In the era of AI, what should a reimagined DingTalk look like?

Author | Zhao Weiwei

The familiar and controversial No Zhao took the stage, still donning a black T-shirt with the iconic phrase "Be Crazy" emblazoned on the back.

This marked No Zhao's first appearance at a press conference since his return to DingTalk.

During the announcement of AI DingTalk 1.0, he mentioned rival products twice. Firstly, he highlighted DingTalk's smart voice capability, "AI Listening and Notetaking," which achieved a 97% recognition rate after enterprise-specific training, surpassing competitors. Secondly, he emphasized DingTalk's AI sheets, which processed 1 million rows of calculation data in 11.6 seconds, compared to 26.48 seconds for the competitor.

No Zhao described the competitor as "very excellent" and "very advanced," and the audience implicitly understood that he was referring to ByteDance's Feishu.

At this conference, No Zhao unveiled over 10 new DingTalk products, including the voice assistant smart hardware DingTalk A1, DingTalk One for seamless work information flow management akin to swiping through short videos, the enterprise search engine AI Ask for single-query answers, an AI sheet with enhanced performance in collaboration with Alibaba Cloud, AI Listening and Notetaking created with Alibaba Tongyi Lab, innovations tailored for customer service, marketing, and education scenarios, and multiple AI assistants for startups.

DingTalk, now 10 years old, is bidding farewell to the past with a new AI presence. The current DingTalk 8.0 version represents the 1.0 version of AI DingTalk. No Zhao stated, "We are determined to clear the past and adopt a mindset of starting from scratch to create a brand-new DingTalk for the AI era."

"We considered Feishu, but the migration cost was too high, so we gave up." A DingTalk customer in Hangzhou told "Lan Dong Business" that his company, specializing in exporting household products, has 3,000 employees and has been using DingTalk since 2018, with an annual payment scale of around RMB 100,000. His company has utilized DingTalk's AI sheets relatively little, with most AI experiences concentrated in products such as knowledge bases and documents.

He looks forward to the new features of DingTalk products bringing changes to his company, and there are 190,000 other paid DingTalk enterprises like his. Over the past four months, under No Zhao's leadership, DingTalk has been contemplating how to reinvent itself in the AI era to meet the future needs of these customers.

With product iterations and team rejuvenation, DingTalk has adjusted its posture to face market competition.

If in the first five years, DingTalk answered how to separate work and life, taking a path distinctly different from WeChat for Work; then in the next five years, DingTalk must answer: How to differentiate itself from Feishu in the AI era?

The DingTalk led by No Zhao is bound to provide a thoughtful answer in the fierce competition.

1. No Zhao's Unwavering Commitment

On the new DingTalk, there's a casual new feature called "Dingtalk Zen."

In "Dingtalk Zen," relaxing and meditative scenes play, aiming to allow users to rest their eyes after working for a long time. Whether it's mountains or seas, these natural sceneries and sounds come from real footage taken by DingTalk employees during global travels. No Zhao said, "DingTalk's colleagues are not only about work as everyone imagines; we also have wonderful lives."

The audience laughed.

This seemed to be a response to the controversy over his "midnight inspection" that emerged some time ago. He also used several numbers to showcase the workload over the past four months: We analyzed and designed a total of 1,850 requirements, fixed 574 issues reported by users, and conducted comprehensive optimizations on more than 20 product lines, resulting in the brand-new DingTalk 8.0.

No Zhao hasn't changed. He remains aggressive in management, possesses strong product capabilities, and is adept at discovering and solving user pain points.

He compared DingTalk four months ago to an old house, and the first thing he did upon his return was to renovate it. He conducted a comprehensive user survey, visited a large number of customers, and received numerous voices. "Feedback indicated that after 10 years, DingTalk's design looked outdated, with too many features crammed onto the screen, making it feel particularly bloated."

Perhaps the best way to solve the bloat problem and showcase the essence of this renovation is DingTalk One, currently in beta testing.

In simple terms, DingTalk One is designed as a unified entry for humans and AI to converse through natural language, turning "people finding tasks" into "tasks finding people," giving users an AI secretary that sorts all the information on the DingTalk interface. Users only need to swipe through the mobile phone screen, similar to swiping through short videos, to process work content one by one without repeatedly jumping between complex information streams.

Behind this way of working is the comprehensive implementation of Agent capabilities, including Message Agent that comprehensively processes messages and analyzes priorities, Approval Agent that analyzes and sorts approvals, and Agent that analyzes all schedules, analyzes meetings, and prepares relevant materials in advance. Users can hold down the One button to engage in voice dialogues, instructing various Agents in DingTalk to handle tasks for them, such as initiating meetings, to-dos, and searching for information.

Work is only one aspect; learning ability is another facet of DingTalk One.

DingTalk One also has 9 learning Agents that bring industry content such as global references and Github selections, allowing the learning content of users to flow within the company, sparking team discussions, and then becoming the company's knowledge assets.

No Zhao hasn't changed. He won't easily give up on things he likes, such as hardware. During his period of re-entrepreneurship after leaving DingTalk, he worked on hardware, and now DingTalk has also returned to the tradition of self-developed hardware. Behind DingTalk A1 is a R&D team of 40 to 50 people at DingTalk. No Zhao said, "They probably sleep for no more than 5 hours a day."

DingTalk previously made hardware such as punch-in machines, attendance machines, and video conferencing machines, but during Ye Jun's management of DingTalk, most of these hardware products were optimized away.

Now in DingTalk AI 1.0, hardware has once again become important. Among the new products unveiled by No Zhao is a smart voice hardware product, DingTalk A1. Its appearance resembles a slim version of Apple's magnetic charging bank, but with an additional small screen. It is equipped with a 6-nanometer AI chip, and after the user presses the button, it can perform visual recording and intelligent dictation at any time.

What seems like a simple hardware product is actually a hardware + membership payment model. This product comes in flagship and youth editions, with complimentary free voice intelligence benefits on DingTalk, including 1,000 minutes of smart voice usage per month and 10GB of storage space, priced at RMB 799 and RMB 499, respectively.

Regardless of the outcome, the obsession with smart hardware that connects AI to the physical world will persist in DingTalk for a long time to come.

In the easter egg at the end of the DingTalk press conference, No Zhao unveiled the next generation of DingTalk - DingTalk Real. Covered in a veil, its appearance resembles a computer mainframe, allowing DingTalk to physically exist by the user's side. Its function is to keep DingTalk working on your computer, accessing all your data within the internal environment, and continuously solving problems and discovering opportunities for you 24/7.

What exactly is the next generation of DingTalk? How can AI get closer to the physical world? This answer can only be revealed by No Zhao again.

2. The Indispensable Role of Sheets

Just one week before the DingTalk press conference, WeChat for Work released its AI-comprehensively-applied 5.0 version, with a smart sheet capable of holding 15 million rows of data per sub-sheet.

Smart sheets and the AI-powered intelligent office behind them are the battlegrounds for the three giants of Feishu, DingTalk, and WeChat for Work. Using smart sheets to manage projects, companies, and customers means that a company has truly achieved digitalization, enabling large models to automatically analyze and produce content, making data "speak" and truly useful for users.

Tencent, Alibaba, and ByteDance are all gathered in the AI office space, but the game between Feishu and DingTalk is the most apparent in this competition.

For example, in July, Feishu announced that its multi-dimensional sheet could handle 10 million hot rows per sheet, becoming the first business system among similar global products to support 10 million rows per sheet. Ma Ruila, the head of DingTalk AI sheets, said at the DingTalk press conference that DingTalk hopes to achieve real-time calculation of 10 million hot rows per sheet by the end of this year, which is also a common goal of DingTalk and the Alibaba Cloud database team, "We will definitely achieve it by the end of the year."

Ma Ruila cited a case where a DingTalk customer had previously decided to use Feishu's multi-dimensional sheet for a project. "He felt that this tool was indeed very powerful, but he didn't know where to start or how to use it," so later DingTalk collaborated with this customer on AI sheets. Ma Ruila believes that AI brings tremendous change, and DingTalk needs to lower the barrier to entry, making a powerful tool truly convenient for everyone to use.

Essentially, DingTalk aims to lower the threshold for users to use AI sheets, allowing users to issue commands using only natural language.

Whether it's adjusting the width of the sheet or highlighting all transactions with negative profit margins in the sheet, AI sheets will translate these natural language commands into corresponding sheet operations, automatically completing calculations, filtering, and visualization. Moreover, every field in DingTalk's AI sheets has out-of-the-box AI capabilities, with over 100 Agents that can directly generate images, translate languages, summarize, analyze resumes, etc., within the sheets.

The competition between Feishu and DingTalk is not only about the AI capabilities of sheets but also about their popularity.

Both Feishu's and DingTalk's sheets are stepping beyond the products themselves. In August, Feishu's multi-dimensional sheet announced that it could be used as an independent product without downloading or registering for Feishu, supporting other platforms. At the DingTalk press conference, Ma Ruila also showed the independent website of AI sheets, meaning that DingTalk's AI sheets can also be used alone without downloading DingTalk.

This competition in the depth and breadth of AI sheet usage will not stop. Ma Ruila previewed subsequent actions for DingTalk's AI sheets, including integration with all DingTalk products, from attendance approval to marketing and education scenarios, and also preparing over 2,000 external data connectors with ecological partners, allowing both internal and external company data to be easily aggregated into a single AI sheet.

The fierce competition between Feishu and DingTalk in the field of AI sheets represents how current large models are changing internet collaborative offices, reflecting the maturity of automation after the integration of AI and products. Regardless of who is ahead, it will inevitably lead to an improvement in product AI capabilities, ultimately gradually driving progress in corporate productivity.

After all, No Zhao has only been back for four months, and the changes at DingTalk have only just begun.

Editor | Chen Qiulin