Closed-door exchange | The next stop for on-device intelligence: Personal AI × Physical AI

Building a great piece of Personal AI hardware means answering three questions: in what form does the product fit naturally into users' lives? How do you build a Personal System that keeps growing from multimodal information? And how do you create value in the target scenario that users can feel and will pay for?

OpenBMB, together with ecosystem partners including Shenzhen InnoX Academy, INNO100, Datawhale, APLUX and the Zhuiguang Community, is hosting a closed-door exchange on "The next stop for on-device intelligence: Personal AI × Physical AI".

We have invited Zeng Guoyang, CTO of ModelBest; Chen Kangda, co-founder of Guangfan Technology; Gao Xinlu, COO of Looki; and the founders of startups including ComindX, Fangcun Lingnuo and Mira Light to share product and technical practice around "how to build a great piece of Personal AI hardware".

Researchers of Personal AI Systems, on-device model and smart hardware developers, and founders of smart hardware products are all welcome to gather in Shenzhen on October 16 to explore the future of Personal AI hardware.

What we will talk about

Product form | How does it fit into daily life?

Earbuds, glasses, a pendant, a portable drive or a desktop terminal? Starting from usage scenarios, we will discuss the trade-offs different product forms make among sensing capability, interaction style, battery life and cost.

Personal system | How does it understand a person better the more it is used?

Which information is worth remembering, when to recall it, how are memories updated and deleted? We will discuss how the model layer, memory layer and application layer work together, and how to achieve efficient device-cloud collaboration.

Scenario value | What will users pay for?

Life logging, personal assistant, desktop companionship... How much value can each create for users? Drawing on real product feedback data, we will discuss users' willingness to pay and the business model in each target scenario.

Speakers

Zeng Guoyang

ModelBest | Co-founder and CTO

Tsinghua University Department of Computer Science and Technology; NOI gold medalist; 2026 China AI Person of the Year; led the development of the VoxCPM series of on-device speech models.

ModelBest was founded in 2022 and is an AI foundation model company. With the strategic core of driving the AGI terminal scenario revolution and the "density law" as its technical guide, it focuses on full-stack R&D, industrial application and ecosystem building for efficient large models, so that intelligence reaches every industry, every device and every user. ModelBest independently develops the MiniCPM series of on-device models, which have been downloaded more than 50 million times worldwide.

Chen Kangda

Beijing Guangfan Technology Co., Ltd. | Co-founder

Joint PhD student of Tsinghua University and UC Berkeley; co-founded Guangfan Technology in 2024, dedicated to building AI wearable hardware and an AI-native OS for the next generation of human-computer interaction.

Guangfan Technology is an emerging AI company founded in October 2024, dedicated to building AI wearable hardware and an AI-native OS for the next generation of human-computer interaction. Its core team comes from Xiaomi, ByteDance, Alibaba, MiniMax and other companies, with deep AI software and hardware capability. Within a year of founding it completed four funding rounds totaling nearly RMB 300 million, and in 2026 it beat Apple, OpenAI and other global giants to mass-produce and ship the world's first AI earbuds with visual perception, topping JD.com's AI earbuds bestseller list for consecutive periods.

Talk | The last mile of Personal AI, and how an AI OS empowers wearable hardware

Gao Xinlu

Looki | COO

Responsible for operations and commercialization. Long-term focus on how AI can fit into real-life scenarios in a more natural, low-intrusion product form and keep creating value for users.

Looki is a technology company focused on Personal Intelligence, committed to deeply integrating multimodal AI capability into the physical world and bringing human-centered AI into daily life.

Talk | How Personal AI fits into real life: Looki's exploration of Personal Intelligence

Liu Senhao

Wuwei Erzhi Technology | Founder and CEO

PhD from Renmin University of China. The company has received investment from leading venture firms and the industrial arm of a listed company.

Wuwei Erzhi Technology focuses on Memory and on-device self-evolution. It has developed an on-device user-level self-evolving intelligent terminal based on graph memory, addressing the difficulty AI has in continuously adapting to individual change, and coordinates complex cloud capabilities so the system evolves with the user and delivers services that understand each person better and fit their needs. The core team comes mainly from Tsinghua University, Renmin University of China, the National University of Singapore and other top universities at home and abroad and from leading technology companies. The company has completed a seed round from leading institutions and industry investors.

Talk | "Wuwei er zhi": is non-action the highest state of intelligence?

Luo Jingmin

LYNOOK | Founder and CEO, Shenzhen Fangcun Lingnuo Technology Co., Ltd.

An early member of Tencent AI Lab and a serial entrepreneur with startup experience in AI, games and consumer hardware products. He is dedicated to exploring entertainment-oriented forms of Personal AI.

LYNOOK is an AI Agent life-simulation company serving the global market, exploring the next product form of Personal AI and Physical AI. LYNOOK wants Personal Agents to have a continuously running "simulated life", so that AI cares not only about work but also experiences life alongside people and creates new game and social experiences. LYNOOK is building a desktop physical device where agents live — a physical "Stanford town".

Talk | LYNOOK: from Work Agent to Life Agent

Wang Jianle

Mira Light | Founder and CEO

Rick, a serial entrepreneur and interaction designer, University of Wisconsin-Madison F25, first place nationally in FBLA international marketing, winner of a Google go-global hackathon. He has won multiple international graphic and interaction design awards.

Mira Light is a desk lamp-shaped desktop companion robot. It can perceive proactively and act autonomously, staying by your side at just the right distance to interact and gather context. It has a distinct personality and accumulates long-term memory built together with you, offering care, interaction and atmosphere at just the right moment through body posture, light and voice. Some care needs no words — we want to make technology full of love again.

Talk | Mira Light: seen, heard, lit up

Guangfan Technology

Looki

Mira Light

Wuwei Erzhi Technology

LYNOOK

Agenda

Who should attend

AI hardware founders: smart hardware founders or teams whose products are already in mass production or shipping, or who plan to enter crowdfunding or mass production in the next six months

On-device model and hardware developers: makers with experience in on-device model deployment, chip adaptation or model optimization, or with smart hardware development experience

Personal AI researchers: technical talent focused on personalized modeling, memory mechanisms and multimodal interaction, and interested in where Personal AI goes next

Join the event

This closed-door session uses a registration-and-review process. Please fill in the questionnaire below; once approved, staff will send an attendance notice to the phone number or WeChat ID you provide.

📝 Scan the code to fill in the registration questionnaire

Hosts and partners

Hosts

Ecosystem partners