Wang Mengqiu on AI Applications, the Funding Winter, and Choices in Life

Zhang Xiaojun invites Wang Mengqiu, a founding partner of ClearVue Partners and former Baidu technology vice president, to discuss the investment market and AI applications in spring 2025. Their conversation moves from work to travel, books, film, feminism, and personal life.

Asked whether the funding winter has passed, Mengqiu says she sees no substantial recovery in either primary-market capital or investable projects. Some investors previously on the sidelines are willing to act, but she considers their capital limited relative to state-backed funding.

She observes continued concentration on hard technology, AI, and embodied intelligence, while consumer investors favor companies with established scale and profits. That is her view of the market she encountered, not a description of every investor.

Zhang asks whether DeepSeek significantly boosted Chinese AI application startups. Mengqiu thinks open models lowered some barriers, but compute still costs money, and the central challenge is finding genuinely compelling use cases.

She argues that technology and user needs must advance together; waiting for stronger models will not automatically create use cases. Founders need to identify unmet needs and possible leaps in experience over the following six to twelve months.

Travel planning is her example: people revise budgets, locations, and even destinations while comparing hotels and flights. An agent taking on this work cannot merely execute a plan fixed at the outset.

Mengqiu says small improvements may not support an independent business; a valuable product needs to save substantial time or surprise users with its results. Understanding demand matters as much as understanding model capabilities.

On her own use, she says DeepSeek’s answers are not always accurate, but she finds its Chinese conversational style approachable. She likens it to a source of emotional reassurance while recognizing that its answers may not be reliable.

They discuss chatbots that flatter users too readily. Mengqiu speculates that feedback and post-training might contribute, while acknowledging she does not know how the products are implemented; Zhang says constant praise can feel unnatural.

Mengqiu compares the standalone Yuanbao app with its WeChat entry point. The same assistant can create different expectations in different settings: a tool should provide answers, while a contact-like interface also shapes expectations for interaction.

She hoped an assistant could collect background information about a person, but found that forums may restrict access and information still needs verification. Zhang warns that online rumors can be unreliable; the example prompts Mengqiu to reconsider the gap between user expectations and actual access.

She dislikes a WeChat assistant first producing an overly enthusiastic short message and requiring a click for the answer. She wants direct, honest responses and emotional support when needed, rather than automatic praise every time.

Personalization might improve interaction, but Mengqiu’s scenario depends on user permission. Even without personalization, she sees substantial product work in answer quality and the design of a friend-like interaction.

She was surprised by Tencent’s quick adoption of DeepSeek but welcomes its willingness to use an outside model. A platform, she argues, need not insist on using only its own technology.

Zhang raises the issue of information silos: platforms have their own content, and assistants may not be able to retrieve it across boundaries. Mengqiu thinks different platforms may develop assistants for different user expectations.

For example, Xiaohongshu might embed travel planning in lifestyle use cases, while WeChat is a frequent communication entry point. Mengqiu thinks users’ choice depends on where a task is easiest to complete.

Asked how an agent in WeChat should work, Mengqiu admits she does not yet have an answer. She questions whether the distinction between general and vertical agents will persist and notes that some teams are building frameworks for other applications.

In her definition, an important distinction between an agent and an information-research tool is the ability to take action. Even clicking web links was unreliable in some systems she tried, leaving a gap between demonstrations and dependable work.

She recalls an early trial of Manus and says both the examples and her own tasks produced rough results that did not exceed her expectations. This reflects her limited experience at the time, not every version or task.

Mengqiu sees many agent startups targeting businesses or professionals by starting with existing workflows. She is particularly interested in video editing because ordinary creators also need it, although different kinds of video require different product design.

She observes many young application founders but rejects choosing entrepreneurs by age alone. Individual ability, product judgment, and industry timing matter more to her than birth decade.

At the same time, she thinks small teams may be able to build larger businesses, potentially reducing the organizational burden on young founders. Small teams do not necessarily need less funding: scarce talent and compute are expensive.

Embodied intelligence is another area she follows. Mengqiu says limited real-world data makes simulation important infrastructure, but transferring from simulation to reality remains difficult, as does building a good simulator.

She distinguishes software agents from robots. The former build on models and software frameworks and need essential workflows; the latter must also handle physical environments and data collection. Both involve action but require different capabilities.

Could a central planner eventually call multiple specialized agents? Mengqiu thinks so, but says the provider would depend on how people actually use their devices; it would not necessarily be WeChat.

If people continue to open separate apps, assistants may live inside each frequent app. If the phone’s entry point changes fundamentally, handset makers could gain an advantage. She cautions against assuming that asking a voice assistant to open an app is always easier.

On wearables, Mengqiu begins by asking what problem the product solves. She has used Meta smart glasses and likes their appearance, camera, and audio features, but does not conclude that she needs to wear them all day.

She says people may not want another device on their face merely to receive information everywhere. Zhang adds that notifications from multiple wearables can be irritating. Both emphasize comfort and concrete uses over a growing feature list.

She thinks smart rings might feel more natural to people who already wear rings and could monitor health metrics, though usefulness remains decisive. This is her preference, not a universal conclusion.

On consumer investing, Mengqiu says some existing companies are doing well, but she sees few new projects. Investors favor profitable, more mature businesses, while international expansion faces an external environment she finds hard to predict.

They review Chinese model companies and companion products. Based on what she saw at the time, Mengqiu discusses acquisition, retention, and expansion beyond an initial audience, while acknowledging that she lacks detailed data on every company.

Outside work, Mengqiu says she traveled often in the previous year, visiting Saudi Arabia, Germany, and Iceland. She sees exposure to different cultures as valuable and believes experiences that seem useless now may matter later.

She does not think investors must stay constantly busy to make good decisions. She says she made few investments the previous year without feeling she had obviously missed a major opportunity; in a changing industry, she sees no need to rush into every wave.

On how her travel has changed, she says that when younger she wanted to visit many places in a day; now she prefers settling somewhere without rushing. A different environment also helps her notice her own reactions.

Her reading includes life’s evolution, mental health, history, neuroscience, and biographies of entrepreneurs. She says friends have experienced anxiety or depression, prompting her to learn more, while acknowledging she cannot establish the causes.

She describes adopting a calmer approach to life and investing after leaving Baidu. She reflects on dealing with things she cannot obtain and admits that she still feels anger when confronted with things she cannot understand.

She compares biographies of Jensen Huang, Sam Altman, and Elon Musk, stressing that she does not know them personally and can only analyze their portrayals in the books. Her theme is that greatness is hard to plan in advance but can be studied.

On online fiction, Mengqiu finds the plot construction of a popular novel clever but stopped reading because of its length and constant pursuit of intense twists. Zhang asks whether AI could write something similar; Mengqiu thinks the plotting is not simple.

Turning to film, Mengqiu says cinemas must answer why people would leave home to watch a movie when short video, short dramas, social activities, and streaming compete for limited time. She distinguishes films themselves from using theaters for concert screenings and other events.

She thinks producers should consider their intended audience and the theatrical experience more carefully. She offers her own explanation for the success of Ne Zha: familiar characters lower the barrier to understanding, alongside market and audience factors.

Her response to Her Story is different: it recalls the lives of women she knows. She likes its story and visual language and believes its feminist narrative resonated, while expressing reservations about some lines and portrayals of men.

Discussing feminism, Mengqiu says she is relying mainly on personal experience and has not studied the topic deeply. She prefers evaluating people as individuals by their abilities and actions; her position does not represent every woman’s experience.

She explains why she decided early not to have children: caring for a pet brought intense worry, and she feared difficulty balancing respect for a child’s independence with her own expectations. She presents this as a choice suited to her.

She recalls marrying her husband when they were young; later, both felt that a life without children gave them more freedom over their time. She also describes conversations with friends about raising children.

At the end, Mengqiu reflects on her own and her friends’ life choices, and Zhang closes a conversation that ranged from investing to personal life.