A 10-step AI action plan that the CEO can implement immediately.

The wave of artificial intelligence is profoundly changing the business landscape. Many enterprises have shouted the slogan of "embracing AI" and actively introduced various AI systems. However, in many companies, these AI systems, which require huge investments, are like ornaments and fail to truly integrate into the enterprise operations and workflows. Not to mention bringing about efficiency improvements and transformation and upgrading. This phenomenon of "much ado about nothing" is a pain point faced by many enterprises at present.

As the helmsmen of enterprises, CEOs cannot and should not stay out of this transformation. To make AI truly effective, CEOs must get personally involved, set an example, and personally experience the changes and challenges brought by AI. Only in this way can they deeply understand the potential and limitations of AI, thereby promoting enterprise transformation and improving operational efficiency.

Based on the practical experience of Ren Xin, a senior AI industry investor and serial entrepreneur, the author of this article, Hundun Jun, shares "Embracing AI: Ten Small Things CEOs Must Do Themselves" with you. As a partner of the US dollar fund Sky9 Capital, Mr. Ren Xin invests in global AI innovation products. He is also the head of the Sky9 Build incubator, deeply incubating AI-native applications. Mr. Ren Xin provides 10 specific actions for enterprise CEOs to help them build a "physical perception" of AI. Starting from small things, they can lead the organization to seize the opportunity in the AI era and achieve collaborative evolution between humans and machines.

Download 10 AI Applications

Action is the starting point of cognition. CEOs should put aside their obsession with "learning" and just "get started". The first step is to search for "AI" in the mobile app store and download all the AI applications on the first four or five screens. Only by using them personally can they have a preliminary perception of AI.

It is recommended to place commonly used AI applications (such as ChatGPT, Tongyi Tingwu, DeepSeek, etc.) in the shortcut bar at the bottom of the mobile phone desktop or pin them on WeChat so that you can see them every day and be constantly reminded to use AI to develop the habit of using it. If you think randomly downloading is too "rough", you can go to AI tool navigation websites, such as AI-BOT.CN or AIGC.CN, which list various AI applications by category. Most of them offer free trials.

If you say you don't look at your phone screen very often, what software do you use most? WeChat. Search for a friend named "Yuanbao" and add it. It is Tencent's DeepSeek. Pin the conversation with it so that you can see an AI there every day. Whatever you say to it, it can immediately give you a reply. This is absolutely doable without any difficulty, and it must be useful as long as you do it.

Buy 1 Overseas AI Account

The second thing is to buy an overseas AI account. It is strongly recommended that you do so. Specifically, what should you buy? ChatGPT, Claude, and Gemini.

As for DeepSeek, you can think of it as a student from Tsinghua University. It's good. But if you recruit someone from Harvard, MIT, or Stanford and compare them, you may find that in some aspects, Tsinghua's is better, but in other aspects, they are indeed stronger. Currently, we are still catching up as a whole. So when using advanced technology as a productive force, it is recommended to use the best and the most expensive.

There is another benefit. Once you spend money, you will be more active in using it. For example, I bought a $200 ChatGPT account. If I don't use it much one day, I feel like I've lost out. It's the same principle as a gym membership. Within your economic tolerance, it is recommended to buy the more expensive one so that you will be more willing to use it.

You should experience at least one of these three because the actual feelings are indeed different. Personally, I think ChatGPT has good logic and a good memory, like a warm-hearted guy; Claude is strong in coding, has good writing skills, and has a strong humanistic sentiment; Gemini has strong instruction-following ability and a large text processing capacity. They each have their own strengths, and their intelligence is at the international first-class level, but they also have their own advantages.

I strongly recommend that you buy one of these three in addition to DeepSeek. You can enjoy world-class intellectual support for only $20 a month. Think carefully. How much does it cost to invite someone to a meal? Hiring a world-class AI must be more useful than inviting someone to twelve meals because you can harass it every day, and it will do whatever you ask it to do. This money must be spent. It is a talent cost, and the invoice should be made out for human resources expenses, not subscription fees.

Use DeepResearch (Deep Research) 3 Times

The third thing is to use Deep Research three times. Currently, both Chinese and foreign AIs basically have this function. If you haven't used it, you must experience it at least three times.

What is Deep Research? You can understand it as an AI that will search for information online according to your task and then give you a report after the search. It is different from ordinary searches. During the research process, it will reflect on whether it has achieved its goal and adjust the research direction in a timely manner. Usually, DeepResearch takes a relatively long time, maybe ten or twenty minutes, but it can provide a large amount of high-quality content that has been deeply thought about, far superior to the results that come out in one minute.

Why do I recommend that you try this? Because Deep Research is the first mature Agent. It helps you collect text and then integrates it to answer your questions. It has invested a lot of effort, so it is more mature. One of the core tasks of a CEO is to gather intelligence and conduct analysis and research. The emergence of Deep Research is equivalent to outsourcing this task to AI, solving half of the CEO's problems.

How to use it specifically? For example, if you have a technology at hand and want to explore its application fields. Recently, I have been paying attention to thermal imaging technology, including infrared imaging and heat conduction. I once asked an AI: "We are researching how to apply thermal imaging and sound imaging technologies to the US consumer market. What are the reference objects? Which products are worth paying attention to? Which niche markets have great potential? Where are the opportunities?" The AI will ask if you need to supplement information to optimize the research. I usually answer "All are concerned", and then it starts the research.

Finally, it will generate a detailed report containing a large number of links, introducing relevant products and their historical performance for your reference. I usually ask the AI to provide factual content related to my product, such as product history, and attach links so that I can conduct my own research. I now have a special monitor set up in a vertical screen display to facilitate continuous operation and viewing of such reports.

If you find that you don't have any questions worth in-depth research, this is actually a bigger problem. It means that your daily work is limited to execution and lacks strategic thinking. A CEO who doesn't have in-depth thinking about problems is a bigger hidden danger to the company.

Create 1 AI Agent

The definition of an Agent is broad. The most basic intelligent agent is the repeated use of prompts. Advanced Agents can plan independently, reflect, use tools, and evolve on their own. As a popular science introduction, let's start with the simplest: save and reuse effective prompts to form a simple and efficient standard operating procedure (SOP). Most people haven't started using this method yet, but it can transform one-time work into reusable assets.

For example, if you usually have the habit of taking notes and have specific requirements for them, hoping that they can provide more inspiration, clarify the application boundaries, scenarios, relevant knowledge, and applicable and inapplicable situations of the principles, you can write a prompt to achieve this goal. At this time, the simplest way is to use Doubao because it is more widely used and easier to access than ChatGPT.

Open the "Create AI Agent" button in the upper left corner of Doubao, name it, and define its function: every time you input a note, it will provide relevant knowledge, more inspiration, application scenarios, and inapplicable situations. Click "Create", and this agent (i.e., the saved prompt) can be used next time. After that, every time you take notes, you just need to open the agent dialog box and input your notes.

In addition to the note-taking scenario, it can also be used in the work scenario. For example, if you need to communicate with foreigners, not in a formal occasion but on WeChat. If you directly use GPT for translation, the output is often too formal. This is a daily trivial matter, but it's annoying to write prompts by yourself every time. You can create a colloquial agent to help you convert Chinese into colloquial English and learn English by the way. Every time you need to send a long text, I just give it to the agent. For short sentences, it doesn't matter. Even if there are mistakes, foreigners can understand them.

All in all, when you identify a repetitive task in your work that has clear input, processing, and output, you can create an AI Agent to handle it.

Creating an AI Agent requires effort and overcoming the "new manager syndrome" - always wanting to do it yourself and thinking "forget it, it's better if I do it myself". However, if you are willing to spend three hours "teaching" it, you can save three hours every week.

Learn 1 Piece of Knowledge with AI

Students who love learning usually regard learning as a single thing. In fact, there are two key links before and after the learning process: filtering and in-depth exploration.

First is filtering: With limited time, how to choose valuable content? Have you ever felt that you wasted time after watching a video or listening to a course? AI can help you filter information.

For example, I prefer podcasts because most of the cutting-edge AI information is presented in the form of podcasts, especially in Silicon Valley where there are a large number of high-quality podcasts. If I want to know about a certain podcast, a plugin will display its approximate timeline on the right side of the screen, and I can quickly scan it within a minute to judge its value.

The focus of a podcast may not be the same as what I'm concerned about, which varies from person to person. For example, you may only care about the application of AI in the clothing or industrial fields. Even if a podcast is excellent overall, but only a small part mentions it, what you need is not the whole but an in-depth elaboration of that specific part.

So you can create an intelligent agent, set it as an AI entrepreneur who is also interested in personal improvement and productivity. It will highlight and elaborate on the key content in the video and provide timestamps so that I can choose to listen selectively. Every time I just paste the link to the agent, and it will automatically generate a long text elaborating on the counterintuitive insights, systematic methodologies, and cases in it. I quickly scan it, and if I'm interested, I directly jump to the corresponding timestamp to listen. This makes my efficiency in watching podcasts extremely high. On average, it only takes 2 to 5 minutes for each podcast, and I only focus on the content I'm interested in. In this way, I can quickly screen and filter a large amount of information.

In addition to filtering, learning also includes in-depth exploration. When you are interested in a certain point, you should dig deeper. Even if it is only mentioned briefly, you should continue to ask questions and let AI elaborate in detail. AI will analyze the overall discussion of this view, relevant theoretical schools and viewpoints, and how to apply it to your specific situation. This is in-depth exploration.

So it is recommended that you use AI to learn something. You can find the information sources by yourself, and then let AI filter out two or three pieces of content that you may be interested in. If there are inspiring points during the learning process, you can let AI dig deeper and ask how to apply them to yourself. Leave all the tasks of learning, filtering, in-depth exploration, and application to AI. We only need to act as navigators in the middle. The above operations do not require any coding skills, only clicking buttons and paying fees. This is completely feasible.

Record for 1 Day and Then Let AI Analyze

The next task is also feasible, but relatively less common. Recently, I found a very cost-effective application: record for one day and then let AI analyze it. This function is very practical and free.

You can use Tongyi Recording because it is convenient to operate on the mobile phone. Open the real-time recording function, and it will record synchronously and generate a transcript