Qi Junyuan, former head of Doubao's PC division, founded Today AI, which has raised over $200 million in cumulative funding

Another post-90s genius has emerged.

Yesterday, according to 5Y Capital's disclosure, the Shanghai personal AI assistant company Today AI has cumulatively completed more than $200 million in financing, equivalent to approximately 1.336 billion yuan. Investors include IDG Capital, Tencent, 5Y Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Venture Capital, and Shanghai Jiao Tong University Hanyuan Assets, among others.

In this era of "the AI stage is big, come if you have a dream," 1.3 billion in financing is not remarkable in itself. But this company was only registered and established in March this year, and the domestic version of its core product Today only launched on September 25—less than a month before this financing was made public. Not to mention that dollar funds, internet giants, phone manufacturers, and university-backed capital are all at this table.

The founder is Qi Junyuan, a post-90s entrepreneur, former vice president of product at Feishu and also the person in charge of Doubao's PC client.

If you ask what the best AI to use right now is, the answer may vary from person to person—Claude, GPT, DeepSeek, Kimi... But if you ask what is the most commonly used in daily life, I think it must be Doubao.

No matter what tricky question you have, asking Doubao first is always a safe bet—this is probably the true portrayal of most people. That is also what has made Doubao the most frequently used AI. And Qi Junyuan's identity as the person in charge of Doubao's PC client may also have become the point that most attracts investors' attention.

In his junior year, he wrote 220,000 lines of code

Many people know Qi Junyuan through Teambition. But if you tell his story, you have to start from the laboratory.

Qi Junyuan's father is a university professor doing scientific research, and he had been soaking in laboratories since middle school. In high school, he participated in a project at the Institute of Cells of the Chinese Academy of Sciences to identify the sex of parrots, and he succeeded.

By conventional logic, this should have been the starting point of a young scientist. But later he found that a single device abroad could solve similar problems, so he let go of his scientific dream.

After entering Shanghai Jiao Tong University, he wanted to build a company like Xerox that could change society. In his view, entrepreneurship could help and influence far more people than staying in a laboratory. His first project was a health records platform that let people upload medical records and manage them online. It raised 2.5 million yuan in angel investment and hired more than 30 people, but ultimately failed.

He put the remaining more than 1 million yuan into his second project. That was 2011, when he was a junior. The project was called Teambition, a lightweight collaboration tool for small and medium-sized teams and one of China's earliest collaborative office SaaS products. With a background in economics and management, he spent half a year teaching himself programming and wrote 220,000 lines of code.

In the following years, Gobi Partners, IDG Capital and other institutions successively came in: a $5 million Series A, and a $12 million Series B led by Northern Light. It took about three years to go from angel round to Series B. In 2016, Tencent made a strategic investment; in 2019, Alibaba acquired Teambition for about $100 million.

He was selected for the Forbes China 30 Under 30 list and was regarded as one of the representative figures of post-90s entrepreneurs.

After the acquisition, he stayed at Alibaba and became the person in charge of Alibaba Cloud Drive, at grade P9. Alibaba Cloud Drive officially launched in 2021, and the original Teambition cloud drive service was subsequently discontinued, with user data merged and migrated over. After leaving Alibaba in 2021, he went quiet for a period.

He returned to public view in January 2023. Qi Junyuan joined ByteDance as vice president of product at Feishu, reporting directly to Feishu CEO Xie Xin. A year later, he moved to the Flow team responsible for AI applications and took over Doubao's PC client.

That was precisely when large-model applications were at their most intense, with ByteDance placing heavy bets on Doubao, and the PC client was an important front. In September 2025, ByteDance adjusted its AI line, and the Doubao PC client, mobile client and model strategy teams began reporting to Zhao Qi.

Two months later, media reported that he had resigned. From joining ByteDance to leaving, it was less than three years.

From Teambition to Alibaba Cloud Drive and then to ByteDance's Doubao, Qi Junyuan's resume covers almost the three finest product battlegrounds in China's internet: collaborative office, cloud storage, and large-model applications.

This kind of product experience spanning SaaS, cloud computing and AI is extremely rare in the current entrepreneurial environment.

More critically, he personally experienced Teambition's complete cycle from 0 to 1 and from 1 to acquisition. The caliber of this kind of "all-around" entrepreneur is precisely the core reason top institutions dare to make heavy bets at the angel round.

1.3 billion, betting on a "proactive" assistant

About four months after leaving ByteDance, on March 4, 2026, Today AI (Shanghai) Intelligent Technology Co., Ltd. was registered and established, with registered capital of $10 million and Qi Junyuan as legal representative. The company is called "Cijian," while the product is simply called Today—both landing on the word "present."

When the company first emerged, there was plenty of outside speculation. Z Finance reported at the time that the company had completed two rounds of financing, with investors including top dollar funds such as IDG, at a valuation of several hundred million dollars. Other reports said the company's direction was large-model-driven intelligent terminals and GUI Agents, leading many to speculate that Qi Junyuan wanted to build an "Agent phone."

Half a year later, what is in front of users is a personal AI assistant, usable on iOS, Android, Windows, Mac and Linux. Today had previously been in public beta overseas, and the domestic version launched on September 25. The basic version is free, and new users can receive 100 million tokens and one month of Pro membership.

Compared with common chatbots, Today's biggest difference lies in the word "proactive." According to the official description, it can understand users over long-term interaction and offer assistance before the user even asks. The official list of uses is wide-ranging, from schedule manager and travel planner to study partner, and even a listening companion.

Specifically, after the user authorizes it, Today continuously reads emails, calendars, notes and files, remembering important people, the user's preferences, things promised to others and tasks not yet finished. When something progresses, it proactively reminds the user; when action is needed, it can also invoke a cloud computer to continue working in web pages, files and other applications.

The domestic version is connected to tools such as WeCom, WeRead, Baidu Netdisk, Kingsoft Docs and Tencent Meeting, and can record work group messages, organize reading notes, find documents in the cloud drive and generate meeting minutes. Another media outlet once ran a test: it asked Today to arrange an interview. It found the contact from the mailbox, checked the calendar, gave a time suggestion and drafted the email, and only sent it after the user confirmed. The next day the other party replied, and without waiting for a new instruction, it connected this email with the previous day's task.

Checking email, reading calendars and sending messages are not new capabilities. What Today wants to do is keep a matter "alive" after the conversation ends. Users also do not need to open a separate window or create a project for each matter; work and life matters can be put into the same dialog box, and it sorts them out on its own.

But being proactive has another side. A media writer used it for ten consecutive days, and Today sent him topic suggestions at 9 a.m. every day, and even proactively asked how far the draft had progressed, which surprised him considerably. At other times, though, it seemed to lack a sense of proportion.

When installing a game mod, it gave several unverified solutions in a row, and the author fiddled with it for over an hour without solving the problem, until he asked it to check the official documentation to find the issue. A 60-minute football training plan that was meant to be read once and used was treated as a background task and ran for nearly an hour, finally producing a web page.

The author's view is that it still cannot tell which matters are light and which are heavy. Moreover, the more an assistant remembers and the more proactive it is, the more users will care about its accuracy, permission boundaries and privacy protection.

Others overseas are already on this path. On September 22, 2025, OpenAI's proactive feature ChatGPT Pulse launched in preview, pushing personalized briefings to users every day. Exactly one year later to the day, Today's domestic version launched.

According to a review, Five Source Capital also invested in AirJelly, a proactive desktop Agent built by another ByteDance-affiliated team. Tencent and Ant Group both hold their own super entry points, yet this time they invested together in an external startup; Transsion's main battlefield is phones in emerging markets, which more or less matches the earlier "Agent phone" rumor.

Someone who has worked on collaboration tools, cloud drives, Feishu and Doubao—this resume itself may be the most weighty part of this funding round.

Young prodigies who left big companies to start businesses

Qi Junyuan is not the first entrepreneur to come out of a big company, nor will he be the last.

Statistics show that in this wave of large models, at least 30 ByteDance mid- and senior-level employees have left to start businesses. Previously, someone exposed the unique betting method of large-model investors: staking out near ByteDance office buildings in Dazhongsi and the North Third Ring Road in Beijing and in Yangpu, Shanghai, tracking personnel movements on LinkedIn and Maimai, with investment managers' spreadsheets recording candidates' job levels, reporting lines and products they had led.

Some had not yet left their jobs when term sheets (TS) were already delivered to them, and some institutions directly handed out blank TS for entrepreneurs to fill in themselves. Within the circle, a term was coined for "Ex-ByteDance": the "valuation multiplier."

The names could fill a long list. Chen Mian, who came from CapCut, founded LiblibAI and the design agent Lovart, completing a nearly $300 million B+ round in the first half of this year at a post-investment valuation of over $2 billion.

There are younger ones too. Pan Yuyang, born in 1997, experienced the 0-to-1 journey of Coze and Doubao smart glasses at ByteDance, then left to build the AI smart necklace OdyssLife, raising nearly 200 million yuan across multiple rounds, with Sequoia China and Monolith Capital leading successively.

Liang Chenqi, the person in charge of the AI companion product "Maoxiang," turned to AI comics and secured money from IDG Capital and Meituan in the angel round. Outside ByteDance, Lin Junyang, former head of Alibaba's Qwen, founded an AI lab after leaving; the company had no product yet, but its first post-investment valuation reached about $2 billion. Even interns became sought after: 17-year-old Chen Guangyu had just posted farewell to Kimi when invitations came in.

Most of these people are young and have led a recognizable product at a big company. What investors value is often not entirely what they are doing now, but where they honed their skills. The basic judgment is that when the technology path is still unclear, investors treat a big-company background as a risk-control tool.

To some extent, big companies have become way stations for prodigies.

More than a decade ago, Qi Junyuan was the student who secured funding in his junior year and, with a collaboration tool, raised all the way until being acquired by Alibaba. Put in today's terms, he would probably be exactly the kind of "prodigy" investors talk about. More than a decade later, after spending several years each at Alibaba and ByteDance, he walked out with a longer resume.

But no matter how star-studded the list or how high the valuation, it is only the beginning. Today is still learning how to be an assistant that "knows its limits," and this bet is much bigger than back then.

The answer, in the end, must be left to the product.

This article is from the WeChat public account "Rongzhong Finance" (ID: thecapital), author: Wang Tao, editor: Wuren, published by 36Kr with authorization.