Baichuan's “Tail - Cutting”: Abandon Finance, Protect Healthcare, Streamline the To B System | Exclusive Report from Intelligence Emergence

Text by | Zhou Xinyu

Edited by | Su Jianxun

(This article is an exclusive report by Intelligence Emergence. If you need to quote or reprint, please indicate the source. Any infringement will be pursued!)

Intelligence Emergence has learned from multiple independent sources that the following changes have taken place in the To B team of the AI unicorn "Baichuan Intelligence", founded by Wang Xiaochuan recently:

On March 3, 2025, the B - end group mainly responsible for the To B business in the financial industry was laid off, and all employees signed the separation agreements on the same day.

On February 19, 2025, the PE team (prompt engineering team) belonging to the To B business team of Baichuan Intelligence was transferred to the product R & D group that manages the algorithm team.

As of the time of publication, Baichuan Intelligence has not responded to the above information.

Within Baichuan, the To B business used to be the heart of blood - making. The B - end group established at the beginning of Baichuan's founding was mainly responsible for the To B business in the financial field. Led by Deng Jiang, the former vice - president of Zhongguancun Kejin, it reported to Ru Liyun, the co - founder of Baichuan Intelligence.

△ Deng Jiang, the commercial partner and president of the financial business group of Baichuan Intelligence. Image source: Internet

Since 2024, the B - end group has brought customers such as iSoftStone, Sunyard, and Daguan Data to Baichuan and provided them with financial - related products such as the AI fiscal and tax Q&A system.

At that time, Baichuan Intelligence was quite active in its commercial layout in the financial field. In October 2024, Baichuan launched a "one - stop large - model commercialization solution". In December of the same year, the company also released the large model Baichuan4 - Finance for the financial field.

On February 17, 2025, Baichuan held a company - wide meeting for Baichuan Intelligence around DeepSeek. At the meeting, Deng Jiang, the person - in - charge of the B - end group, also mentioned how the current B - end business of Baichuan Intelligence should be combined with DeepSeek.

However, for the employees of Baichuan Intelligence, the lay - off of the B - end group had long been foreshadowed.

Two days after the company - wide meeting, on February 19, the PE team (prompt engineering) under the B - end group was transferred to be managed by the product R & D group.

In Baichuan Intelligence, the product R & D group is the heart of all businesses and has the model algorithm team under its jurisdiction. After that, the cooperation between the To B team and the PE execution team needs to be approved by the product R & D group. At the same time, some employees of the B - end group were also laid off.

The adjustment of the PE group was regarded by many employees as a sign of the contraction of the B - end business. "Within Baichuan, the PE group is equivalent to a bridge that serves the business needs on the front line and coordinates the business and algorithms." An employee told Intelligence Emergence that since the company's founding, the PE group has belonged to the B - end group.

On March 3, the axe finally fell on the B - end group, and all employees of the B - end group were forced to sign the separation agreements. Another insider told Intelligence Emergence that in the afternoon of that day, Ru Liyun explained the lay - off of the B - end group: the company wants to focus on healthcare.

"Focus" has become the core proposition of the "Six Little Tigers" at the beginning of 2025 in terms of business. The important variable behind this is the fierce momentum of DeepSeek.

It is obvious that the open - sourcing of high - performance models such as DeepSeek R1 has had a direct impact on the To B business of the Six Little Tigers. The traditional To B models mainly based on Token - based billing, API sales, and private deployment are being replaced by the integration of DeepSeek.

Facing the challenge of DeepSeek, the current solution of the Six Little Tigers is to simplify and focus resources on the most differentiated and monetizable businesses to strengthen their barriers.

For example, Intelligence Emergence has learned that MiniMax is shrinking its B - end business team and focusing resources on its C - end products Talkie and Xingye. Similarly, Moonlight is reported to have cut its advertising budget and focused on polishing its Kimi product.

In the financial industry, Zhipu, Jieyue Xingchen, the "Four AI Dragons", and large - scale enterprises have already taken their positions. Many people told Intelligence Emergence that apart from the charisma of Wang Xiaochuan and Ru Liyun themselves, Baichuan Intelligence has no obvious competitiveness in the financial field, and there is no significant gap between its models and products and those of its competitors.

Currently, the healthcare model is the best card that Baichuan can play on the large - model table.

In an interview in January 2025, Wang Xiaochuan mentioned to LatePost the implementation plan for healthcare: the AI pediatrician will start work at Beijing Children's Hospital; in the first quarter, Baichuan's "AI doctor assistant" will be implemented in Haidian District.

An investor believes that there are currently few players in the healthcare field, which is the core barrier that allows Baichuan to differentiate itself from the other Six Little Tigers and is also the key for Baichuan to continue to obtain financing.

However, at the same time, healthcare is a hard nut to crack. Take the AI consultation system, which is commonly used as an entry point by manufacturers, for example. Although the diagnosis efficiency of AI has increased, the accuracy has not improved. The reason is that most AI manufacturers in the market cannot obtain sufficient clinical data.

At the same time, the business of each department in the hospital varies greatly, and the general system cannot meet the needs.

A practitioner told Intelligence Emergence that the core data for training the diagnosis and treatment model is not only the existing literature and books but also the clinical data accumulated in the actual diagnosis and treatment scenarios.

To enter the healthcare industry, Baichuan Intelligence has also made great efforts. For example, at the end of 2024, Baichuan invested in the healthcare data service provider "Xiao'er Fang". At the annual meeting, Wang Xiaochuan specially introduced two healthcare experts he hired from overseas and Hong Kong at a high salary, and they are responsible for leading the professional healthcare team.

Now, the greatest confidence for Baichuan to survive by cutting off the "tail" is the relatively abundant funds on its account, which has also been confirmed by many insiders. How to focus resources and build barriers in the healthcare field is something that Wang Xiaochuan needs to think about before realizing his goal of "creating doctors".

(Wang Fangyu, the author of Intelligence Emergence, also contributed to this article)

Welcome to communicate!

Welcome to follow!