(Image source: Pixels)
Sanweijia explained a truth: how digital enterprises can use their existing advantages to build AGI capabilities.
@商科星球original
Author丨Yuan Jing
Edit丨Big Rabbit
AIGC has been hot for more than half a year, and people are no longer unfamiliar with ChatGPT.
From Internet giants to the previous generation of AI companies, no one wants to miss this round of gluttonous feast. But objectively speaking, as AIGC becomes more universal, the dividend period of technology is constantly shortening. As a result, people began to re-examine the actual value behind this wave.
From the perspective of Digital Planet (ID: digital-planet), new AGI companies are racing against time, and whoever can build a moat in a short period of time will have the last laugh. As a digital company, Sanweijia has a clear view of AIGC.
The company's insights in the home improvement industry are profound and logical. At a time when countless companies are attacking the vertical field model, we believe that the experience of this company is worth learning. This time, Digital Planet (ID: digital-planet) talks to Cai Zhisen, the founder of Sanweijia. We will explain the profound aspects of the home decoration industry in simple terms, which will be beneficial to AIGC practitioners.
01
The particularity of digital home improvement
The home improvement industry involves a wide range of areas, from design and construction to home materials and furnishings, with many product SKUs and strong user customization. In the lives of ordinary people, they spend a long time at home, and the home environment has a very direct impact on daily life.
No one wants to live in an uncomfortable environment, and this has made the digitization of the home improvement industry very professional. For a long time, the digitalization of the home furnishing industry has not been well known to the public. Only in vector graphics design, the industry requires designers to master a higher level of graphic design and computer technology, as well as higher knowledge of aesthetics and home decoration.
Can AIGC take root in such a large industry? Can it replace traditional home improvement software? Faced with this question, Cai Zhisen gave the answer: "Yes, but the general-purpose Vincent graph or Yitu graph software are 'toys'."
A certain designer revealed to Digital Planet (ID: digital-planet) that he had tried to use Vincent diagram software for design, but the embarrassing thing was that after the prompt word was changed, the size or outline of the generated picture also changed. . As a result, the finished product is difficult to deliver. In addition, if you want to modify the details of the picture, the existing AIGC products do not support it.
In fact, the current AIGC software only cuts into the tip of the iceberg of the home improvement industry. From the perspective of the industrial chain, the software ecology of this industry is composed of To C-end apartment design software, cloud design platform and design + manufacturing products. In the current Vincent diagram software such as Midjourney, etc., only 2D image-based renderings can be produced. Since there are no specific parameters, it is difficult for such pictures to generate more value, and it is impossible to actually participate in the physical manufacturing process.
In Cai Zhisen's thinking, software such as GPT can realize part of the "layout" function, but the rendering level has not been involved: " Behind the rendering is controllable shape and material. The finished picture needs to combine different material textures and lighting effects. However, these functions are not yet available in general AIGC products."
In order to achieve "controllability", a large amount of vector data accumulation is required. It can even be said that without vector data, it is impossible to make digital products in the home improvement industry.
Cai Zhisen's confidence comes from the fact that Sunway Home has a library of 1.63 billion+ design schemes, covering 28 million+ residential area data across the country. The huge and real vector data set makes the application of large models possible.
02
Vector parameter data + AIGC
What is a vector diagram and what does it look like?
People who are familiar with home decoration may have an impression. Usually, designers use professional 2D design software to generate home plans for customers. In the later stage, this kind of black-and-white design drawing with rules and regulations can also generate 3D effects according to the house type parameters. Similarly, in the home furnishing mall, the sales staff also use this type of terminal software to display the effects of home decoration elements such as panels and furniture.
Regardless of the complexity of the design drawing, the designer must draw it bit by bit. From the perspective of AIGC, this kind of work is a typical repetitive labor scenario. In principle, at least in the field of home decoration design, there is huge room for upgrading.
Therefore, the essence of AIGC upgrade in the home improvement industry is the AIGC of vector data. To expand, in the 3D modeling process, digital manufacturers need to have a large amount of house floor plan data; in the design of building materials and home decoration units, it is necessary to accumulate parameter data for each category and each consumer product.
Judging from the results, the home improvement industry after the AIGC upgrade is very attractive: it will mean that consumers no longer need to visit model rooms offline to determine their needs, and are no longer subject to the red tape consumption routines of the offline home improvement industry. At that time, even if consumers do not understand home decoration, they can plan their own home design through massive generative content.
"If ergonomics, light and temperature data are taken into consideration, the system can even recommend the optimal solution for the home improvement plan without taking a position." A senior industry practitioner told Digital Planet (ID: digital-planet).
"To put it simply, AIGC has changed home decoration from 'fill in the blank' to 'multiple choice'." Cai Zhisen said.
03
Changes in the home improvement industry after AIGC
It is worth noting that AIGC will not only change the field of home decoration design, but also have a profound impact on the industry as a whole.
In the past, C2M logic ran through the manufacturing industry. Consumers make demands directly to producers, bypassing distributors for cheaper prices. Today, because AIGC companies have accumulated a large number of data sets in their products, combined with algorithms, they can provide manufacturers with more valuable data.
In addition, manufacturers can combine data for new production scheduling. After marketing activities, users can get feedback on browsing and ordering on the AIGC platform. Finally, these data are combined into different production lines, so that the capacity of the production line can be fully utilized.
"For example, in Zhibang Home Furnishing, we helped them increase the utilization rate of raw materials by about 2%, which saved tens of millions of costs every year for a company with a scale of 2 billion." Cai Zhisen replied .
In addition to enabling manufacturing, AIGC products can also empower logistics. Similar to the reason that home furnishing needs vector parameter data, trucks and containers also rely on spatial layout data during the loading process. By entering the price of the goods being shipped into the system, freight companies can use space more efficiently.
As far as digitalization itself is concerned, due to the real-time and accurate data provided by upstream AIGC companies in the industry, such data can be reused by all links in the industry chain, so that there is no need to "reinvent the wheel" and improve the efficiency of data use among enterprises .
For To B users, while enterprises use data, they can generate desired content based on the minimalist dialogue system provided by generative AI, which simplifies hundreds or thousands of click operations in the past into one input.
The impact on Sanweijia is that after the AIGC wave, the company's strategy is more focused and more AI-oriented. Cai Zhisen reorganized the focus of the enterprise, and he said to Digital Planet (ID: digital-planet): " At present, we focus on three issues, namely, what problems should be solved in combination with what scenarios, how to generate data labels, and how to Build an algorithmic model."
Conclusion: At present, no matter at home or abroad, the combination of AIGC has become a trend. In our understanding, enterprises represented by Sanweijia are actively embracing changes. Under optimistic expectations, the AIGC of 3D vector data has a promising prospect, and the changes it has caused are not limited to the home improvement industry. secondary creation.
However, objectively speaking, due to the particularity of the home improvement industry, AIGC of 3D vector data is difficult. This requires converting the data previously stored in the database into a mathematical model, and using deep learning to classify, identify, segment, etc., so as to realize automated data processing and analysis. In Cai Zhisen's words, it is: "We need to extract them from the database and turn them into 0 and 1".
For domestic "stuck neck" industries, the impact of AIGC will be far-reaching. Digital Planet (ID: digital-planet) believes that after digitizing the above product parameters and AIGC, the productivity it will drive will be extremely amazing.
In the future, AIGC will become the fulcrum of leverage, multiplying the existing productivity several times. Cai Zhisen hopes that relying on AIGC as the fulcrum, Sunway Home will start from the home improvement industry and become a world-class general industrial software company.