Already seated at the very front of the table.
Recently, Hege (Shanghai) Intelligent Technology Co., Ltd. (hereinafter referred to as "Today AI") announced the completion of a cumulative funding round of over USD 200 million. The investors include IDG Capital, Tencent, 5Y Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Ventures, and SJTU Hanyuan Assets.
The Today AI team is dedicated to building personalized, proactive AI systems, providing everyone with an AI assistant that understands them better and takes more initiative. From internal testing to its domestic launch, Today AI has become a schedule butler, life helper, work assistant, travel customizer, personal recorder, health assistant, and learning sparring partner for many users.
This funding comes at a time when the Personal Agent track has been completely ignited by a single product.
Personal Agent: A New Trend
On September 8, Meta officially launched the personal agent Muse, defining it as a Personal Agent for ordinary users, which is a personal intelligent agent.
The release of Muse quickly caused a sensation. It not only topped the App Store charts but also became one of the only dozen or so mobile apps in history to surpass 10 million downloads in the US within 40 days of launch, with some metrics even exceeding the early performance of ChatGPT.
The explosive popularity of Muse has pushed Personal Agent from a cutting-edge product concept to a new competitive direction where tech giants must secure their positions.
Before this, the main battlefield for AI Agents was mostly concentrated in office scenarios. The operational logic of office agents usually revolves around "tasks." For example, if you need to fix a bug, it will read the relevant code, error files, and test records; once the task is completed, this set of contexts is cleared, waiting for the next instruction.
However, in real life, user needs are often vague and long-term. What users tell the AI might just be "I want to go out and play for a few days next month" or "I want to make my life more regular recently." This requires the AI to not just perform "one-off tasks," but to remember the user's status, habits, and preferences over the long term, building the context around the "person."
Take Muse as an example. It runs on an independent cloud virtual computer, has its own browser, and can open web pages, fill out forms, and book trips on behalf of the user.
If a task takes a long time, it can continue to execute in the background even if the user exits the app. This ability to continuously provide services from the context of the "person" is precisely the core barrier of a Personal Agent.
This category quickly triggered a chain reaction. On September 28, Manus launched the independent app Cue; on September 29, OpenAI released Dots integrated into ChatGPT; on October 13, Qwen launched "Qwen Go," and at the same time, news emerged that Tencent Yuanbao is developing "Yuanbao Assistant," which may be made into an independent app.
Among domestic startups, Qi Junyuan's Today AI did not enter the market following this string of releases. The global version of Today AI had already opened for public beta testing on September 8, the same day as Muse, while the Chinese version was opened for download on September 24.
What Product is Today AI Building
Currently, Today AI has integrated with commonly used domestic applications such as WeChat and email, covering all platforms including Mac, Windows, Linux, iOS, and Android, with the basic version being permanently free.
As a proactive Personal Agent, the most core underlying design of Today AI is "long-term memory."
The product has a built-in dedicated Memory system that remembers the projects the user has been working on for a long time, the people they know, personal preferences, and unfinished tasks. Under this system, all conversations between the user and the AI take place in the same long-term conversation. Users do not need to open a separate dialog box for each new task, nor do they need to repeatedly explain the background.
Based on long-term memory, Today AI can proactively intervene in the user's workflow. When a new email arrives, the schedule changes, or a task reaches a new status, it can proactively reach out to the user and invoke tools such as the browser and files to continue processing.
This proactivity is also reflected in the daily interaction rhythm. Every morning, Today AI will organize a Morning Brief based on the schedule and recent changes; in the evening, it will re-summarize the unfinished tasks of the day. If it finds that something related to the user has changed in the meantime, it will also proactively remind them.
From a technical architecture perspective, what supports this experience is a complete set of Agent infrastructure: model orchestration, task planning, memory retrieval, task queues, and cross-device synchronization.
In terms of product form, Today AI maintains a continuous conversation flow on the left side, while the right side is a personal master panel that centrally displays the status across three dimensions: "Today," "Tasks," and "Memory." Most requests are still thrown into the same main conversation, scheduled by the system in the background. Projects, relationships, preferences, and unfinished tasks are therefore kept in the same thread, rather than scattered into newly opened dialog boxes.
The Today AI team believes that this accumulation will become a barrier. When a personal AI assistant serves a user, it will continuously accumulate the user's long-term data, and thus become increasingly adapted to them. After using it for a period of time, such products are not easily replaced. When the basic capabilities provided by various assistants become similar, the experience generated from long-term use and the switching costs become the primary barriers.
From Teambition to Today AI
The founder, Qi Junyuan, is not an unfamiliar name in the venture capital circle.
While still a junior at Shanghai Jiao Tong University, he partnered with two classmates majoring in information security to register a company, initially working on health records. The team once had over thirty people and received an angel investment of about CNY 2.5 million. However, due to the immaturity of the online medical market, the project ended in failure.
Teambition was the second startup that grew out of this team's internal tools, officially released in June 2013. It is considered one of the earliest batch of team collaboration SaaS products in China. It received strategic investment from Tencent in 2016 and was acquired by Alibaba in 2019.
After Teambition was acquired, Qi Junyuan joined Alibaba, taking charge of products such as the Alibaba Cloud Developer Suite, DingTalk Collaboration Suite, and Alibaba Cloud Drive. In 2023, he joined Feishu as Vice President of Product; subsequently, he transferred to ByteDance's Flow department, responsible for the PC version of Doubao.
It was not until November 2025 that he chose to leave ByteDance and embarked on a new entrepreneurial journey.
In fact, as early as 2014, Qi Junyuan bought the domain name Today.ai. Now, this domain name has finally found its owner.
The Personal Agent track is now crowded with giants. ByteDance and Alibaba are both increasing their bets, and startups are emerging one after another. Yet Qi Junyuan still secured an extremely luxurious funding lineup: IDG Capital, Tencent, 5Y Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Ventures, and SJTU Hanyuan Assets.
Behind this investment lineup, there is a clear interweaving of logic from all parties. The rare joint appearance of Tencent and Ant Group indicates that the two giants have reached a consensus on the judgment that "Personal Agent is the next-generation entry point"; the entry of IDG and Sequoia represents the recognition of the track and the team by top financial institutions.
And in this list, the appearance of SJTU Hanyuan Assets is particularly logical. Qi Junyuan is an alumnus of Shanghai Jiao Tong University. This kind of capital with a university background usually means a deep binding between the founding team and the alma mater in terms of technical resources and talent delivery.
From Teambition to Today AI, Qi Junyuan has spent more than a decade pushing the matter of "helping people manage information and tasks" from a collaboration tool to a more imaginative personal intelligent agent. On this track that has just been ignited, Qi Junyuan and his team have already taken their seats at the very front of the table.
Text | Zhou Yu Editor | Yida