88Z AI, 17 November — Google has open-sourced FashionMind, a multimodal real-time AI voice try-on assistant built on ADK and Gemini Live 2.5 Flash, supporting full-duplex voice conversations, real-time outfit feedback via camera, and three-layer cross-session memory.
At a glance
- Built on ADK and Gemini Live 2.5 Flash, supporting full-duplex voice, real-time visual feedback, and cross-session memory
- Eliminates latency by injecting product catalogs into context and using signal tools, enabling immediate highlighting with zero extra turns

The update
Google has officially open-sourced FashionMind, a multimodal real-time AI voice try-on assistant. Built on the ADK framework and the Gemini Live 2.5 Flash model, it features full-duplex voice conversation capabilities, provides real-time outfit feedback via camera, and supports three-layer memory across sessions. The core innovation of FashionMind lies in directly injecting the product catalog into the context and converting the recommendation mechanism into a signal tool. This design eliminates tool invocation latency, achieving an interaction experience with zero extra turns. When users mention specific items, the system highlights them immediately, ensuring a smooth and natural communication flow.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- FashionMindSource published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.