Doubao Phone 2 Targets Daily Tasks over Raw Specs

ZTE and ByteDance have launched the nubia NaviX Ultra, officially positioned as the first mass-market AI agent phone. Unlike its predecessor, which was a limited-run engineering sample, this device is a standard consumer product available in four colors and starting storage configurations that meet mainstream expectations. The launch signals a pivot from proving technological feasibility to demonstrating commercial viability in a crowded flagship market.

At a starting price of 5,999 yuan, or 5,499 yuan with national subsidies, the phone competes directly with other high-end devices. Pre-order reservations on major Chinese platforms exceeded 370,000 units before the official sale began. This level of interest suggests that early adopters are eager to move beyond chatbot interactions and toward an interface that executes multi-step tasks autonomously, such as booking travel or summarizing work communications.

Hardware Serves The AI Core

The specifications are competitive but not revolutionary, featuring a Snapdragon 8 Elite processor and a large 7,100mAh battery. The hardware exists primarily to support the phone’s defining feature: a dedicated physical AI button on the side frame. This button does not open a specific app but activates a system-level agent. This design choice prioritizes immediate access to voice and task execution over manual app navigation, creating a distinct user experience compared to traditional smartphones.

A notable inclusion is the use of domestically produced high-speed memory chips, marking the first time this technology has appeared in a mass-produced device. While the camera system includes a 200-megapixel main sensor, the marketing focus remains on the phone’s ability to process and act on user commands. The hardware is the stage, but the software agent is the performer.

From Screen Reading To Protocol

The previous generation relied on simulating user actions by reading the screen, a method that often triggered security blocks from app developers. This caused significant reliability issues for early users. To address this, the new model introduces the Screen Automated Execution Protocol, or SAEP. This framework establishes standardized rules for how the AI agent interacts with applications, aiming to reduce friction and prevent the malfunctions that plagued the prototype.

The shift from a hacky workaround to a structured protocol is the critical difference between a demo and a product. By defining clear boundaries for automation, the developers hope to ensure that tasks like booking hotels or comparing prices execute smoothly. This move represents a maturation of the technology, moving from a novel trick to a functional utility that can handle complex, multi-step instructions without breaking application security.

Market Reality Checks The Hype

According to reporting by GN technics/mobile (en-US), the market is now evaluating whether this new category can sustain itself. The technology is no longer the primary barrier; the challenge is user adoption. Consumers must decide if the convenience of an agent that handles tasks autonomously outweighs the cost and the learning curve of a new interaction model.

The trade-off is clear: users gain time and automation but cede some control to an AI that must be trusted with sensitive data and financial transactions. If the reliability of the new protocol holds up, the Doubao Phone could define a new standard. If not, it risks becoming another expensive gadget that fails to deliver on its promise of seamless, agent-driven productivity.