At present, many companies are exploring the implementation of AI, and everyone is crossing the river by feeling the stones.
Today, the editor has compiled the AI dry goods sharing of the exclusive live broadcast of “Starting Point Classroom Members”, let’s take a look at how senior product consultant @ Xia Yehua teacher analyzes some problems encountered in the process of enterprise AI landing.
The starting point classroom has been inviting product experts from all walks of life who are exploring the implementation of AI to share their AI understanding and practice. Join the starting point classroom membership, 100+ AI era, thinking about product application implementation, AI product manager capability model construction guide, all free to learn!
Click here to learn more about the starting point classroom membership:https://996.pm/7mvqk
Three major problems in the implementation of enterprise AI
1. Whether to introduce AIGC
Enterprise business diversification, such as smart city business, smart home, smart industry, smart health business, etc., whether it should be introduced in all fields needs to be carefully considered.
2. If you decide to introduce AI, where to start? How to cut in?
As the person in charge of planning, the product manager must bear the brunt of it, so where to open it? There needs to be a logic of thinking.
3. How to do it?
There are so many large model manufacturers in the industry, including foreign and domestic, so who do I cooperate with, whether open source or closed source, and then what level of technology do we do, etc.
These questions are very complex, there are no standard answers, and we need to have the ability to analyze, and then understand the business, understand the technology, to connect with internal and external technicians. In this process, the product manager must be the coordinator
Whether enterprises want to introduce AIGC is not a multiple-choice question, but a compulsory question
Why are many companies still waiting and not making decisions today? There are two reasons:
First, the industry to which it belongs is not at a high level of digitalization. Let’s take an example, agriculture, including many government systems now, is its own information foundation is very weak, and you cannot do decoration on a very poor house frame, which is unrealistic. Therefore, they need to build their own technical base, digital systems, accumulate data, and then talk about how to introduce large models, which is determined by industry attributes.
Second, let’s look at the fast-growing industries, which are several industries with relatively high penetration rates of large models. For example, like finance and industry, why can they do it? It is because finance itself is data-driven, there is already a very strong data accumulation, and in the past, this so-called traditional AI model was used to drive intelligent decision-making of data.
So this is a fundamental proposition, a mandatory question, it is impossible to say not to do it, just to say how to choose the timing and how to do it.
Enterprises introduce the quadruple realm of AIGC
The first step is to reconstruct cognition, that is, the organization must receive relevant training from top to bottom.
There may be training related to technical architecture, and then a series of training related to strategy and organizational capabilities related to the large model product ecosystem. First of all, all managers need to have an understanding of large models at the cognitive level. After the bosses build basic cognition, they will issue the following tasks.
The second step, strategic restructuring and organizational restructuring, are usually done together by middle and senior cadres.
Then in strategic restructuring and organizational restructuring, there must first be a basic strategy to pass, that is, in the strategy of the large model, whether it will affect the business strategy, and then whether the product form will change, in terms of talent gradient, which people need to be cut off and which people are added.
So you can observe where your company is now? Is the manager still training, or is your company constantly holding strategy conferences, and then giving various publicity and implementation within the company, what changes have been made in the company’s internal strategy? If this is the step, then it is exciting to the second step, including the action adjustment, layoffs or recruitment.
The third step is to start introducing common product tools.
For example, how does the ideal work? At the earliest time, the LLaMA large model introduced was an open source large model. Then he used his own technology to develop ideal applications at the upper level based on the LLaMA open source large model.
Further on, you can customize and develop your own products. For example, the ability optimization of ideal marketing and customer service tools, including AI human-computer interaction solutions for intelligent cockpits, can be achieved through large models, and this part has been being done. For example, when a new technology emerges and DeepSeek appears, then Ideal can also access DeepSeek as soon as possible, which is also the third step.
Therefore, the action of the third step will depend on the strategy and organizational construction. Then those who have the strength and funds can start to pick up the tools of these large models. Either directly connect to general tools, or do secondary tool development based on the underlying model, and apply it in various fields of business.
The fourth step is business process reshaping.
There is a fundamental difference between the third step and the fourth step. The business attributes of the third step have not changed, from business strategy to business process, to all business structures, including the underlying support product form, there is no change, but some new things have been developed on the original basis. The biggest change in the fourth step is to rely on the decision-making ability of large models to reconstruct business processes.
These four levels can also be summarized as the first level is called the cognitive layer, the second level is called the strategic layer, the third level is the tool layer, and the fourth level is the core layer.
This is a common problem in the implementation of enterprise AI, in this process, we must make it clear that the implementation of enterprise AI must be top-down, not initiated by our employees, but must be responsible from the top CEO and willing to do transformation. Then he will drive the team to transform and set goals at all levels from the cognitive level, strategic level, organizational level, tool level, and core level.
If you’re an AI product manager today, or you want to transition to an AI product manager. Then you need to understand that AI products have different levels and attributes in the process of integrating into the business.
Join the starting point classroom membership and learn massive AI content for free
For Starting Point Classroom members, we are always committed to helping product owners and operators improve themselves. In this AI era, we have kept up with the pace of industry development and updated and launched 100+ AI-related skills and knowledge points to help everyone stay at the forefront of industry development and expand their AI-related knowledge and skills.
Join the starting point classroom membership to learn cutting-edge AI content for free! It is the 11th anniversary of the starting point classroom, and you can get 6 months for buying a 1-year membership!
Click here to learn more about the starting point classroom membership:https://996.pm/7mvqk
