Hey Tuya

Hey Tuya is an AI life assistant developed by Tuya Smart, a global AI cloud platform service provider, that integrates multiple AI agents with physical devices in smart environments. It operates as an orchestration system that connects conversational AI, vision processing, and IoT intelligence through a unified Physical AI Engine (PAE), enabling real-time interaction between users, cloud intelligence, and hardware devices within Tuya’s ecosystem. Hey Tuya supports multimodal inputs such as voice, vision, and sensor data, and coordinates multiple AI agents to execute context-aware, multi-step tasks across home, energy, security, health, and productivity scenarios.

Hey Tuya uses a Tuya Real-Time Communication (T-RTC) and AI-Device Real-Time Network (AD-RTN) to reduce latency and support reliable device-to-cloud interactions. It includes long-term and short-term memory mechanisms to model user preferences, device behavior, and environmental context over time. 

Hey Tuya Features  

Hey Tuya is currently available through a controlled beta program within the Tuya App. 

App-Level Deployment and Entry Points

Hey Tuya is delivered as an AI assistant feature inside the Tuya App, with a dedicated entrance that opens a conversation page and voice interaction flow. It will be positioned to be integrable not only in the first-party app but also within OEM apps built on the same ecosystem, with support for customizing the assistant naming at the brand level.

Multi-Agent Orchestration Across Life Scenarios

The system coordinates multiple AI agents that map to functional domains (e.g., home, security, energy, lighting, notes, memo, fitness, health, petcare, podcast). These agents are intended to operate across “all life scenarios” and can issue commands to both software services and connected hardware.

Physical-World Integration Through Device Ecosystems

Hey Tuya is designed to extend beyond a cloud-only assistant by coupling AI behaviors to a large device ecosystem spanning multiple categories and product series. This integration is described as enabling heterogeneous sensing, cross-room context awareness, and coordinated multi-device actions.

Real-Time Infrastructure and Physical AI Engine

The core runtime is the Physical AI Engine (PAE), which uses a global AI-Device Real-Time Network (AD-RTN) for edge acceleration and a real-time communication framework built on T-RTC (Tuya Real-Time Communication). The goal is low-latency, reliable real-time connectivity between AI services and device endpoints.

Multimodal Engines and Long-Term Memory

PAE is layered with a Conversational AI Engine (natural-language + multimodal interaction), a Vision AI Engine (visual understanding), and an IoT Intelligence Engine (device communication and control). Long-term personalization is supported via OmniMem memory technology, described as learning across multimodal interactions, device behavior patterns, and scene intent to model user preferences and spatial habits over time.

Edge Processing and Data Locality Controls

Beyond cloud orchestration, Hey Tuya incorporates edge-computing mechanisms that allow selected data processing to remain local to devices. This reduces unnecessary cloud transmission, lowers latency, and supports privacy-preserving AI execution for sensitive workloads. 

Domain-Specific AI Modules

Dedicated AI modules are defined for lighting, notes, calorie tracking, memo management, fitness, pet care, parenting, podcasts, and security. Each module is implemented as an independent agent with domain-specific workflows, data models, and device integrations.