps.

The update follows a rapid expansion for Opal. Google first introduced the tool for U.S. users in July 2025, allowing anyone to create mini web apps or remix existing apps. In October 2025, the company rolled out Opal to users in 15 more countries, including Canada, India, Japan, South Korea, Vietnam, Indonesia, Brazil, and Singapore. A month later, in November 2025, Google made the tool available in over 160 countries. In December, Google added the tool to the Gemini web app, allowing users to create custom apps through a visual editor without writing any code.

Beyond Google, many other startups are also building tools that let users build apps via natural language prompts. Lovable and Replit are among the more popular ones, but other startups like Wabi, Emergent, and Rocket.new are also gaining prominence.

Why it matters

By enabling non-technical users to build complex, task-executing mini-apps, Google is lowering the barrier to entry for software creation and directly competing with a growing ecosystem of AI-native development startups.