The arrival of AI technology not only opens up people’s imagination, but also brings more possibilities to individuals in specific positions. So overall, what specific problems related to marketing and operations can AI solve? How does AI solve these specific problems? Let’s take a look at the author’s answer.
The title is the question I was asked most when I was recruiting students for the first batch of action camps.
Facing the fierce AI technology, as a member of private domain operations, marketing, product operations and other partial marketing positions, I believe you also have such questions.
I am afraid of missing the trend, and I don’t know what specific value AI can provide.
After completing the delivery of the first phase of the action camp, I have a relatively clear answer:
- For those standardized, mechanical, and repetitive “production-type” work (such as marketing copywriting with a fixed form but the product needs to be adjusted, such as morning newspapers with a fixed time but need to change the content body in real time, such as a fixed purpose but need to change the style to keep it fresh interactive topics), the value of AI lies in directly liberating human power.
- For those “brainstorming” jobs that require a lot of creative collisions and where the computing power of the human brain is not enough to exhaust all creative ideas (such as creative ways of marketing activities, the expandable angles of topic interpretation, and the cross-analytic dimensions of data analysis), AI’s The value lies in the continuous replenishment of creative ammunition.
- For those “strategic” work that can produce better output once there is practical theoretical support (for example, using mental accounting theory to write more impressive copywriting, such as using the Six Thinking Hats model to produce more comprehensive Decision-making, such as using the diode headline method to mass-produce high-quality explosive articles more efficiently), the value of AI lies in maintaining both quantity and quality.
1. How AI solves specific problems
The AI we use today , whether domestic or overseas large models, are essentially GPT, which is Generative Pre -Trained Transformer (generative pre-training Transformer model).
All capabilities are based on the knowledge learned by AI to “generate answers”. The difference in usage lies in what you expect AI to generate, that is, what requirements you put forward.
Writing mechanical copywriting, providing creative ideas or theoretical support are essentially generating content, but the generated content is at different stages of your work chain:
The first is ” you know how to do it, teach it to AI and let it do it “; the second is ” you know the direction but can’t think of the details, let AI help you exhaustively “; the third is ” you know The requirements for the final result do not know the production process, so AI can provide theoretical support .”
Among the first batch of trainees, there are front-line operational executors who master the specific methodology of their work, so AI has become an amplifier of value.
Once the AI training is completed for this kind of work, the biggest gain is: saving (days) and saving (days) a lot of (energy) (touch) time (fish).
The other is decision-making or strategy work, which essentially tests the decision-making model mastered by the decision-maker or the way he thinks about problems.
In the first period, there was a very powerful BOSS-level student. When the course was only one week old, he had already started using AI to screen and evaluate applicant resumes.
From the last sentence of the conversation, you can see that the boss’s rank is very high. After achieving the stability and controllability of AI production capacity, the next step is the professionalism of the industry (these problems were solved in the second week of the course) ).
For smart people, once they start thinking that “AI can be taught”, they will definitely enter a crazy learning “feeding” journey: “feeding” themselves to call more powerful AI.
Thinking models are the underlying values that allow humans to work and collaborate more efficiently. “A person who can multiply” and “a person who can only add” are not at the same level in terms of arithmetic efficiency. “Multiplication” is a model. At work, there are also the “diode headline method” for writing popular headlines, the “pyramid theory” for writing reporting frameworks, and the “mental accounting theory” for making it easier for users to place orders. These are all thinking Model.
Mastering, or even just knowing, these models can double your efficiency when thinking about and solving problems.
So I “forced face-to-face output” in the course, so that everyone was “forced” to learn some “models” that they will definitely use:
2. What can be added to the so-called AI+?
I spend half an hour every day browsing the “AI Drilling Station” circle of the Instant App. Sometimes I can’t finish reading the updated news in this circle in the past day in half an hour.
This circle should be the most active place on the Chinese Internet today for AI application practices, new gameplay sharing, and new technology discussions.
I’ve seen here people who decided not to hire video directors after using AI, people who have been separated from AI and can’t write code or search for information, people who have used AI to improve their thinking efficiency, people who have written a The prompt word allows AI to help review the contract…
If you ask what specifically can be added to “AI+”, then my answer is ” how bold a person is, how productive AI is .”
In the first action camp, because I was too “stuck” in using AI to improve the efficiency of private domain operations, I only left two classes to talk about the application of AI in daily work and personal growth, so that everyone They are all “complaining” and dissatisfied.
So in the second period, I changed all the 7 days of class time originally used for answering questions to “AI + everything”, and added the use of AI to disassemble and write imitation short video live broadcast scripts, write public account Xiaohongshu and other text content, Office 3 Software packages, meeting summaries and other usage demonstrations on daily office work, data analysis, personal growth and even advanced use of the API to do many things.
In its internal open letter after the coaching change on September 12, Alibaba talked about Alibaba’s two major strategic priorities: user first and AI-driven .
If you do business in the Alibaba ecosystem, you will find that Alibaba has launched a large number of AI-driven support tools in the past two months. On the homepage of Alimama’s marketing platform alone, there are more than 5 AI-based copywriting generation, video and text editing and creative planning tools, as well as the Vientiane Lab, which is still in internal testing and can generate main images with one click.
Among the first batch of students, there is a manager of a business line of Alibaba. His KPI for the rest of this year is to train a tool adapted to his own business based on the Tongyi large model to provide support to downstream merchants.
Not only Alibaba, but Tencent, Baidu, iFlytek and Byte, which have self-developed large models, are also using AI to reconstruct their original sub-business line interaction models.
In the case display on the official website of Tencent Hunyuan Large Model, I saw a shopping guide assistant embedded in the sidebar of Qiwei. The AI pre-trained by the knowledge base can help customer service answer customers’ questions more efficiently.
Baixing.com’s Chato intelligent assistant supports uploading your own company’s knowledge base to “train” your own robot. This robot can be embedded in all conversation scenarios such as web pages, Feishu, official accounts, WeChat customer service, and Douyin.
This means that as AI is embedded in more and more scenarios, most of the original business may be reconstructed .
As a marketer, how can you use AI to produce copywriting more efficiently; as a customer service officer, how can you use AI to better respond to customers; as a manager, how can you use AI to make better decisions; as an analyst, How can you use AI to help you better research the industry…
Please note that I use “utilize AI” here, not “use AI”.
3. Use AI or utilize AI
The best way to write AI prompt words today is called “structured”. This prompt word is not a “toothpaste-squeezing” method of asking and answering questions, but telling AI how to execute the entire process of a certain work, achieving “write once and repeat. ” Use “.
Taking writing private domain marketing speech as an example, I told AI all the requirements and methods for writing speech through structured prompt words. Once the prompt words are determined, the next work is to provide the product name, original price and private domain purchase price. AI directly outputs words.
The so-called “using AI” refers to the person who wrote this structured prompt; and “using AI” refers to the person who outputs the product name, original price and in-app purchase price to the AI.
Friends who are surprised, please become nervous immediately, because this is already a realized AI usage, not the future, and does not require special technical prerequisites.
The prompt words shown above are the results of crowdfunding by the students of the first action camp in the last class. We spent two days “after class” and iterated 11 versions and got this one that can produce 80 points in batches, stably and efficiently. Prompt words for marketing copy.
4. Make progress together
After reading this, your mind has actually been opened.
But if you want to join the fast train of applying AI to practice and making specific work more efficient, you can pay attention to the second issue of “AI + Private Domain Efficiency Improvement Action Camp”.
As mentioned before, in the new issue of Action Camp, I have added more demonstrations, practices and thoughts on AI implementation scenarios. My ultimate goal is not just to let everyone use AI, but to open up a “new thinking model in the AI era.”