TMTPost AGI learned that Glodon recently announced its new AI model AecGPT for the construction industry，which has a total of 32 billion parameters, as well as a complete set of industry AI platform solutions.

The announcement was made at the China Digital Construction Conference 2024.

In particular, AecGPT is aimed at industry-level applications, covering 20 sub-expertise areas in seven fields of the construction industry, including planning, design, transactions, costs, construction, operation and maintenance, and comprehensive management. It has professional capabilities such as automation, analytical decision-making, and assisted generation. The model has billion-level Tokens Industry Knowledge Enhancement (RAG) technology, with an average score of 97% in China’s first-level construction engineer examination, higher than that obtained by the general large model GPT-4. By using AI models to analyze and generate work items, efficiency is improved by 5 times; dynamic planning and tuning improve efficiency by 10 times.

Yuan Zhenggang, Chairman and President of Glodon, said that the future has come and AI is already changing society. The core capabilities of AI in the construction industry are the ability to automate repetitive work, analyze and make decisions based on data insights, and understand and generate professional content. The AceGPT large model can make BIM applications richer, computational efficiency higher, and data more secure and reliable, enabling us to make more accurate and faster decisions.

After the conference, Yuan told TMTPost AGI, etc., AI technology needs iteration, data, and application. “If you develop an AI application, but it is not used in scenarios and not deeply applied, it cannot be iterated and may die in the future,” he said.

"Glodon is more willing to do what it is good at. In areas where we have no accumulation and are not good at, we will open it up and work with partners," Yuan  said.

Yuan emphasized that "the current cost of AI large models mainly includes computing power and talent. Glodon did not create basic large models, but instead created industry models with billions of parameters. The cost of computing power has decreased significantly, and the investment in this area is within a reasonable range. Therefore, we believe that this investment is worthwhile and will continue to invest."

According to the Titanium Media App, Glodon is likely to announce a technical plan for industry MoE large models with trillions of parameters in the future.

Glodon, established in 1998, went public on the Shenzhen Small and Medium Enterprise Board in May 2010，making it the first A-share listed company in China's construction engineering informatization field. The company, positioned as a digital construction platform service provider, focuses on the construction industry and providing customers with digital solutions for the entire life cycle of construction projects.

Currently, Glodon has eight global technology research and development centers, over 80 global subsidiaries, business operations in more than 100 countries and regions, and serves over 370,000 enterprise customers, over six million service construction projects, and over ten million end users.

Yuan said that tThe current digital transformation of the construction industry is encountering challenges, including ineffective implementation of smart construction sites, inaccuracies in data reporting for enterprise informatization, high costs and limited effectiveness of Building Information Modeling (BIM) software, and a lack of clear direction in the construction industry's internet integration. At the same time, globally, Labor productivity in the construction industry lags far behind the average level, with only 30% of on-site work time directly contributing to value creation. The remaining 70% of time is wasted due to delays in progress plans and inefficiencies in managing production materials.

"Previously, there were many misconceptions about digital transformation, and every company would take some detours. For the construction industry, we believe that the era of 'high speed' development is already  gone, and we have reached the stage of 'high quality' development, requiring continuous deep and thorough digital application," said Yuan.

In Yuan's view, digital transformation with refined management as the goal is the new model for the development of the construction industry. At the same time, the key to enterprise digital transformation lies in the integration of connectivity, data, and algorithms, breaking through the barriers of business and data silos. In addition, platformization is also a necessary path for industrial digital transformation.

In response, Glodon provides the core PaaS capabilities for digitalization in the construction industry, namely the "Glodon Construction Business Platform". Furthermore, it focuses on the refined management of cost, production progress, safety, quality, materials, and labor services. Therefore, Glodon hopes to systematically build an integrated solution for digital transformation in the construction industry, based on the three levels of point, line, surface, volume, data-connection-algorithm, and business line focus.

As of now, Glodon has more than 500 ecological partners, including those using core technology platforms such as BIMFACE, a lightweight BIM technology and GDMP, a numerical dimension design platform, eveloped independently. It also has more than 5000 enterprise customers, including design institutes, general contractors, and construction parties, using its applications.

Regarding AI, according to Glodon, the new AecGPT has basic, industry, and professional capabilities. Based on the large-scale AecGPT model, AI can reduce the calculation time for billion-scale projects from three months to two weeks. In addition, the AI-driven design platform, Concetto, has functions such as AI estimation, AI 3D model generation, and AI copilot.

Yuan pointed out that as early as 2015, the company formed (established) an AI research and development team and has continued to invest heavily in research and development. It now has more than 100 intelligent application scenarios. By 2023, Glodon has launched the "All in AI" strategy, strengthening research and development investment in generative AI in the construction field to promote intelligent innovation in the construction industry. So far, Glodon's large-scale models and AI products have been integrated with other traditional products.

Meanwhile, using its large-scale model, Glodon has also developed an all-in-one AI-native application development platform, the Glodon Industry AI Platform. This platform comprises three layers of AI technology: the construction industry AI large-scale model layer, the tool platform layer, and the product application layer. These layers enable the AI enhancement of software products and applications within the construction industry. Glodon is committed to an "All in AI" approach to facilitate the AI-driven transformation of software products throughout the entire lifecycle of construction, including AI intelligent design, AI calculations, AI project progress tracking, AI material management, AI transactions, AI safety, and AI education.

"By using AI technology to quickly generate renderings, we have achieved impressive results at high speed. AI has thus opened a new chapter in the digitization of the construction industry, bringing us numerous new possibilities. What once seemed distant and confined to the realm of imagination has now become reality. The rapid advancement of AI has brought this future much closer than we anticipated," said Yuan in his speech.

"I think the era of (AI) has just begun, and the potential is huge. Each company can do what they are good at, and we are doing what Glodon does best.

Our strength lies in deeply integrating our software, applications, products, and data. We don't venture into completely new territories; instead, we continually refine and enhance our existing products. It's important not to pursue projects outside our expertise solely because of AI, as it may not be worth the effort.

From our perspective, we will continue to immerse ourselves deeply in the industry. We aim to be thorough and specialized because doing only general things would not establish a significant competitive edge. If other companies can quickly develop the same technology, it means there is no real barrier to entry, and our competitiveness would be insufficient.

Therefore, in our opinion, startups aiming to leverage AI should first identify their unique advantages. Every company, including startups, needs to understand their strengths and focus on a specific field, developing their own distinct characteristics, rather than solely pursuing quick success and rapid monetization.

Secondly, this industry still has huge potential, with unique and innovative areas and many specific scenarios, such as decoration. When AI is integrated with these products and applications, significant value can be achieved if we remain committed to thorough exploration. AI needs to be continuously applied to achieve advancement; without application, there is no data, making iteration difficult. This is why OpenAI can develop and refine its entire process: through continuous iteration, it becomes better and better. If you don't iterate and create a product that is widely, deeply, and extensively used, your AI cannot improve.

We also see many newly established AI companies in this industry, some of which face heavy pressure just a few months after their inception. Therefore, we are willing to open up many of our capabilities. We understand that Guanglian Da has not accumulated expertise in these fields, but our strength lies in technology. By sharing our technology, we can collaborate with partners to achieve common success. We believe there are many opportunities in each segment, and start-ups should begin with small projects and gradually grow. Starting with a large company and a major project, such as a large-scale industry model, requires a significantly different level of investment." Yuan told TMTPost.

Yuan  emphasized that Each company should focus on its strengths, and what Glodon excels at is the deep integration of professional applications and industry data. Once the large model is introduced, it addresses problems more effectively, including those in construction stages, safety management, progress tracking, and transaction processes.

"We don't see ourselves merely as a software company; instead, we offer comprehensive solutions to our customers. We provide the industry with complete digital solutions to help companies overcome challenges, which we consider our responsibility. By adopting a holistic approach, we assist companies in navigating difficult times and enhancing their competitiveness." Yuan said.

In the end of his speech, Yuan emphasized that the essence of digitization lies in data, connectivity, and algorithms, aiming to genuinely assist every worker and manager. He underscored the importance of gradually implementing the "point, line, surface, body" structure within the entire construction company and integrating it more effectively with the business operations.

"We anticipate not only technological transformation but also ideological and business transformation in our digital journey. Despite the complexity of digitization, as long as individuals and companies pursue it diligently, we are confident that the construction industry's digitization will significantly improve within the next five to ten years," Yuan confessed.