Yicai: Tezign Technology's Enterprise-Level Intelligent Agents Solve Long-Term Problems for Businesses, Self-Developed Subjective World Model Enables AI to Truly Understand Humans

Below is the original text from the Yicai interview:

Enterprise-level intelligent agent systems have become the focus of enhancing quality and efficiency in the real economy.

As a benchmark enterprise in Shanghai's artificial intelligence industry, Tezign Technology showcased its self-developed enterprise intelligent agent architecture GEA (Generative Enterprise Agent) under the theme "Digital Forest," focusing on large-scale implementation in core business scenarios such as product innovation and brand marketing, exploring a path to transform enterprise data assets into AI's core productivity.

"There are many procedural tasks within enterprises, long processes, such as new product development and product marketing, which actually involve multiple departments and long durations. So in this process, we call it AI-native workflows to transform these processes, making them more suitable for development in the AI era," said Fan Ling, a professor at Tongji University, director of the Design Artificial Intelligence Laboratory, and founder of Tezign Technology.

By empowering enterprises' long-process work, this self-developed enterprise-level intelligent agent architecture GEA is attempting to address the industry pain points of how large models can further streamline business flows in real enterprise contexts.

It is worth noting that Tezign showcased its self-developed Subjective World Model technology breakthrough at this conference, enabling AI to understand and simulate the decision-making processes of real consumers, achieving a behavior simulation accuracy rate of 85%. This means that for the first time, enterprises can use AI to predict "why consumers make such choices."

"We actually helped enterprises create a system called context, which uses large language models and what we refer to as 'ontology' to understand the 'various relationships' within the enterprise. So on one side, there is an understanding of users externally, and on the other side, there is an understanding of the enterprise internally, helping enterprises innovate new products more efficiently, and secondly, helping enterprises market and sell more efficiently."

It is understood that the orchestration layer of the GEA architecture is driven by Tezign's self-developed Creative Reasoning Model, which is the first large model registered in the field of divergent reasoning in China. This orchestration layer can coordinate over 30 foundational models and call upon more than 400 modular skills.

Currently, this architecture has served over 180 enterprise clients, including more than 60 Fortune Global 500 and China 500 companies, deployed and operating in over 50 countries and regions worldwide, with a monthly resource deployment volume exceeding 10 billion.