Google Upgrades Opal with AI-Powered Automated Workflows

Google has unveiled a significant update to Opal, its “vibe-coding” application, introducing new AI-driven agents designed to automate complex tasks. This enhancement allows users to build mini-apps capable of planning and executing workflows entirely through natural language prompts.

At the core of this functionality is the Gemini 3 Flash model. The new agent acts as an autonomous assistant that intelligently selects the appropriate tools for a given task. For example, it can interface with Google Sheets to maintain data persistence across different sessions, such as managing a dynamic shopping list for an e-commerce application. The system is designed to independently plan and initiate the subsequent steps required to complete a user’s request.

Image Credits:Google

The company emphasizes that these agents are built to be natively interactive. If an agent requires additional clarity to proceed, it will prompt the user for specific input or provide a selection of options to guide the workflow. This capability aims to lower the barrier to entry, allowing individuals without formal programming expertise to construct sophisticated applications.

The Evolution of Opal

Since its inception, Google has rapidly expanded the reach and functionality of Opal:

  • July 2025: Initial launch for users in the United States.
  • October 2025: International rollout to 15 countries, including Brazil, Japan, India, and South Korea.
  • November 2025: Global expansion reaching over 160 countries.
  • December 2025: Integration into the Gemini web app, introducing a visual editor for no-code development.

The move places Google in direct competition with a growing ecosystem of startups focused on natural language app development. Platforms such as Lovable and Replit have already established a presence in this space. Additionally, new entrants like Wabi—founded by the creator of Replika—along with Emergent and Rocket.new, are increasingly competing for users looking to build software without writing traditional code.

Deixe um comentário

O seu endereço de e-mail não será publicado. Campos obrigatórios são marcados com *