At a time when the AI wave is sweeping in, many product managers are both excited and anxious, as if they are holding a “dragon slaying knife” in their hands, but they don’t know how to find the real “dragon”. This article will combine the insights of Liang Ning’s “Real Needs” and Fu Sheng’s views, and through an actual product case, we will delve into how to start from the real needs of users in the AI era, avoid the trap of “AI for AI’s sake”, and truly transform AI technology into user-perceptible value
The recent AI circle is still hot, and it is not an alarm clock or a dream that wakes me up every day, but a bunch of new AI terms and screen-level applications.
It feels like it’s constantly highlighting – if you don’t add an “AI-Powered” label to the product solution, we are embarrassed to say hello when we go out.
The same goes for bosses, opening their mouths is to be called a big model, and closing their mouths is to benchmark OpenAI, although the way to embrace AI every day is still only DeepSeek half a year ago.
According to my observation of the students around me, under this wave, most of our product managers are both excited and anxious – excitedly, there seems to be an extra “dragon slaying knife” in their hands; Anxiously, looking around, it seems that there are dragons everywhere, and it seems that there is no dragon to be found.
As a result, a strange phenomenon appeared: AI for AI’s sake.
Many teams treat AI as a KPI rather than a tool. As a result, the functions made look full of technology, but users vote against them mercilessly with their feet.
Brother Jing felt that it was like putting an old tractor with an F1 engine for a racing car, which sounded fierce, but in fact, it ran two laps on the road, either falling apart by itself or plowing the ground into a mess.
Two days ago, I just listened to a podcast by Cheetah Fu Shengfu, and he has a point that I deeply agree with: in the AI era, applications and experiences are the real value highlands.
No matter how good the technology itself is, if it cannot be transformed into user-perceptible and real value, then it is just a castle in the air.
This also reminds me of what Mr. Liang Ning repeatedly emphasized in the book “Real Needs” – to gain insight into the real needs of users, rather than staying on the surface of “what to want”.
How can product managers do a good job in B-end digitalization?
All walks of life have taken advantage of the ride-hail of digital transformation and achieved the rapid development of the industry. Since B-end products are products that provide services for enterprises, how should enterprises ride the digital ride?
View details >
The so-called real needs are often rooted in the depths of human nature, “greed, hatred, and foolishness”, or more mildly, those things that can make users “addicted”, “improve efficiency”, “feel safe” or “get rid of fear”.
You see, technology is changing, but the underlying logic that drives people’s hearts has never changed for thousands of years.
Today, Brother Jing wants to use this topic, combined with a product case we have personally experienced, to talk about how we can clear the fog of technology and find and design those “real needs” in the AI era.
1. From “no one uses it” to “inseparable”, the evolution history of AI for a repair work order
In my opinion, the best way to learn is to review.
Let’s take a look at a project that the team did in the early years, optimized the front-end time, and now seems particularly interesting – a vertical field repair work order module.
Stage 1: Scan the code to fill in the “self-hi”
In the earliest version, we also think of the “Internet” – if the customer’s device is broken, the mobile phone will scan the QR code on the device, and an H5 page will pop up.
Above is a bunch of forms: fault type, device number, problem description, upload images…… The design logic is clear, the fields are complete, and the data background is neat.
We thought at that time that this was so good, how rigorous the logic was, and the data was directly stored in the warehouse, saving the trouble of manual entry, which was perfect!
Then, the product is gorgeously …… It is online.
The result? There is very little data in the background.
Later, when we went to the site to investigate, we found that users would rather call the 400 phone number printed in the corner of the device than scan the code.
Why?
An elderly maintenance master woke us up with a sentence: “Young man, my fingers are very thick, my eyes are not good, you let me poke on this small screen, and I have to take pictures and upload them, and when I finish this, I have made it clear to the dispatcher three times on the phone.” What am I trying to do? ”
This sentence is a basin of cold water that extinguishes the flame of our “taking it for granted”.
The process we think is efficient, but in the real scene of users, it is a “high-friction and anti-human” design. We solved the company’s data specification problem, but it caused a lot of trouble for users.
In Brother Jing’s view, this is a typical “pseudo-demand” in the laboratory –It only satisfies our desire for management and imagination, but does not respect the user’s usage habits and real pain points.
To paraphrase Mr. Liang Ning’s words, we did not find the “pain points” or “itch points” of users, but became the “pain points” of users.
Stage 2: “Breaking the ice” of AI voice repair
Later, the project was shelved for a period of time until the AI wave came, and some students in the team asked, can it be transformed with AI?
This time, we learned well, and instead of drawing a pie as soon as it came up, we returned to the core question: What is the most natural and lowest cost interaction method for users when applying for repairs?
It is “saying”.
So, we did the first AI iteration:AI voice repair。
After the user scans the code, it is no longer a cold form, but an interface similar to WeChat voice. The user held down and spoke and directly described the problem: “The No. 3 compressor in workshop A seems to be not turning, and the sound is a bit abnormal, please send someone to take a look.” ”
Then, a miracle happened.
The system background converts speech into text through speech recognition, and then calls the large language model to “extract information” from this natural language.
The model will automatically identify “Car A-No. 3 Compressor” as the equipment name and “Abnormal Sound” as the fault description, and then automatically populate it into the work order system, and directly assign the work order to the corresponding maintenance engineer according to our preset rules (for example, “Workshop A” corresponds to Zhang San being responsible).
This small change has an immediate effect.
The open rate and ticket creation success rate of the new version have directly doubled several times. Why? Because we greatly reduce the interaction cost of users.
I often say this in the team:The same is information input, and the experience of the user input process directly determines the life and death of product functions.
The logic is also very simple, if you ask the user to do fill-in-the-blank questions, he will instinctively feel irritated and anxious. But if you ask him to do oral questions, he will feel relaxed and natural.
If the information said by the user is incomplete, such as not saying the device number, the AI will also ask through voice like a real assistant: “Hello master, which device is in workshop A?” Please let us know the device number so that we can locate it faster. ”
You see, this is the first value that AI brings:It allows machines to accommodate people, not to adapt to machines.It turns a complex, structured information entry task into a simple, human-like conversation.
This solves the user’s “laziness” and satisfies their nature of pursuing convenience, which is a huge improvement.
Stage 3: The “endgame” of AI telephone operators
Although the data came up, when colleagues went to the scene for a return visit, they discovered another “silent truth”.
Most of the older masters are still used to making phone calls. For them, when they encounter a problem, they take out their mobile phones, find the familiar number from the address book, and dial it out, which is a muscle memory engraved in their bones.
What to do? Do you want to give up this part of the “stubborn” users?
No.
Brother Jing always believes,When your user behavior does not match your product design, the first thing to reflect on is always your product, not your users.The user’s habits are the optimal solution screened by the market.
We should embrace it, not try to “educate” it.
At this time, we came up with a bolder and more thorough plan:Let AI answer the phone.
We have introduced the ability of intelligent outbound calls and voice robots – when users call the traditional repair call, the answer is no longer a busy operator in the dispatch center, but an AI robot with a sweet voice and quick response.
“Hello, this is the XX equipment repair center, how can I help you?”
The entire conversation process is almost the same as our second stage of voice repair: AI guides users to say questions, understand semantics, AI extracts information, creates work orders, and completes distribution. The whole process, the user’s experience is almost the same as making an ordinary human call.
But it’s a huge leap forward for both the customer and the company.
In the past, the data of telephone repair was precipitated offline, and the operator needed to manually enter the system twice or even three times, which was inefficient and had a high error rate.
Right nowUsers’ front-end habits haven’t changed at all, but the data they generate makes its way into our data platform seamlessly, in real time, and structured the moment it happens.
In my opinion, this is the true user-driven “killer” app.
It doesn’t create any new and cool interactions, on the contrary, it perfectly explains what it means to “follow the user, follow the market, not follow the intuition of AI”.
It is true that this solution has costs, interface fees, and development resources, but the value it brings is decisive, because it not only solves the pain points of users, but also solves the management pain points of enterprises, and opens up online and offline data islands.
This is spending money on the blade.
2. What does this case tell us?
In my opinion, the real opportunity in the current era of AI is often not to create new demands, but to use AI as a new technology to “redo” those old scenarios that already exist but have a bad experience.
First, go deep into the scene and do “fieldwork”, not “keyboard man”.
The most real needs are never in the office, not in the competitive analysis report, but in the hands, mouths and complaints of users.
If it weren’t for the maintenance master’s sentence “what am I trying to do”, we might never realize the root cause of the problem.
This is especially important in the age of AI. Because the boundaries of AI’s capabilities are very wide, but also very vague.
Only when you truly understand the ins and outs, interest entanglements, and user habits of a scene can you judge what role AI can play here, whether it is “icing on the cake”, “sending charcoal in the snow”, or “adding to the snake”.
Another example:
Many products are doing “AI intelligent summary”, summarizing a long article or a video into hundreds of words with one click. Is it worth it?
Of course. But how much is it really worth? You need to analyze the scene.
In order to quickly write a reading report, a student uses AI to summarize a famous book, which is a real need.
In order to quickly screen project information, an investor uses AI to summarize hundreds of pages of business plans, which is also a real need.
But if you ask a user to use AI to summarize a movie he has been waiting for for a long time, he will most likely think you are crazy – because the user wants an immersive experience, not an efficient acquisition of information.
SoDon’t fantasize about what AI can do in the conference room, combine your scenarios to see what they are doing and how painful they are.Where the pain lies, the opportunity for AI lies.
Second, embrace “laziness” and be “efficiency paranoia”
Human nature is “lazy”.
Any design that requires users to pay additional learning, operation, and cognitive costs will most likely fail.
On the contrary, any design that allows users to be “lazy” and complete tasks “without using their brains” has the potential to become a hit.
In this regard, AI is a natural “efficiency artifact”.
In this regard, Brother Jing has recently shared a lot of AI efficiency improvement tools, such as Cursor’s efficiency improvement for prototyping; Gemini’s efficiency in meeting summaries the efficiency of the buckle agent for the working day report; Genspark improves the efficiency of work reporting.
And so on.
You see, AI has not created new needs, the product manager’s need to “understand the user’s voice” has always been there, and what AI does is to turn an extremely tedious “physical work” that used to take 10 hours into a “mental work” that only takes 10 seconds.
It frees product managers from repetitive labor to think about more important things.
And this is the real need.
All we need to do is try to find those repetitive, inefficient, and painful links, and then use the “bull knife” of AI to cut it down accurately.
Having said so much, the core is actually one sentence:AI is a hammer, but the eyes of product managers must always keep their eyes on the nail.
And this requires us to put aside the blind worship of technology and return to the most basic and core values of product managers: insight into human nature, understanding of scenes, and persistent pursuit of user value.
Therefore, future product managers do not want to become AI technology experts, but to become a “scene translator” who understands business and users better: they can accurately translate the powerful capabilities of AI into actual experiences that users can understand, are willing to use, and cannot do without after using them.
Experience and application may be the greatest value of AI at present.
