The company needs a set of autumn and winter commuting jackets with a style similar to last year's, but with lighter fabric and no price increase. We hope to see the first version of the design plan tomorrow morning

In his years of experience in the clothing industry, Liu Chen, CEO of Lingdi Technology, has frequently encountered similar business scenarios. Before a new clothing product is launched, a typical production path is "manufacturing companies do clothing design and manufacturing - clothing brands provide requirements and make selections - formal launch". Usually, after clothing brand manufacturers propose design requirements, manufacturing companies need to quickly come up with design solutions.

On the one hand, the seemingly simple clothing design requirement mentioned above requires manufacturing companies to quickly mobilize multiple positions and systems. This is because brand owners' previous clothing design information was usually scattered in sales WeChat and emails, with styles, fabrics, patterns, and project files in the file library, while process judgment and delivery relied more on personal experience.

On the other hand, the clothing industry is based on a predictive spot business model, and there is a certain lag and possibility of errors in the prediction of new clothing. In order to bring product development closer to the market stage, clothing brand manufacturers need to extremely compress research and development time to reduce the probability of producing incorrect clothing and greatly optimize clothing inventory.

Under the contradiction between supply and demand, it is particularly important to use clothing design tools to improve design efficiency. In 2015, when Liu Chen started working on Style3D, an industrial software for the textile and clothing industry, he found that even though the sales end of the clothing industry had rapidly informatized with e-commerce at that time, the digitalization level of clothing design, research and development, sampling, and production processes was still relatively low, and a large amount of work still relied on human experience, 2D drawings, and physical samples.

On September 13th, in an interview, Liu Chen stated: "Ten years ago, small and medium-sized clothing enterprises had already felt problems such as fragmented orders, accelerated style updates, and rising inventory pressure. There was an urgent need to improve efficiency and reduce costs, which drove the rapid popularization of digitalization in the clothing industry. Nowadays, under the extreme demand for cost compression and efficiency improvement in the clothing industry, injecting AI (artificial intelligence) capabilities into the digitalization of clothing has ushered in a period of demand explosion. ”

In 2024, Lingdi Technology was selected as a national specialized and innovative "little giant" enterprise. In 2022, the company completed nearly $100 million in Pre-B+round financing, with continuous investment from multiple institutions such as Hillhouse Capital and CDH. At present, the company's clients have grown to over 3000, including ANTA, Bosideng URBAN REVIVO、 Taiping Bird, NetEase, MiHoYo, etc. Previously, the company also launched a platform product Style3D that integrates "AI+3D" technology, mainly used for digital design in the clothing industry. On a technical level, Lingdi Technology holds 106 authorized patents worldwide.

Liu Chen stated that the clothing industry is going through a critical stage of transitioning from mechanization and digitization to digitization. AI perception of demand, 3D digital twin verification, and the implementation of intelligent manufacturing will jointly promote the industry's evolution from "prediction based spot trading" to "real-time production based on determined demand".

The boundary of AI's ability to overlay digitalization

As a well-established textile and clothing import and export enterprise established in 2002, Hangzhou Jinhui Trading Co., Ltd. has long been plagued by uncertainty in the clothing design process. For example, in the process of cooperating with PRIMARK (an Irish clothing brand) through ODM (Original Design Manufacturing) mode, the selection rate of new clothing styles is usually only about 10%.

During an interview, Zhou Bin, the person in charge of Hangzhou Jinhui Trading Co., Ltd., stated that in the competitive landscape of multiple suppliers in China, if an ODM enterprise faces a low selection rate for new clothing styles, coupled with long cycles of repeated design changes, sampling, and shooting, and high cost constraints, the enterprise will fall into a disadvantage in order competition.

In response to such pain points, Hangzhou Jinhui Trading Co., Ltd. has adopted Lingdi Technology's AI products. Afterwards, the company used AI to quickly generate multiple versions of solutions for each clothing design for customers to choose from. After customers determined the direction, AI visualization videos were produced to optimize the customer docking process.

Zhou Bin introduced that originally it took 4-8 hours to render a clothing video online, but with the help of AI capabilities, it has been shortened to 25 minutes, increasing efficiency by about 10-19 times. The time required for clothing sampling has also been reduced from two weeks to 8 hours due to AI. Overall, the cost of clothing design has been reduced by about 80%.

The improvement in efficiency also translates into an increase in orders. He said that with AI empowerment, customers can expand their options from ten to dozens. When customers are still willing to look, the number of options that clothing companies can offer often determines how many orders they can receive. Higher efficiency and precise demand have increased the selection rate of new clothing styles to 30% -40%.

This case is not an isolated one. Liu Chen stated that promoting the integration of AI capabilities into the company's existing 3D industrial software stems from the practical needs of the clothing industry. The clothing industry has long relied on a large amount of manual experience, with long design and development processes and repetitive work. At the same time, it faces problems such as accelerated speed of new product launches, rising costs, and inventory pressure. The company hopes to use AI to lower the threshold for digital content production, improve design, redesign, review, and collaboration efficiency, and enable enterprises to respond to market changes more quickly.

From the accumulated customer cases, Liu Chen summarized and analyzed that in the fashion design industry, AI capabilities can play a role in the following four business scenarios:

The first is product planning and design. AI can integrate trend data, analyze brand styles and market information, assist in theme planning, series planning, cultural and creative products, integrated creative products, pattern design, and local redesign, helping designers quickly transform inspiration into executable solutions.

The second is R&D collaboration and asset management. AI can support the recognition, retrieval, and recommendation of styles and fabrics, assist in product review, quality inspection analysis, and automatically complete fields such as category, color, pattern, and fabric, making enterprise images, styles, and fabric information searchable and reusable digital assets.

The third is product visual and marketing. Based on digital samples, AI can complete virtual fitting, model and background replacement, product image and short video generation, and adapt to the content specifications of e-commerce and different channels, reducing commercial shooting and content production costs.

The fourth is enterprise level business collaboration. Through AI intelligent agents and workflows, enterprises can connect trend planning, promotion of products and materials, design and development, product launch and review, and collaborate with their existing product management systems.

In order to better integrate AI capabilities into the daily workflow of enterprises, Lingdi Technology recently released the fashion industry AI all-in-one workbench StyleWork. It can combine the understanding ability of general large models with the company's long-term accumulated knowledge of clothing verticals, 3D simulation technology, process experience, and digital assets, allowing AI to understand the company's products, processes, standards, and business habits, evolving from a passive "tool" to an active collaborative "colleague".

Of course, the limitations of AI application in the fashion design industry are gradually exposed as the number of users increases. Liu Chen stated that general AI can improve the efficiency of creative generation, but industrial scenarios not only require good appearance, but also meet the requirements of pattern, fabric, process, and manufacturability. Therefore, AI must be combined with 3D simulation, industry knowledge, and real production data to improve the accuracy and controllability of the results. In addition, AI needs to be embedded into the existing design and development processes of enterprises, while balancing usability, investment costs, and actual benefits. Only by truly reducing repetitive work, shortening research and development cycles, and continuously calibrating through production feedback, can AI transform from auxiliary tools into stable productivity.

The first person to 'eat crabs'

In the past few years, clothing foreign trade company Suhao Fashion (600287. SH) has gradually accumulated the company's static data into digital assets such as 12918 fabrics and 3959 3D silhouettes, transforming them into searchable, reusable, and collaborative resources. After partnering with Style3D, Suhao Fashion fully integrated these digital assets into the product design, sampling, and production processes.

Yang Yi, assistant manager of Suhao Fashion R&D Innovation Business Unit, said that the experience brought by AI to downstream customers is mainly "fast and accurate". For the company, the greatest value is that suppliers and brands can resonate with each other to a higher degree - designers' ideas can be presented to customers more accurately and intuitively, significantly reducing communication costs and improving decision-making efficiency.

This change has also brought many unexpected orders to the company. In the interview, Suhao Fashion provided a practical case - a Canadian client whose 90% of production is in Bangladesh had a low dependence on Suhao Fashion at one point. In order to impress customers, Suhao Fashion not only presents the patterns, silhouettes, and arrangement plans of clothing online, but also provides suggestions in visual marketing, store display, and other aspects, becoming a deep participant in customers' business and sales processes.

Yang Yi stated that the amount of orders placed by Suhao Fashion through 3D modeling assistance increased significantly last year, and this year this amount is expected to double by about. Nowadays, most designers at Suhao Fashion have integrated 3D technology into their daily workflow. If we don't have the 3D tool, our work efficiency will be greatly reduced. ”

In addition to listed companies like Suhao Fashion, Liu Chen explained that based on current data, the most widely accepted and applied Style3D platform is still among brand manufacturers, ODM/OEM manufacturing companies, material and accessory companies, and clothing e-commerce customers in the textile and clothing industry chain.

Liu Chen stated that manufacturing enterprises have the highest proportion, largely due to the short clothing manufacturing chain, especially ODM manufacturing enterprises that only need to do some design and development for brand owners or customers to obtain orders. Compared to manufacturing enterprises, brand enterprises involve more links such as product planning, design, research and development, sampling, and marketing, and digital applications usually require more departmental collaboration.

In Liu Chen's vision, with the rapid development of AI technology and the improvement of AI applications in the clothing industry, the potential users of clothing design software that integrates AI capabilities will exceed market expectations, and the creativity of the clothing industry will also be greatly released. He said: "Even if individuals have never been engaged in design work, such as small sellers on Tiktok and the proprietress of Hangzhou Sijiqing Clothing Market, they may create a clothing brand in the future."

(Author Tian Jin)