Company News

On November 5, the award ceremony of the 2026 "Straight to Wuzhen" Global Internet Competition was held at the Wuzhen Internet International Convention Center. Gtrontec's "Autonomous Decision-Making Silicon-Based Employees for Manufacturing Scenarios Based on Octopus Brain" project won the second prize in the AI + Manufacturing track. This year's competition attracted 1,347 projects and OPC registrations from 31 countries and regions, setting a new record for participation scale.

As an important event of the World Internet Conference Wuzhen Summit, this competition set up the AI + Manufacturing track for the first time, focusing on technological innovation and industrial application. After years of digitalization, advanced manufacturing enterprises have accumulated massive industrial data, but cross-system analysis and complex exception handling still rely on engineers' experience. How to enable AI to make reliable decisions under strict production constraints has become a key issue for the large-scale implementation of industrial AI.

In the finals defense on November 3, Yang Li, Marketing Director of Gtrontec, systematically demonstrated the operating mechanism of the autonomous decision-making silicon-based employees driven by Octopus Brain. The platform integrates industrial data, business rules, and expert experience. Events trigger the silicon-based employees to obtain data across MES, EAS, SPC and other systems, identify root causes, form solutions, and, according to authorization, coordinate with business systems for handling, creating a closed loop of "perception—analysis and decision-making—execution."

The stringent requirements for reliability and determinism in industrial scenarios constitute an important technical threshold for AI to enter core production links. Yang Li introduced that the project has created an integrated mechanism of multi-source data governance + collaboration between large and small models + structured accumulation of industrial skills. Among them, large models handle semantic understanding, task decomposition, and scheduling; small models combine industrial mechanisms to complete professional analysis; and industrial software ensures deterministic execution. At the same time, implicit knowledge such as engineers' experience is further encapsulated into callable Skills, combined with manual confirmation, permission control, sandbox isolation, and security auditing to ensure that decisions are evidence-based, execution is controllable, and processes are traceable.

Quantitative results on real production lines provide strong validation of the technical capabilities. In semiconductor production exception management, the closure time for important exceptions was shortened from several hours to half an hour, and root cause localization efficiency increased by 90%; in semiconductor crystal growth quality analysis, efficiency increased by 60% and the exception rate decreased by 10%; at a leading lithium battery factory, energy consumption prediction accuracy reached over 97%, enabling second-level instruction issuance. Regarding commercialization, Gtrontec uses a standardized platform, scenario-based silicon-based employees, and an FDE expert co-creation model to distill models, Skills, and implementation experience validated at benchmark factories into reusable capabilities. To date, the company has completed the implementation of more than 300 industrial AI projects.

It is worth noting that the technical system of autonomous decision-making silicon-based employees driven by Octopus Brain has won three competition honors this year. Related homologous achievements previously won the second prize in the Hubei Division and the third prize in the national finals of the 2026 "Data Elements ×" Competition. Winning the second prize again in the "Straight to Wuzhen" Competition further demonstrates its comprehensive strength in unlocking the value of industrial data, manufacturing scenario applications, and industrialization.

The "AI + Manufacturing" Special Action Implementation Opinions proposes launching 1,000 high-level industrial intelligent agents by 2027. Transforming industrial software, on-site equipment, and engineering experience into trustworthy intelligent decision-making capabilities and achieving large-scale replication across scenarios and factories is becoming the key to competition in the industrial AI industry. Relying on continuous deep cultivation of advanced manufacturing scenarios, Getrontec provides an important practical sample for building replicable autonomous decision-making capabilities for China's industrial AI.