With Over US$200 Million Raised, Today AI Wants to Turn “Artificial Intelligence” into “Intelligence for Everyone”

The funding comes amid growing interest in personal AI assistants. Over the past year, AI has progressed from answering questions to carrying out instructions, with searching emails, accessing calendars, and operating computers gradually becoming basic capabilities. In the past month alone, personal AI assistants have emerged overseas as another highly sought-after application category after AI coding, with technology giants entering the field in succession and capital flowing in faster.

This wave of interest seeks to address a longstanding, unresolved question: whether AI is answering questions or following instructions, users must first explain their needs and provide context. Today wants to try a different approach: remembering the goals and recent circumstances users have already shared, then proactively suggesting the next step at the right moment.

Within China’s AI sector, Today AI has chosen to build an independent product centered on a personal AI assistant, making this the company’s core focus. Today AI is among the earliest startup teams in China to explore this direction. The team is dedicated to building a personalized, proactive AI system that gives everyone an assistant that understands them better and takes more initiative.

“The capabilities are here, but the relationship has yet to be built,” the Today team once said in explaining its starting point.

AI can do more and more, yet using it effectively still takes considerable effort: first deciding what to ask, then explaining the background, and often starting all over again the next time. Today wants to spare users that effort and proactively lend a hand when the time is right.

Achieving this requires gradual adjustment through everyday use. That is also why Today chose to put the product in users’ hands first: it began overseas public beta testing in September, launched its domestic version on September 24, and only now formally disclosed its funding. Before this announcement, many users were already using it to plan their work, organize trips, schedule workouts, and manage their calendars, putting the product to the test in real-life situations.

1. Remembering Your Past While Keeping Up with Your Present

The help a person needs often depends on their circumstances at that particular moment.

When traveling, they may not start thinking about where to go the next day until the evening. After a holiday, their plan to resume exercise may also need adjusting based on how they have been doing recently. Many needs emerge gradually as life changes, making them difficult to turn into a clear AI instruction in advance.

This places greater demands on a personal AI assistant: it must remember the experiences and preferences users have shared while also keeping up with their latest plans and circumstances. Only by connecting the two can AI bring relevant matters to their attention in time.

One user staying in Tibet had previously asked Today to organize travel photos and other materials. Recently, he was considering his itinerary for the following day. That evening, Today proactively recommended a list of lesser-known temples frequented by local Tibetans in his briefing, complete with bus departure times, opening hours, and the best lighting conditions for photography.

The recommendation came as a pleasant surprise because the temples matched his interests. The photos and travel experiences organized on the earlier occasion proved useful for this outing.

According to the Today team, the product primarily remembers basic information users voluntarily share, along with preferences expressed in previous interactions. Users also manage these memories themselves. They can review the personal details and preferences Today has recorded, correct inaccuracies, and delete anything they do not want retained.

But being helpful also requires understanding a user’s current circumstances.

One user who has practiced boxing for a long time regularly analyzes her exercise records with Today and has connected her health data. On her first day back at work after the holiday, Today noticed that her activity levels had dropped significantly during the National Day break. In her morning briefing, it reminded her not to rush into higher-intensity exercise when resuming training.

The reminder was simple, but it took changes in her recent routine into account. Even for the same person, appropriate advice may vary with time and circumstances.

Proactive assistance, however, also demands better judgment from AI: which information is worth remembering over the long term, and which needs updating promptly? When should it send a reminder, and which matters do not warrant interrupting the user?

These small judgments determine whether AI can help at the right moment and how much of the user’s everyday burden it can actually ease.

2. Lowering the Cost of Delegation So More Tasks Are Worth Handing Over to AI

Delegating many tasks to AI still involves repeatedly explaining the background and checking progress, potentially requiring more effort than doing the work yourself. The Today team calls this the “cost of delegation” and hopes to reduce it, making it more natural for users to hand tasks over to AI.

Scheduling is one example: plans often struggle to keep pace with change. One user preparing for the College English Test Band 6 regularly uses Today to organize her study plan and daily schedule.

On one occasion, she received an email accepting her as a volunteer and learned that she needed to attend a training session the following morning at short notice. Taking her existing study plan and the new appointment into account, Today suggested moving her vocabulary study session to the evening.

She did not have to explain her study arrangements again or work through her calendar to reschedule the session herself, reducing the cost of delegation.

This is also why Today organizes information around the same individual: arrangements across study and everyday life often affect one another, and a last-minute change can disrupt an existing plan.

The same thinking is reflected in its interaction design. Today uses a single, continuous conversation in which users can introduce different matters one after another, while individual tasks proceed in parallel in the background and their progress can be checked separately. Users do not need to switch frequently between conversations, and they can pick up where earlier discussions left off.

In the Today team’s own words: “You hand it over, then go live your life. When you come back, it’s still there.”

A personal AI assistant’s initiative also requires clear boundaries on execution. In the study-planning example, Today only suggested an adjustment; the user decided whether to make it. Payments and important decisions likewise require user confirmation before the relevant actions are carried out.

Reducing the cost of delegation is intended to help more people use AI effectively and reserve their energy for more important things. Achieving that requires far more than helping out just once.

3. Lasting Help Depends on Work Behind the Scenes

For this kind of assistance to continue, AI needs to remember users’ circumstances accurately, provide timely reminders, and complete tasks successfully.

The Today team continues to invest in these areas. In memory management, it uses a Dreaming mechanism, compression, and updates to reduce the influence of outdated information on judgment. For proactive reminders, it adjusts its notification strategy based on whether users read and respond to messages, reducing unnecessary interruptions.

Meanwhile, the team is refining its system infrastructure, task queues, and communication across devices so that long-running tasks do not stall or get lost and users can check their progress at any time.

In Today’s view, a personal AI assistant needs a good product, suitable models, and solid infrastructure working together. Creating an impressive demonstration once is not difficult; being helpful every day is the hard part. Much of that work is not readily visible from the outside.

For a personal AI assistant that needs to understand users over the long term, another question deserves an answer: does AI really get to know you better the longer you use it?

A one-off AI task can usually be evaluated by its output. A personal AI assistant, however, also needs to be assessed on how it changes over time: as historical information accumulates, does its next suggestion become more relevant? Can it identify needs and complete tasks more accurately?

To explore this, Today has built a simulated-user evaluation system. The simulated users have continuous life histories, and the team compares AI performance under three conditions: with no historical information, with only a personal profile, and with 30 days of accumulated interaction history. It then observes how the effectiveness of its assistance changes.

Through this approach, Today hopes to verify whether memories accumulated over time can translate into more accurate and effective help. The team says it will gradually make its research methods, test scenarios, and interim results public, and welcomes more researchers, developers, and users to examine them together.

Today still has many areas to refine. Ease of operation, conversational fluency, and the clarity of feature guidance are all aspects the team continues to improve based on user feedback.

Today acknowledges that this is only the starting point. Much work remains across the product, models, and engineering to enable AI to understand a person over the long term and provide help at the right moment.

The more than US$200 million in cumulative funding gives Today the resources to continue this work, which requires sustained investment. It also allows the team to keep moving step by step toward its vision of “Intelligence for Everyone”: enabling more people to afford AI and use it effectively, receive assistance of their own even without technical expertise, and have more time for the lives they care about.

The team is also recruiting more people to join it on this journey. If this future excites you, and you are willing to commit for the long term and help build it yourself, Today welcomes you to join.