Google Labs has introduced a new agent step in Opal that transforms static workflows into dynamic, interactive experiences. This update enables Opal to move beyond predefined sequences by allowing agents to understand user objectives, autonomously select the best tools and models, and adapt in real time to achieve desired outcomes. With the new agent step, users no longer need to manually choose specific models or define rigid paths. Instead, the agent analyzes the goal and determines the optimal sequence—triggering tools like Web Search for research or Veo for video generation—automating complex tasks with minimal setup. For example, creating a storybook in Opal used to require predefining page counts and user questions. Now, a Visual Storyteller Opal can use the agent to dynamically suggest plot points and story developments, turning the process into a collaborative, evolving narrative shaped by real-time creative decisions. In interior design, the agent-powered Room Styler Opal becomes a true design partner. After uploading a photo and describing a mid-century modern vision, the agent generates an initial concept with period-accurate decor and color schemes. If the result isn’t quite right, users can give feedback on specific elements. The agent learns from this interaction, refines its understanding, and even researches niche sub-styles to deliver a personalized design that feels unique, not generic. This shift is made possible by several new capabilities. Memory allows Opals to retain user preferences—like style choices or brand identity—across sessions. In a Video Hooks Brainstormer Opal, for instance, the agent remembers a user’s brand voice and preferences, enabling instant, tailored content ideas without repetition. Dynamic routing lets users define multiple workflow paths based on custom logic. In an Executive Briefing Opal, the agent automatically adjusts the briefing based on whether the meeting is with a new or existing client—pulling in web data for new clients or referencing internal notes for ongoing ones. Interactive chat gives agents the ability to ask follow-up questions or present options when information is missing. In the Room Styler Opal, if a user hasn’t provided enough detail, the agent can prompt for clarification or show examples to guide the process. Google Labs designed Opal to balance automation with user control. New users benefit from a seamless experience where the agent self-corrects and optimizes. Power users still have access to traditional fixed steps for precise prototyping or deterministic workflows. By combining intelligent agents with customizable workflows, Opal now supports a broader range of creative and technical applications. The platform remains simple to use while unlocking deeper capabilities, empowering users to build smarter, more responsive, and truly personalized AI experiences.