Google Opal Adds Agent Step and 3 Tools, Expanding AI App Automation

1 articles · Updated · KDnuggets · Sep 15

Google Labs has added an Agent option to Opal’s Generate step, letting the no-code app builder choose models and tools at runtime instead of following a fixed, prewired workflow.
The update ships with three new capabilities—Memory, Dynamic Routing via “@ Go to,” and Interactive Chat—so apps can retain user details, change paths mid-run, and pause to ask for missing information.
Opal’s model lineup has also broadened beyond Gemini, Imagen and Veo to include Gemini Flash, Gemini Pro, Nano Banana, Nano Banana Pro, AudioLM, Veo and Lyria 2, making agent-led orchestration more useful.
Google has also moved Opal onto its Google for Developers site from the Labs catalog, a sign of deeper commitment even as the product remains experimental and lacks export-to-production, published quotas and enterprise features.
The rollout follows Opal’s July 2025 launch and, according to the report, comes amid expansion to more than 160 countries within roughly seven months.