Article reprinted from: AI Navigator
Source: AI Whale Selection
Author: Brother
Image source: Generated by Unbounded AI
Just as OpenAI's Sora became popular and Google's Gemini was released, domestic large-scale model companies began to fall before the dawn of commercialization.
On February 21, the well-known AI startup Emu Intelligence suddenly announced that it would suspend operations for six months due to reduced business demand, which brought serious cash flow pressure and challenges to the company.
The AI company Emu Intelligence has a notable background. It was founded in 2015 by Jian Renxian, former vice president of Microsoft (Asia) Internet Engineering Academy. It claims to have developed a "multimodal emotion recognition model" independently through human-computer interaction technology using text, images and voice.
Brother Jing also visited the company's booth at the 2023 Shanghai World Artificial Intelligence Conference. At that time, the booth was bustling with people. Unexpectedly, six months later, the company went into a state of "shock".
The highlight moment of Bamboo Smart
In Caohejing, Shanghai, many innovative AI companies are located here. Well-known AI companies such as ByteDance, SenseTime, StarRing Technology, and Bamboo Intelligence gather here. Bamboo Intelligence is one of the companies that has developed well.
Judging from the financing history, there are rounds of financing of over 100 million yuan almost every year, with a total financing amount of 100 million US dollars + 400 million RMB, which makes this company's financing amount exceed 1 billion yuan.
Zhujian Intelligence has developed a series of NLP technology platforms, such as Bot Factory, Knowledge Factory, Emoti Coach, etc. Based on these technology platforms, Zhujian Intelligence has also launched a series of products, which have been used by seven or eight hundred customers from industries such as banking, securities, insurance, manufacturing, and consumption.
Realizing early on that the era of big models was coming, in 2023, Ezujian Intelligence released a 1+4 product strategy driven by AI big language models.
Zhujian Intelligent 1+4 Product System
This system is actually to help domestic large models build the last mile of commercialization. On the EmotiBrain large model, you can choose general basic models such as Huawei Pangu, ChatGLM, MOSS, MiniMax, LlaMA2, Baichuan, etc., and use fine-tuning technology to train your own proprietary model.
Because Zhujian Intelligence's NLP technology can help companies better train data knowledge, and then through its AIGC products: KKBot, AI Coach, Knowledge Factory and Magic Writer, it can be applied to marketing, customer service, sales training, knowledge management and copywriting scenarios.
In other words, in the process of helping enterprises apply big models, they can choose better big models to train enterprise-specific big models. Because they incorporate the enterprise's own knowledge, its several AIGC applications will also have better results.
Therefore, from the promotional caliber, we can see that Zhujian Intelligence has created a "model factory" that can help 90% of companies deploy AIGC within one year.
Suddenly fell before the dawn of commercialization
Public information from the past one or two years shows that this company has quite a few partners.
In March 2023, Zhujian Intelligence announced a strategic partnership with Inspur and Zhipu AI to further promote the application and commercialization of artificial intelligence technology.
At the same time, Zhujian Intelligence is also a deep cooperation partner of Huawei Cloud and has participated in the core part of Huawei Pangu Big Model. Zhujian Intelligence COO described it as an excellent AI industry landing partner of Huawei Cloud.
The upstream companies are large companies such as Huawei and Zhipu, and the downstream customers include more than 300 large enterprises including China Minsheng Bank, GAC Group, China UnionPay, etc. The number of ecological partners exceeds 50, and the cumulative reach is hundreds of millions of end users.
In the era of big models, Jian Renxian, CEO of Zhujian Intelligence, emphasized that enterprises do not need models with too large parameters. "Our general big models and model factories are deployed to the client. Combined with the customer's data and industry know-how, and then parameter adjustments, we can fine-tune and train a 7B-65B big model."
From this point of view, there seems to be no major problem with the customers and ideas. But in fact, the company's problems may have historical roots. As an AI company that started with NLP, CEO Jian Renxian emphasized the "NLP big model dual engine". Industry insiders analyzed that when the ChatGPT wave hit, Zhujian Intelligence was still unwilling to give up NLP, which may have caused problems for the company's business.
Therefore, Zhujian Intelligence mentioned in an internal open letter that it is adjusting its business direction. During this period, it will give priority to serving existing customers and strengthen its exploration of new technologies and new products. We will also optimize the loss-making business lines, and some departments and positions will be suspended from now on and reorganized.
Big models and AIGC are not a panacea
Enterprises are still in the wait-and-see stage of applying big models. The cost of privatizing big models is 2-3 million yuan, which is not something every enterprise can afford at any time. Zhujian Smart's model is cheaper, but it also needs to wait for enterprises to have the necessary understanding and decision-making.
But after the last round of D+ financing before December, it was obvious that Zhujian Intelligence did not wait until the commercialization wave of large models and AIGCC hit, and suddenly announced the suspension of work and production.
Zhujian Intelligence still retains the opportunity for employees to return:
The work and production will be suspended for 6 months from February 20, 2024, and the suspension time is: 10:00 on February 20, 2024 - 19:00 on August 19, 2024. During this period, we will continue to pay attention to market trends, actively seek opportunities for business development, and ensure care for every team member. The company will keep in touch with you at any time and convey information about returning to work or other relevant arrangements in a timely manner. Wage standards during the suspension and vacation period According to Article 12 of the "Shanghai Municipal Human Resources and Social Security Bureau's Notice on Issuing the <Shanghai Enterprise Wage Payment Method>", the salary payment cycle is calculated from the first day of the suspension and vacation to the end of one month, and the employees' wages are paid as agreed. A full month of vacation is deemed to exceed one salary payment cycle, and the company pays the Shanghai minimum wage standard in accordance with regulations.
From the current perspective, in China, only large-scale AIGC companies such as Baidu, Alibaba and heygen (an overseas digital human company) have maintained a monthly revenue standard of over one million US dollars.
More large-scale models and AIGC companies still find it difficult to find market scenarios with outstanding commercialization effects, which results in them gaining popularity but not making money.