As AI continues to become the central focus of the tech industry, companies like OpenAI and nubia are racing to release AI-powered smartphones. However, it seems that Chinese startup StepFun may have beaten them to the punch. Enter StepX Neo, a device that may have just claimed the title of world’s first mass-market agentic smartphone.
Before we dive into the nitty-gritty details, it’s important to clarify what sets such a product apart from existing phones. After all, modern handsets already come with AI features. That said, these tools are embedded in apps. So if a user wants to complete a complex task with multiple different components, they will need to open these apps one at a time. Conversely, agentic smartphones leverage AI agents to handle multi-step, end-to-end tasks with just a single prompt.
Or at least, that’s the theory anyway. None of these devices have landed in a user’s hands just yet, though StepFun is looking to change that. Founded in 2023 by former Microsoft employees, the company developed its own AI agent, called Amoo. This agent is integrated into the StepX Neo’s operating system and works fully offline for all core tasks, requiring no constant network connectivity.
Much like many other consumer-facing AI models, Amoo relies on a chatbot interface for user interaction. It’s designed to take a single natural language request from users and perform tasks across multiple apps and tools, delivering full, finished results. Some examples of its capabilities include translating live conversations, phone calls, messages, and text captured through the camera across 32 languages and regional dialects. Furthermore, it can access local transit schedules, find nearby events, and access saved trip itineraries.
Other than that, the StepX Neo can book flights, automatically fill customs forms, identify visa requirements for destinations, and send reminders for check-in times. The company also claims that the agent can remember the user’s preferences over time.
Powering these capabilities is StepFun’s custom operating system, Step AOS. Rather than adding AI on top of Android as an app, the software uses rebuilt pieces of Android, Linux, and RTOS. Created with what the company calls an atomic capability engine, the OS splits the phone’s functions into communication, apps, files, and system tools using the Model Context Protocol (MCP) standard. This enables the agent to combine the different functions to complete tasks.
Meanwhile, its Natural User Interface (NUI) accepts voice and visual inputs while learning the user’s habits. Additionally, three different systems support the agent in terms of memory, decision-making, and security. All of this is powered by Step Edge, a proprietary foundation model designed for smartphones.
Finally, StepFun revealed that the StepX Neo will offer integrations with platforms like Ctrip, Alipay, Didi, Meituan, WPS, and CapCut. That said, the company did not offer any information on the device’s specifications, though it seems that it features two cameras and a secondary display on the rear. Similarly, details on its availability and pricing remain uncertain.
(Source: Wallstreetcn via GizmoChina)
