Home About us

What is Agentic AI? What is the difference between AI Agent and AI Agent?

Wang Jiwei 2024/06/28 13:27

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

Text/Wang Jiwei

Just when China was still talking about AI Agent, the wind direction of foreign technology circles changed. Instead of talking about how and how AI Agents are, they turn their heads and start talking about Agentic AI.

While the concept of Agentic AI can be traced back to the 90s when IBM's Deep Blue chess-playing system appeared, it was the real application of large language models that brought it back into the public eye. In particular, the specific application of AI Agent and Autonomous Agent has made Autonomous AI hotly discussed again, and the workflow including AI Agent has directly made Agentic AI a hottest topic in the AI field.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

This process and changes are still thanks to OpenAI. In June 2023, Lilian Weng, head of applied research at OpenAI, wrote an article titled "LLM Powered Autonomous Agents", which defines the mainstream technical framework of AI agents that many developers are currently using.

Article address: https://lilianweng.github.io/posts/2023-06-23-agent

In December 2023, OpenAI published a white paper titled "Practices for Governing Agentic AI Systems", introducing the differences between Agenticness, Agentic AI Systems, and Agents, and officially defining Agentic AI Systems. Since then, Agentic AI has officially entered the field of vision of technologists.

White paper address: https://openai.com/index/practices-for-governing-agentic-ai-systems/

What really made Agentic AI out of the circle was the Agentic Workflow proposed by Professor Andrew Ng, a leader in the field of artificial intelligence and founder and CEO of DeepLearning.AI and Landing AI. On March 26, 2024, Professor Andrew Ng delivered a speech entitled "Agentic Reasoning" at Sequoia Capital's AI Ascent Summit, sharing the four mainstream design patterns of AI Agent.

At the recent Snowflake Summit 2024 Developer Day, he gave a speech titled "How Al Agentic workflows could drive more Al progress than even the next generation of foundation models", further unveiling the mystery of Agentic AI. It may be a promising AI development direction than the next generation of basic models.

Video address: https://www.youtube.com/watch?v=q1XFm21I-VQ

The two speeches made Agentic AI truly a topic that technical people are keen to discuss.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

AI Agent and Agentic AI look so similar that it looks like the two words "AI Agent" are swapped places, and the average person can't even see the difference. However, if you taste it carefully, you can appreciate the slight difference between Agent and Agentic just from the difference between nouns and adjectives in the nature of words. As a concept or technical term of artificial intelligence, the difference in meaning between the two is still very obvious.

So, what exactly is Agentic AI? What are its innovations and features? What is the difference between AI Agent and AI Agent? What is Agentic Workflow? What are the characteristics? What are its four main design patterns? In this article, from Agentic AI to Agentic Workflow, Wang Jiwei's channel will explain it clearly for you at one time.

(Note: The relevant papers, white papers, and other resources mentioned in this article have been packaged, and the background sends a message agentic to obtain them.) )

What is Agentic AI?

In the white paper "Practices for Governing Agentic AI Systems", OpenAI argues that Agentic AI systems are characterized by the ability to take actions that consistently contribute to the achievement of goals over a long period of time, without having to specify their actions in advance.

The white paper defines the degree of agenticness of a system as the degree to which a system adapts to achieve complex goals in a complex environment with limited direct supervision, and subdivides this intelligence into four components: target complexity, environmental complexity, adaptability, and independent execution.

artificiality defines Agentic AI Systems as systems capable of perceiving, reasoning, and acting in varying complexities to extend the human mind beyond our current experience. This definition places more emphasis on the three abilities of perception, reasoning and action.

Based on the above two definitions of Agentic AI Systems, combined with the various views of the industry on Agentic AI, it is not difficult to summarize the profound meaning of the concept of Agentic AI.

Agentic AI, also known as autonomous AI, refers to a system designed to perform tasks independently and proactively through natural language input by understanding targets, navigating complex environments, and performing tasks with minimal human intervention. Often designed to be more autonomous and adaptable, not only processing data, but also making decisions, learning from interactions, and taking proactive steps to achieve complex goals.

Agentic AI can set goals, learn from interactions, and make decisions autonomously to transform business operations and customer interactions. Functions are very similar to those of human employees, who can grasp the nuances of their environment, set and pursue goals, reason through tasks, and adjust their actions to changing conditions.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

For example, while traditional AI may assist in creating customer support scripts or even generating personalized responses based on customer input, Agentic AI takes it a step further. It can handle customer inquiries autonomously, resolve issues from start to finish, and even follow up with customers based on their responses. Agentic AI can mimic the reasoning, execution, and corrective heading mechanisms that humans typically use to achieve goals, embodying a more refined approach to technical operation and management.

In fact, the idea behind Agentic AI is to empower machines, which means they can set goals, plan, and take action to reach those goals. With the ability to anticipate needs, recommend actions, and make decisions that align with set goals, Agentic AI functions more like a partner than just a tool. As such, it represents a fundamental shift in artificial intelligence to autonomously understand and manage complex workflows with minimal human intervention.

Innovations and features of Agentic AI

Here, in order to reflect the innovation and characteristics of Agentic AI, it is necessary to compare it with traditional AI.

Traditional AI, also known as narrow AI, operates primarily on specific algorithms and set rules. These systems are designed to perform well-defined tasks, such as sorting data, recognizing faces in photos, translating languages, performing predefined processes, or answering frequently asked questions based on databases. The scope of traditional AI is limited to its programming, lacking the ability to deviate from its given instructions or learn new experiences independently.

Traditional AI excels at narrow tasks that need to be run under clear instructions. It thrives in a structured environment with clear rules and operates effectively in scenarios where processes are tightly divided, but is limited by its limited scope, reliance on human guidance, and difficulty adapting to unforeseen changes.

In addition, traditional AI is primarily designed to automate specific repetitive tasks, increasing speed and efficiency to a limited extent, but they fall short in handling complex workflows that require holistic understanding and strategic judgment.

Agentic AI provides a more dynamic and flexible approach by leveraging advancements such as large language models (LLMs), scalable computing power, and huge datasets. It combines reinforcement learning (RL) and decision theory to learn from interactions and optimize over time. Not only to be responsive to situations, but also to actively participate in the decision-making process.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

Arguably, Agentic AI is built for autonomy, designed to navigate complex real-world scenarios and be able to adjust its strategy as the situation evolves. This marks a profound shift from AI as a tool or specialized system that requires human input to as a collaborative partner capable of acting independently and interacting with the real world.

Agentic AI functions more like a human employee, mastering the complex context and instructions provided by natural language, starting to set goals, reasoning through subtasks, and adjusting decisions and actions based on changing conditions.

Therefore, the key innovations of Agentic AI are mainly reflected in the following points:

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The main features of Agentic AI can also be summarized as follows:

Now, groundbreaking frameworks like CrewAI, Langraph, and Autogen are paving the way for the development of Agentic AI. Developers can design and deploy teams of AI Agents on these platforms, each with unique skills, knowledge bases, and communication interfaces. Through coordinated collaboration, these agent teams can autonomously navigate and execute complex workflows, adapting to dynamic conditions and changing needs.

Extended reading: AI agents build an intelligent future, and a large inventory of 80+ AI agent construction platforms around the world

These advancements enable Agentic AI to go beyond just following instructions to set independent goals, strategize, and adapt, providing a dynamic approach to achieving complex goals.

Difference between Agentic AI and AI Agent

Although the phrase AI Agent is very similar to Agentic AI, which is basically the difference between "AI in the back" and "AI in the front", the two concepts are still very different.

Regarding the words Agent and Agentic, Professor Ng mentioned in the article that instead of choosing whether a system is an Agent in a binary way, it is more useful to think of the system as having different degrees of Agent characteristics. Unlike the noun "Agent", the adjective "Agentic" allows us to think about such systems and bring them all into this evolving field.

Rather than having to choose whether or not something is an agent in a binary way, I thought, it would be more useful to think of systems as being agent-like to different degrees. Unlike the noun “agent,” the adjective “agentic” allows us to contemplate such systems and include all of them in this growing movement.

Original link:

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

In other words, the noun "Agent" is only used to indicate whether a product or project is an AI Agent and whether it has intelligent features, while the adjective "Agentic" means how strong the Agentic features of an AI product or project are, and whether it can reflect greater initiative, autonomy, and adaptability. The former is still talking about the relevant features of the agent product or project, while the latter is talking about how intelligent the product is, and obviously the latter is more meaningful.

From AI Agent to Agentic AI, even though the current discussion is still about AI Agent-related technologies, products, or solutions, the foothold has been completely different, which is a major shift in cognition. If AI Agent is still a product mindset, Agentic AI has risen to strategic thinking. Agentic AI further represents a holistic collection of AI technologies, products, solutions, ecosystems, and even strategies, and will inevitably be included in their strategy reports by more organizations like GenAI.

By definition and conceptually, an AI agent is an intelligent entity that can perceive the environment, make decisions, and perform actions. They are often based on machine learning and artificial intelligence technologies and are autonomous and adaptive, with the ability to learn and improve autonomously in a specific task or domain. Its core function can be summed up in a three-step cycle: sense, plan, and act.

Agentic AI are AI systems with a higher degree of autonomy that are able to actively think, plan, and execute tasks rather than relying solely on pre-set instructions. It emphasizes that the system can have different degrees of "agency" (agentic characteristics) and is not limited to passively executing instructions.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The main differences between the two are as follows:

So much we have said about the difference between the two. But when you explore Agentic AI, you'll find that more content is still related to AI Agent, or "Agent + Workflow". Professor Ng also mentioned that the key to the realization of Agentic AI lies in the "Agentic Workflow", which gradually optimizes the results through cyclic iteration, which is closer to the human problem-solving mindset.

Therefore, if Agentic AI focuses on the strategic level and points to the organization's technology development goals, AI Agent is the main way to achieve this goal at the tactical level.

Since the concept of Agentic Workflow is involved, let's talk about it here.

What is Agentic Workflow?

Since Professor Andrew Ng shared four design methods for Agentic Workflow, Agentic Workflow has become more and more popular. While this concept is not entirely new, Agentic Workflow is becoming hot with the growing use of large language models and AI agents.

After summarizing the opinions and opinions of various parties, the following explanations can be made about Agentic Workflow.

Agentic Workflow can be translated as agent workflowAgent workflowActive workflow, at its core, is an agent system in which multiple AI agents work together to complete tasks by leveraging natural language processing (NLP) and large language models (LLMs). Capable of autonomously sensing, reasoning, and acting in pursuit of specific goals, these agents form powerful collective intelligence that can break down silos, integrate disparate data sources, and provide seamless end-to-end automation.

As a complex system and iterative process, Agentic Workflow aims to improve the efficiency and effectiveness of business processes. It uses AI agents to seamlessly integrate with business settings, and the AI agents deployed in the Agentic Workflow are able to collaborate and perform complex tasks with high precision.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

Agentic workflow architecture diagram Source: codiste

From a development perspective, Agentic Workflow refers to an iterative, interactive approach to AI development that uses large language models and AI agents to perform tasks, where the AI agents are able to engage in a more dynamic and self-reflective process, which is a more iterative and multi-step approach.

From an automation perspective, Agentic Workflow represents a significant shift from traditional automation approaches. Traditional automation methods often rely on strict predefined scripts or human-machine interaction processes, and by leveraging the ability of multiple specialized AI agents to work together, Agentic systems can dynamically navigate and adapt to the complexity of enterprise workflows, thus promising to take productivity and innovation to new levels across industries.

In simple terms, Agentic Workflow is a complex iterative and multi-step process for interacting and instructing large language models to accomplish complex tasks more accurately. In this process, a single task is divided into smaller, more manageable tasks and leaves room for improvement throughout the task completion.

In addition, Agentic Workflow involves deploying multiple AI Agents to perform specific roles and tasks. These agents are equipped with specific personalities and attributes that allow them to collaborate and perform defined tasks with high precision.

Another key highlight of Agentic Workflow is the use of advanced prompt engineering techniques and frameworks. The process includes techniques such as chain of thought, planning, and self-reflection that enable the AI Agent to:

For example, if you write an article about Agentic AI directly using LLMs. In the traditional approach, a prompt would be entered instructing the LLM to write an article on this topic. It's like asking someone to write an essay from start to finish without reviewing research sources, checking the outline, and improving the tone and quality of the content.

This traditional zero-shot method uses LLMs, leaving no room for iteration, feedback, and improvement during the writing of the article, greatly reducing the accuracy and quality of the output.

However, Agentic Workflow does not need to give a prompt to write an article, only needs to put forward the goal requirements, it can decompose the task into smaller tasks, and generally has the following task decomposition steps:

In the Agentic Workflow working mode, the LLM is instructed to complete larger tasks in a step-by-step process, with the output of each step acting as input to the next task.

This means that Agentic Workflow, an iterative and collaborative model, translates interactions with LLMs into a series of manageable, improvable steps, allowing for continuous improvement and adaptation throughout the task process.

The main features and three pillars of Agentic Workflow

Based on the above analysis and examples, we can summarize the main features of Agentic Workflow as follows:

Agentic Workflow offers several advantages over traditional workflow automation. They can handle more complex, multi-step processes that require context-aware decision-making and can adapt to new situations without extensive reprogramming. In addition, the use of natural language processing allows for more intuitive interaction between humans and systems, reducing the need for technical expertise.

In Agentic Workflow, the AI Agent is an autonomous, driven, dynamic problem solver for complex and evolving tasks that increase productivity.

AI Agents, Prompt Engineering Techniques, and Generative AI Networks (GAIN) are the three pillars of Agentic Workflow. Their role in Agentic Workflow is briefly described as follows:

AI Agents: At the heart of Agentic Workflow are AI Agents, which are essentially complex instances of large language models (LLMs).

Prompt Engineering Techniques & Frameworks: A key aspect of Agentic Workflow is the use of advanced Prompt Engineering techniques and frameworks.

Generative AI Networks (GAINs): Agentic Workflow has been significantly enhanced by the deployment of Generative AI Networks (GAINs), which embody the principles of multi-agent collaboration.

For the details of the three pillars, you can refer to the mind map below.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

▲ Click to enlarge the image

In addition to the three pillars, the core components of Agentic Workflow include AI augmentation, ethical considerations, AI interaction, and adaptive learning, which are explained in the figure below.

In addition, for a survey on LLM-based agents: common workflows and reusable LLM-profiled components, you can refer to the paper: A Survey on LLM-Based Agents: Common Workflows and Reusable LLM-Profiled Components.


The shift from traditional processes to Agentic Workflow is a sign of a shift in the business processes that will enable us to achieve better outcomes through AI. Experiments have shown that even less advanced LLMs can produce significant results when involved in these complex, multi-layered workflows. In this regard, you will have a deeper impression on the following introduction to the four design patterns of Agentic Workflow.

In addition, Agentic Workflow also allows domestic large language models and various open source large language models to have more use, which is still very important in the current international environment.

Of course, we should also recognize that these enhanced workflows require a lot of patience from the user under the current state of technology. Because of the iterative, collaborative process inherent in Agentic Workflow, it is also more time-consuming, often taking minutes or even hours to complete a task. Excessive task execution time is also one of the main problems encountered by Agentic Workflow, and it is also an important factor that is complained about the lack of experience.

However, compared to the depth of analysis, creativity and problem-solving ability that it can provide far beyond traditional methods, people are still willing to try it in many application scenarios, which also indicates a huge market potential.

The four main design patterns of Agentic Workflow

Before introducing Agentic Workflow, consider the question, why do we need AI Agent/Agentic Workflow?

At present, most people still ask questions directly in simple sentences (partly because they can't write structured prompts) when using large language models such as ChatGPT, Wenxin Yiyan, and Kimi, such as: Help me write an article about Agentic AI. This type of questioning is called a zero-shot prompt in the technical field.

Zero-shot prompting refers to the ability of an LLM model to perform a task by relying only on prompts and extensive language knowledge obtained from pre-training without special training for specific tasks, which can be a good test of the ability of large language models. This approach is flexible, versatile, and eliminates the need to prepare specialized training data for each specific task. However, due to the lack of training for specific tasks, the quality of its generation cannot be guaranteed.

Specific to the user's interaction with the large language model, if the user asks ChatGPT to write a paper about XX, ChatGPT will give a one-time reply. In this process, it will only perform the task of "generating".

This process is different from the process of completing work tasks in the real world. For example, when writing a thesis, a first draft is usually drafted, then evaluated, analyzed, revised, and iterated for a second or third version until you are satisfied. This is the case when we deal with tasks such as work and study, and we will break it down into step-by-step processes and operate according to the process to ensure the quality of the completion of the work.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

▲ The difference between zero-sample prompts and Agentic Workflow Source: Professor Andrew Ng's sharing at the Sequoia Summit

That said, in order for a large language model to do the job we give it better, it's best to use human-like process steps as well.

The AI Agent is here to do this work, it can understand the intent of the zero-shot prompt words entered by the user through natural language, and decompose the user's given demand target task planning into multiple process steps, transform the simple prompt words into more refined institutional prompt words, and can call various tool plug-ins such as networking and code to further execute and complete various subtasks decomposed.

Obviously, this way of working is infinitely closer to that of humans.

The task execution effect of the zero-shot prompting mode and the AI agent mode is very obvious. Professor Ng's team used GPT-3.5 and GPT-4 in "Zero-shot prompting" mode and AI Agent mode respectively, and came to several conclusions:

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

1. In the zero-shot mode, the model only relies on its pre-trained knowledge to perform the task without a specific task example. In this case, the competition is the general basic ability of the model, so it is expected that GPT-4 will perform well.

2. The performance of the AI Agent with the addition of Agentic Workflow has been significantly improved in task execution, regardless of whether the base model is GPT-3.5 or GPT-4.

3. Even if the base model is GPT-3.5, the performance of GPT-4 in Zero-shot mode is surpassed after it is designed as an AI agent by adding Agentic Workflow.

Professor Ng also summarized and introduced four common design patterns, namely Reflection, Tool Use, Planning and Multi-agent Collaboration.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The Reflection design pattern is a way for AI models to improve their ability to perform tasks through self-reflection and iterative improvement. In this model, the model not only generates the initial solution, but also continuously optimizes its output through multiple feedback and modifications.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The Tool Use design pattern is a way for an AI model to augment a task by calling an external tool or library. In this model, the model does not rely solely on its own knowledge and capabilities, but instead utilizes a variety of external resources to complete tasks, thereby improving efficiency and accuracy.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The Planning design pattern is a way to improve efficiency and accuracy by planning and organizing task steps in advance. In this model, the model breaks down a complex task into multiple steps and executes each step in turn to achieve the desired goal.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

The Multiagent Collaboration design pattern is a method to improve the efficiency and accuracy of task execution through cooperation between multiple agents. In this model, multiple agents share tasks and work together to accomplish complex tasks by communicating and collaborating with each other.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

PS: The above pictures marked with Andrew Ng are all from Professor Andrew Ng's Sequoia Summit sharing. Send a message to agentic in the background of the official account to get the PDF file of Andrew Ng's Sequoia American AI Summit talk about Agentic Workflow and 4 mainstream design patterns.

AI Agent/Agentic Workflow can help users better interact with large language models and complete various tasks. This will greatly expand the use scenarios of AI and effectively improve the quality of task completion, so it is crucial for the implementation of AI applications.

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

In this sharing, Professor Ng referred to the process in which the Agent participates as the Agentic Workflow rather than the AI Agent workflow, which obviously focuses more on the workflow in which the AI Agent participates rather than the AI Agent itself. From this point, you can also see the simple difference between AI Agent and Agentic AI.

For the business process automation in which AI agents participate, technology vendors such as RPA hyperautomation, ERP, CRM, and BI have already introduced the AI agent architecture under the existing technology ecosystem, and they have demonstrated stronger task execution capabilities. For this topic, Wang Jiwei's channel will communicate with you in another article.

The concept of Agentic AI has been clearly explained, and to achieve this goal and make it prosperous, it depends on the construction methods of various AI agents and various Agentic Workflow solutions that integrate AI agents.

In terms of building Agentic Workflow, Professor Ng has given four mainstream design patterns, which have inspired developers and enterprises.

At present, many AI agent build platforms already support these four design patterns, and enterprises and individuals can build agentic workflows that meet their needs on these platforms. There are also many open source projects that are further optimizing the build process of Agentic Workflow, which is a great benefit to the private deployment of the majority of organizations.

In addition, many technology vendors have further built Agentic Workflow on top of the introduction of AI Agent, which can make it easier and faster for users to apply various intelligent workflows.

On the construction of Agentic Workflow, Wang Jiwei will also write a few articles to share relevant experiences with you. What do you want to communicate with, please leave a message.

Postscript: Keeping Agentic AI's Business Finger on the Finger Foot

At present, there is a general trend of AI applications, and almost all applications are developing and migrating in the direction of AI agents and RAG. This means that if all AI applications move to the AI agent mode, future workflows will become agentic workflows.

From the perspective of Professor Andrew Ng's zero-shot prompting and Agentic Workflow, any large language model with the Agentic model can be much ahead of the large model itself, which means that the business process efficiency of organizations and enterprises will be doubled in the future.

In Wang Jiwei's view, even if the large language model behind ChatGPT is iterated to a later version after GPT-4, Agentic AI will be a more efficient way to use large language models. Otherwise, OpenAI would not have published the white paper mentioned at the beginning of this article to elaborate on the correlation and difference between large language models, AI Agent, and Agentic AI Systerm.

It has been verified that through various systems and integration of large language models, or the expansion of tools and platforms on the basis of large language models, the application efficiency and experience of large language models will be higher.

Further reading: More organizations are accessing generative AI such as ChatGPT, and generative automation may become the new standard for enterprise operations

Jiwei Wang, What is Agentic AI? What is the difference between AI Agent and AI Agent?

From the perspective of the development history of AI technology, the concept of intelligent twins precedes AI, and AI has always been the technical implementation path of Agent, and large language models are no exception. Therefore, after the popularity of machine learning and deep learning, everyone is paying attention to AlphaGo and AI face swapping technology. Now that large language models are popular, people are paying more attention to AI Agent, Agentic Workflow, and the next Agentic AI.

Further reading: A brief history of the development of AI Agent, from the enlightenment of philosophical ideas to the implementation of artificial intelligence entities

Of course, Agentic AI is not a mystical thing, but it is just a combination of various applications and ecosystems, including AI Agent and Agentic Workflow. However, it is undeniable that it is bound to become a mantra in major organizations and business fields for a long time to come, and will also become an important embodiment of AI at this stage.

At present, everyone is talking about intelligent bodies, which makes AI Agents more and more aesthetically tired, and more foreign countries are talking about Agentic AI. Even Professor Ng mentions in the article that when you see an article that talks about the "Agentic" workflow, you are more likely to read it.

Everyone's eyes are starting to attract Agentic AI, will they pay enough attention to AI Agent? Does this mean that AI agents are already on the wane? For these questions, don't forget what we discussed earlier: AI Agent is the implementation of Agentic AI, and Agentic Workflow is the key to Agentic AI.

Therefore, by paying attention to the strategic development trend of Agentic AI from the level of "Tao", and the various frameworks, technologies and solutions that AI Agent continues to innovate from the technical level, you will be able to grasp the business pulse of the entire Agentic AI and even the AGI era.

End of full text

[Wang Jiwei channel, pay attention to AIGC and IoT, focus on digital transformation, business process automation and RPA, welcome to pay attention and communicate. 】

This article is from Xinzhi self-media and does not represent the views and positions of Business Xinzhi.If there is any suspicion of infringement, please contact the administrator of the Business News Platform.Contact: system@shangyexinzhi.com