Categories
科技報導

Isn’t this the ox that Ali blew two years ago?



If you want to understand something in the Internet industry, it is said that you only need to stay in the media long enough. After the Rhino Factory’s new intelligent manufacturing concept has been fermenting in the past few days, we have always felt that the logic of this business is familiar, but we have fallen into self-doubt, thinking about this “empowerment” for small and medium clothing factories and Taobao clothing merchants. It is indeed different because of the rapid iteration of technology.

Isn't this the ox that Ali blew two years ago? 1

However, in the face of various unseen real production lines, we began to “cheer for the people”, and even rose to the long-form analysis of promoting the development of the entire country’s clothing industry. We found a way out of the “confusion”-to put Ali Zeng Some things I did two years ago are shown to everyone, and discuss with the readers what is the difference between the two.

Don’t flatter, don’t belittle, let us see whether this new carnival made by Ali can bring earth-shaking changes to the clothing industry as many people have predicted.

Picture from Tiger Sniff Pro: Rhino Model Factory
Picture from Tiger Sniff Pro: Rhino Model Factory

In an interview with 36 krypton, Ali Rhino CEO Wu Xuegang clearly pointed out that the new manufacturing of Rhino should do two things:

On the one hand, it is oriented to Taobao merchants, to set production according to sales, to achieve small single quick response, and to solve service and cost problems;

On the other side is empowering garment factories to “arrange production capacity reasonably.” Through the model room of Rhino Factory, the digital production model will be “transplanted” to hundreds of factories in the future, so that the factory production process will be digitalized and transparent.

Obviously, this essentially solves the problem of information asymmetry on both sides of supply and demand and makes information flow more efficient.

This is exactly what Alibaba 1688 (B2B business platform) led the “Tao Factory New Manufacturing Project” two years ago.

In an interview with us in 2018, the person in charge of the Tao Factory project said that “has helped more than 30,000 factories to achieve accurate matching based on massive big data, so that high-quality factories can obtain accurate customer resources, and reduce the time and resources caused by the mismatch of supply and demand. waste”.

The matching method is to cooperate with the Alibaba Cloud IPO team:

First, the Alibaba Cloud technical team will help the factory’s production lines to deploy IOT equipment so that the status of each production line in the garment factory is completely digital.

Second, complete the digitalization of the production process through visual recognition and analysis of the clothing production line.

Third, connect the entire online supply chain data.

Obviously, what they want to solve is also the mismatch between supply and demand. The method is a bit like “digital transformation of factories.”

Isn't this the ox that Ali blew two years ago? 2

Fortunately, we accepted Ali 1688’s invitation in 2018 and visited a small town called Qiaosi in Hangzhou, where hundreds of small and medium-sized garment factories gathered. Some can’t even be called factories. Many of them are hidden in residential buildings and should be called “clothing workshops”——

There is no advanced production line, but a production team of less than 50 people can produce hundreds of clothes every day.

In the small town of Qiaosi, Hangzhou, there are many family-style production workshops.
In the small town of Qiaosi, Hangzhou, there are many family-style production workshops.

At that time, Dianshi Factory, a benchmark project of Tao Factory, was the destination of my trip. Unfortunately, the scene I saw at the time was very different from the “automated model room” in rhino manufacturing. But this is the general appearance of tens of thousands of small and medium-sized garment processing factories in the Yangtze River Delta and the Pearl River Delta.

The fabric, the cutting process and the sewing process are all in one space.
The fabric, cutting process and sewing process are all in one space.

Its “inside” is not tall at all. Clothes materials are piled everywhere in the corners. The so-called production line is actually a number of tailors who “flip their fingers” out of the prototype of jeans or hoodies, and then deliver them. The sewing table on the other side.

The whole scene is hectic and has no sense of rhythm, only the buzzing noise of machines and the flying cotton wool filling a space of several hundred square meters.

The incoming materials are piled aside.
Incoming materials are piled aside.

At that time, the slogan of Amoy Factory was “Use low-cost digital software and hardware to help the factory realize the flow of production data.” The owner of Dianshi, Wang Cunshi, is said to have spent 50,000 yuan to purchase this set of software and hardware equipment that combines artificial intelligence and IOT technology.

Therefore, we see that above the relatively rudimentary workbench, the front and rear sides are equipped with the same hardware as a camera. According to the owner of Dianshi, in this small space, about 20 cameras and edge servers that can meet real-time monitoring are installed in all production links including cutting beds and sewing.

Edge server
Edge server

Ali told us at the time that the main function of these IOT hardware is through computer vision analysis, which enables data processing in every production process.

In this way, if the factory receives 10 orders, each order can be automatically matched by the online system to the factory and the buyer into a group, and the online virtual robot will manage the production plan and automatically track the production plan.

Isn't this the ox that Ali blew two years ago? 3

“One link and one link can save time on both sides.” At that time, Dianshi owner Wang Cunshi told us that before, he often took large orders of more than 5,000 pieces, because there was no time to take small orders, which “wasted” manpower and time. , And not very profitable.

However, it is true that business is getting harder and harder year by year, and orders can only be obtained by grabbing, and small factories themselves can’t figure out the market situation.

He quietly told us that every year, many factories owe a lot of money due to product inventory accumulation due to double 11 betting mistakes. “There was a factory owner who jumped off the building a year ago because the factory had too many down jackets to sell and closed down.” He admitted that market forecasting capabilities and inventory clearance capabilities are not available in their small and medium-sized factories.

At the same time, more and more Taobao stores, especially many so-called “designer brands”, are more inclined to “determine the production quantity based on sales”, so they have been looking for ways to optimize the production process and shorten the output as much as possible. Cargo cycle.

“Many times our conflict with our customers is that they think we are not producing for them, and we queue them to the back, deliberately dragging them; but we actually have a fixed schedule for each order, and it is always clear to explain to them .”

Therefore, all transactions and production steps such as “fabric arrival in warehouse”, “feeding and cutting”, “production sewing”, “off-line withdrawal” and other transactions and production steps are open and transparent online-upload pictures or videos from the factory) , Allowing the client to see online that his product has a progress update, which indeed solves a problem.

Obviously, there is a video uploaded in the column of
Obviously, there is a video uploaded in the column of “Feeding and cutting” as evidence

On the other hand, as the owner of the factory, Wang Cunshi disassembled dozens of small production groups into many different combinations in order to be able to take more small orders. This is very different from the flow group of traditional large factories.

“The large order is a standard assembly line with more than 20 people, and the small order has a separate group of 2, 4, and 6 people. It is equivalent to flexibly arranging each order to different scale production groups according to the scale. .”

What’s interesting is that at the time this project also mentioned the need to learn and surpass Zara’s “quick reverse” model.

The concept of rapid response in the apparel supply chain was first proposed by the world’s famous fast fashion brand Zara. And a technology called “Radio Frequency Identification” (RFID) is fully applied in the entire supply chain-through the “identification code” on the product and the production process, the whole process of tracking the product from the factory to the store is realized. And can monitor the inventory situation in real time.

In the Tao Factory project, “computer vision” and “online big data analysis” have become the key technologies for the clothing processing factory in the promotion to realize small single quick reverse and on-sale production. According to official data at the time, through computer vision algorithms, Dianshi Factory optimized the production process, increased production scheduling by 6%, and shortened the delivery cycle by 10%; Taobao Tmall data could help factories make market judgments.

In addition, on the 1688 platform, Tao Factory referenced ZARA’s supplier rating model to classify factories. The higher the level of the factory, the more opportunities to receive orders from high-quality customers.

Isn't this the ox that Ali blew two years ago? 4

Yes, the transparency of delivery and data sharing in each link can make the parallel operation of multiple orders more orderly, and also “standardize” the delivery process to a certain extent, so that each process can be linked to each other. Carrying around the ground can indeed save time;

The online market and sales big data analysis and offline production and manufacturing link data can allow the factory to arrange different optimal combinations of “large, medium and small orders” in the off-season and peak season, without production gaps, and ensure that the off-season is not weak. ; And Taobao merchants can also be matched to factories with suitable production capacity and technology.

Everything seems impeccable in theory.

But if you think about it carefully, there are actually some visible loopholes and confusing conclusions.

First of all, is computer vision necessary to make the production process transparent? Besides image data, what other data is there?

According to industrial Internet industry professionals who have done similar projects, many clothes are stacked on top of each other, and image recognition is almost ineffective-whether it is to confirm the quantity or confirm the quality, it is not very useful.

“In this environment, it is very difficult. Even if the process is completed, the worker can press the button directly, without the need for a machine to confirm.

And the completion confirmation is not to confirm the appearance of a certain piece of clothing, but to confirm through some execution actions or quantity. In general, image recognition is very tasteless in this complicated environment, but the camera can do real-time monitoring and on-site working conditions confirmation, such as confirming the processing conditions of workers, it is necessary. “

So the question is, in addition to image data, what kind of available data is collected on-site to deploy manpower and adjust inventory? After walking around the workshop, I found nothing else.

Isn't this the ox that Ali blew two years ago? 5

Secondly, although clothes can be “standard products”, in such a small garment processing factory with less than 100 people, the process is actually not standard at all.

During the visit, the boarding team worked in a closed room, where no equipment was installed; and there was no trace of digitization in the corner where the clothes were stored. A group of two or a group of four makes small orders by hand, and its work is even more difficult to quantify with cameras or other sensors.

Originally, the intelligent transformation of industry was based on automation. Without equipment automation, what about intelligence? This transformation is more like a “transaction on the cloud” rather than an intelligent transformation.

The question was raised to the boss at that time that he also admitted that some links “really need to be manually clicked to complete”, and the significance of online orders lies in “avoid arguing on both sides and parallel progress of orders.”

Isn't this the ox that Ali blew two years ago? 6

Third, the so-called less than 50,000, you can use standard software and hardware equipment to help small and medium-sized clothing factories complete the intelligent transformation, which sounds a bit like a hooligan.

Even though the garment factory process is relatively simple, as many people say, the size of the production workshop of each garment factory is different, the number of employees is different, the function and age of the equipment are different, the scale of each production link is different, and the processing technology is different. It is also different, and the flow of personnel is also different…I don’t know how to generalize with the same price and the same amount of hardware.

In addition, the engineer’s labor costs, real-time system updates and follow-up service costs are also very expensive.

Isn't this the ox that Ali blew two years ago? 7

Fourth, to determine production by sales, Ali can indeed help the factory to achieve. For example, what is popular in the season, using artificial intelligence to predict user preferences is indeed more effective than factory bosses. However, technology can only do well the single-dimensional work of the online part, and the offline production workshop is a complex polyhedron. Even if the online data is completely connected with the offline production process data of the factory, it cannot solve the problem of affecting the factory. The essence of survival.

Technologists and R&D technicians also need to have a high level of insight into the garment industry’s production process, master craftsmanship, the flow of factory personnel (workers leaving and staying are very unstable), market changes from foreign trade to domestic sales, and fierce competition. Power and comprehension.

In addition, you may have overlooked a detail of the benchmarking project “Dianshi”, that is, all groupings are arranged by the boss-the ratio between the large group and the group. How to group to maximize production efficiency and achieve sufficient flexibility? According to our on-site observation, there is no trace of system decision-making on site. “People” still play a decisive role in small and medium-sized factories.

In short, don’t try to use words such as “technology” and “reconstruction”, just want to completely solve the long-standing pain points of garment processing factories and even an industry.

Some people say that Ali wants to use the model room to sell digital solutions to domestic factories.  I would like to ask how many factories can buy this robotic arm, and I need to attach a
Some people say that Ali wants to use the model room to sell digital solutions to domestic factories. I would like to ask how many factories can buy this robotic arm, and I need to attach a “recovery strategy”, but Rhino Intelligent Manufacturing has always answered vaguely on business models and revenue issues. Picture from Tiger Sniff Pro

Two years later, when we saw that Ali had created the “model room” of the rhinoceros factory, on the one hand, we were thinking about whether Ali felt that it could not break through the cost and manpower bottlenecks that existed in domestic small and medium factories. We simply built a demonstration ourselves and deployed it. We use advanced equipment to digitize everything from the bottom to the top; on the other hand, we are also wondering if there are some unknown changes in this project two years ago.

Therefore, I contacted Ali 1688 again, and the other party has already stated that this project has not been done, and “upgraded” to the C2M business, connecting the domestic trade wholesale platform 1688 with Taobao special edition. In other words, it is to help the factory do a good job of “connecting” and “matching” with the consumer, focusing on “factory direct sales” and not mentioning the issue of participating in the transformation of the factory.

In addition, when I contacted the Dianshi factory again, there was no response from the other party.

Now, instead of the Alibaba Cloud IOT team, it is made by the Rhinoceros intelligently bred from the Tao Department. It has built a benchmark “model room”. It once again mentioned the application of artificial intelligence, IOT and other technologies to the production workshop to integrate process data and production capacity. Data and production line data are all connected, and the generated data is used to determine the scheduling of workers; once again, it is mentioned that the massive data of Taobao is used to help clothing brands make market forecasts…

It sounds the same as two years ago, but it seems different.

But at least we know that if the simple equipment and workshop layout of the small and medium-sized factories along the coast with meager profits and fierce competition, if you just set up a camera and don’t modify it from the bottom, you will not get enough effective data.

But then again, it doesn’t seem to make much sense for Ali to build a prototype factory with a robotic arm and a spider web hanging system. Whether it can be used as a reference for small and medium-sized factories that are constantly on the verge of bankruptcy.

So back to Alibaba, will the benefits from this model room be enough to support them in establishing a second and third factory?

Perhaps Ali calculates the return on investment ratio himself, and after achieving breakeven in the future, those who intend to join may promote this model. Then we will record the effect again in words.

Produced | Tiger Sniff Technology Group

Author | Utada