Recently, Shenzhen Bell Information Technology Co., LTD. (hereinafter referred to as “Using.AI”) completed tens of millions of yuan Pre-A round of financing, investors for Junlian Capital and Jinshajiang joint Capital, this round of financing will be mainly used for the company’s new product research and development and business development.

“Using.AI” is an innovative enterprise that researches and develops intelligent operating systems for industrial and energy AI. It empowers control decisions in the pan-industrial field through AGI technology (Artificial General Intelligence) and promotes the intelligent upgrading of the industry.

AGI spawns a new model of industrial control

In 2017, the “Using.AI” team began scientific research in AGI related fields. According to reports, co-founder and CEO Dr. Crown has been engaged in artificial intelligence study and research in Stanford, Berkeley, Tsinghua and other institutions of higher learning. More than 90% of the company’s team is R & D personnel, and has the research and development and application experience of multiple AI agents, basic models, reinforcement learning, cybernetics and other technologies. The core technology of “Using.AI” is derived from the Berkeley and Stanford artificial intelligence field laboratories.

When it comes to AGI, the most notable product in recent times has been ChatGPT, a large language model trained by OpenAI. Different from the big language model, which is oriented to “decision-making with low precision requirements in open scenarios”, “Using.AI” focuses on “decision-making with high precision requirements in closed scenarios”.

After years of technology accumulation, the company has been able to achieve AI autonomous control and decision-making of complex tasks in vertical scenarios, and finally selected two major industries: industry and new energy. “This is an important application scenario for AGI, and it is also a scenario where the algorithmic, data, and business barriers are very high.” “Using.ai” team said.

The traditional industrial industry has significant pain points such as low production efficiency, unstable product quality, and high operating costs, and AGI provides a new technical path for the intelligent upgrading of the industry, and promotes the evolution of rigid automation to flexible intelligent stage.

The selection of “Using.AI” directly starts from the difficult and high-value upstream core control link, and reshaps the traditional rule-based industrial control decision-making mode with software-defined control decision through the self-developed AI intelligent operating system.

Traditional industrial control systems often use the “soft and hard one” mode, can only do simple motion control programming, but also can only achieve the logic control of standard equipment, can not intelligently meet the needs of the extremely dispersed non-standard scenes of industry, the pain points are significant. In addition, due to the closed ecology of the industry, the traditional industrial control system has low compatibility and poor scalability for industrial data such as production execution systems (MES) and testing equipment.

In view of this, “Using.AI” through the “soft and hard decoupling” way, the hardware resources are configured by software, and the migration decision-making ability of multi-type non-standard devices is realized. “We built a standardized base that is compatible with multiple systems and precipitates the whole process data in different intelligent production links.” The “Using.AI” team says. The AI intelligent operating system developed by the company is specifically divided into simulation class and control class. The simulation class includes virtual measurement, anomaly detection, etc., and the control class includes real-time control and process optimization.

According to reports, in actual customer cases, the product helps enterprises reduce the defect rate by 80%, reduce the personnel cost by 60%, and create an additional annual profit of nearly 1 million yuan for a single line body, which has been verified by the market.

AI intelligent operating system can be both “general” and “intelligent”, it is relying on its core underlying technology – “Meta Learning”. Meta-learning is the use of previous knowledge and experience to guide the learning of new tasks, so that the machine can learn how to learn, evolve, reason, and abstract new applications by itself, so as to have rapid learning and generalization capabilities.

This technology is suitable for small sample learning, breaking the technical bottleneck that AI deep learning requires a large number of labeled data. Based on this underlying technology, “Using.AI” builds the architecture of “meta-base model + industrial vertical large model” to achieve self-supervised learning and low-cost migration under cross-scenario.

The fragmented nature of the industrial and energy scene further highlights this advantage. After years of research and exploration, the “Using.AI” team found that AGI technology based on meta-learning is more suitable for dealing with closed-loop control tasks in closed scenarios, and in industrial and energy fields with complex tasks and closed scenarios, meta-learning and AGI bring higher value.

“Technology + Business” dual flywheel drive

In the field of industry and energy, “Using.AI” further selected 3C, automotive, new energy, semiconductor and other high-boom and high-value segments to cut into, give priority to creating head benchmark customers, and complete the closed loop of the scene from 0 to 1.

Taking “Using.AI” as an example for the intelligent transformation of anodic oxidation production line of a 3C head enterprise, in the process of anodic oxidation, the traditional solutions mostly manually control the dyeing process of products and maintain the state of dye solution, and the quality control is unstable; At the same time, the relevant data of dye solution cannot be collected and analyzed effectively, and the dosage is not accurate, resulting in a large amount of material loss; In addition, the factory’s production process parameter control ability is poor, it is difficult to carry out quantifiable quality analysis and decision-making, and the process ability is insufficient.

In response to the above pain points, “Using.AI” introduces AGI technology to realize the full parameter monitoring, automatic dyeing and dosing execution, intelligent parameter control and other functions, to help customers improve the process level, improve yield and significantly reduce costs. According to reports, after the customer signed the first order, the second purchase was carried out within six months, and it has expanded from a single production line to five production lines.

At present, “Using.AI” has achieved rapid PMF (Product-Market Fit) verification in process scenarios such as surface treatment and processing molding.

The company plans to precipitate standard controller modules, radiate other large and medium-sized enterprises, and achieve large-scale replication and rapid growth; At the same time, the field of multi-equipment is further expanded, and POC (Proof of Concept) is being carried out in pan-new energy manufacturing, pan-semiconductor process and pan-energy optimization scenarios.