Calculate the application scenario value of AI Agents

Let me share a value thinking formula about "In what scenarios should we use AI":

What determines how much your Al Agent can sell?

Remember this formula:

Scenario value = number of people labor cost per unit time frequency of use * single efficiency saving time

For example, if there is an internal process of an enterprise:

There are 100 people using this function every day, and each person uses it 10 times a day. Through AI, 5 minutes are saved each time. The enterprise labor cost of this person is 15,000/month. Then the scenario value is: 100*10*5=5000 minutes/day, saving 5000*21.75/60=1812.5 hours in a month. How much are these hours worth?

15000/21.75/8*1812.5=156250

This is equivalent to creating an additional income of 156,250 in theory. We use this to calculate the scenario value and evaluate the improvement of income.

Yes, the value of an AI scenario is measured by the product of four numbers: the first is the number of people, the second is the labor cost per unit time, the third is the frequency of use, and the fourth is the time saved by a single efficiency.

Let me break down each term of this formula and you will understand. You can imagine that because it is multiplication, any increase in a number will multiply the total value of the previous scenario.

The first multiplication term is the number of people, that is, in an enterprise, the more people this application scenario can cover, the greater the value of its scenario. For example, what is the most common scenario for content companies? It must be text work, writing various copywriting. So in these companies, the first AI application to be implemented is "typo check". This is because this scenario involves a large number of people, whether it is content, operation or product, it involves writing a lot of documents. If one AI application scenario can be used by 150 people in the company, and the other can only be used by three or five professionals, it is obvious that the former scenario has greater value. By extension, for the first application, you must not first make a scenario that is only used by the boss or a few executives.

The second multiplication term is the labor cost per unit time. This concept is easy to understand, that is, the average salary of the people covered. For example, if you divide your monthly salary, you can calculate that your hourly salary is 80 yuan, then your labor cost per unit time is 80 yuan. This number can be substituted into the formula.

The third multiplication term is the frequency of use. For example, do you use AI multiple times a day, only once, once a week, or once a year? The difference is huge. The value of an AI scenario used once a day is 365 times that of an AI scenario used once a year.

The last multiplication term is the time saved by single-time performance. This is very easy to understand. This job used to take five hours, but now it only takes one hour. As I mentioned earlier, because you have a single labor cost.

Let’s take the example of finding typos in Get just now. In the past, content colleagues had to submit documents to the editorial office experts for review. If typos appeared, it meant that the teachers had to send the manuscript back or make corrections themselves, which was very labor-intensive. But if AI can do this, even if the accuracy rate is only 90%, it will greatly improve the process of the editorial office teachers and content colleagues, and the time saved can be substituted into the formula to calculate.

Why AI programming is so popular is because the cost is very easy to measure. For programmers, if a piece of code takes 3 hours to complete, then after using AI programming, it only takes half an hour to complete the work. In this way, a programmer can save two and a half hours for one task. In addition, the production and research team has the characteristics of a large number of people, high frequency of use, and high salary. Therefore, the production and research team landing AI tools is the easiest to quantify the benefits. Any company with a production and research team, if you don’t know what to let AI do, then let everyone use the most advanced AI coding tools first, which will definitely benefit a lot!

If there are many scenarios within the company where AI can be applied, which scenario should you prioritize? Starting from these dimensions, you can quickly make a quantitative judgment. First, the number of people involved is large enough, second, the frequency of use is high enough, and finally, the cost of employment saved per unit time is large enough. In this way, you can clearly calculate the value of AI application scenarios.

Of course, in addition to these quantitative indicators, there is another angle to consider. When AI solves problems now, the more detailed and precise the problem is, the better the AI performs. Therefore, don't expect AI to solve all problems, but give it accurate instructions in the areas where it excels. Because the instructions you give are clear enough, it knows how to achieve excellent standards. For example, when you ask it to find typos in an article or write an in-depth investigation report, it will do a better job of the former task. Therefore, when applying AI in an enterprise, you should have a grand vision when thinking about things, but start small when executing. In this process, you will have a deeper understanding of the combination of AI and the company's business.

Another detail that needs special attention is to place AI tools in the most convenient place in the company. If you use Slack, then AI should be embedded in Slack; if you use Microsoft Teams or Google Workspace, the principle is the same. Otherwise, a scenario that was originally used by 100 people may end up being used by only 5%. A scenario that can improve efficiency may become bland.