Company News

At the end of 2022, ChatGPT swept the globe at lightning speed, officially ushering in the era of large models. As early as the beginning of 2022, TCL began laying out AI large models and, at the WAVE SUMMIT+ 2022 Deep Learning Developer Summit, jointly released with Baidu a large model for the electronics manufacturing industry—TCL-Baidu·Wenxin, optimizing existing quality inspection algorithm models, improving model accuracy, and reducing data requirements.

As an industrial internet platform strategically incubated by TCL, Getech focuses on smart manufacturing technology and integrated solutions for high-end manufacturing, with deep expertise in AI technology applications. Four years ago, it launched the industrial AI visual inspection product—Tianshu GeekVision, widely used in semiconductors, new energy, automotive/3C, and other industries, serving dozens of large high-end manufacturing customers, helping them reduce costs and increase efficiency, and accelerating smart manufacturing industry upgrades.

OctopusGPT Officially Released

Inquiry and practice are inherent cultural genes of Getech, with an unceasing desire for, exploration of, and application of new technologies. With in-depth exploration of large model technology and leveraging TCL's over 40 years of industrial manufacturing experience and data accumulation, Getech officially launches its independently innovated industrial intelligent large model engine base—OctopusGPT.

OctopusGPT is Getech's latest industrial intelligent large model base engine product, designed to help industrial enterprises accelerate the intelligence of production and operations. Inspired by the biological characteristics of the octopus, the model learns and simulates various rules and processes in industrial production to achieve comprehensive optimization and intelligent control of industrial processes.

A Large Model Base with Octopus-like Intelligence

The octopus, commonly known as the "eight-armed fish," has one main brain and eight distributed small brains, with extensive tactile senses and over 500 million brain neurons. By combining with the octopus image, OctopusGPT symbolizes superior intelligence and learning capabilities. In the field of industrial intelligence, the Octopus large model, like the tentacles of an octopus, can keenly perceive data in complex industrial environments, skillfully capture core information, and tightly integrate knowledge with action, bringing more flexible, efficient, and intelligent application solutions to the industrial field.

The OctopusGPT AI base consists of large models (including commercial public/vertical private models), Octopus Intelligent Hub (Octopus X), Octopus Algorithm Model Development (Octopus M), and Octopus AI Integration (Octopus Z). Adhering to Getech's "Industrial Geek" product design philosophy, it shields complex IT technologies, allowing frontline engineers who understand industrial know-how to easily and efficiently DIY innovative applications.

Efficient, Rapid, and All-Encompassing Intelligent Engine

To accelerate the implementation of large model intelligent scenarios, OctopusGPT includes the intelligent hub capability module Octopus X, which integrates advanced large model technology, intelligent orchestration strategies, and multimodal applications, bringing unprecedented flexibility and efficiency to the industrial field.

Through intelligent orchestration, Octopus X tightly integrates model services, business functions, data resources, etc., to meet the needs of different fields and scenarios. Without strong development, users can quickly implement an AI scenario based on large models through configuration, responding in just 4 steps. With ready business scenario data, a scenario can be launched in just 1 day.

OctopusGPT encapsulates existing assets such as APIs, databases, documents, and files in the enterprise into Plugins with business capabilities through configuration in Octopus X, providing public Plugins for data query, chart generation, data analysis and interpretation, task execution, knowledge extraction, form filling, OCR, industrial mechanism models, and more. At the same time, OctopusGPT has integrated the Octopus industrial private domain base large model and multiple commercial base large models, with targeted high-quality Prompts. Octopus X intelligently selects large models and Prompts based on user-set scenarios and Plugins, and also supports user-defined adjustments to flexibly complete scenario orchestration.

Additionally, Octopus X offers two capability invocation modes. The first is the Chat page mode, which can be configured and directly embedded into other application pages; the second is the API mode, which allows capability invocation through RESTful interfaces.

Rich Base Capability Scenario Templates and Plugins

OctopusGPT readily handles various basic intelligent capabilities, such as data query, report generation, data interpretation, task execution, content generation, knowledge Q&A, etc. As enterprises integrate more capabilities and Plugins become richer, the innovative scenario case templates fused by OctopusGPT will become even more diverse.

We've taken the leap with the parachute and will share the thrill of landing with everyone

Currently, OctopusGPT has been piloted in dozens of industrial scenario projects.

For example, Getech's integrated smart control center solution (1+4+1) for a leading semiconductor factory, based on production execution and company operations, establishes 1 data platform, 4 centers (war room center, alarm center, dispatch center, analysis center), and 1 knowledge base to create a high-efficiency, high-reliability intelligent integrated control center. Using OctopusGPT, it achieves intelligent functions such as automatic report generation in the war room, alarm content analysis, intelligent dispatch management闭环, and a knowledge base based on industry large models, making factory operation management more intelligent and efficient, bringing continuous value and competitive advantage to the enterprise.

Another example is integrating OctopusGPT with existing systems to give them a "second life." In the second phase of the equipment health management system project for a leading semiconductor display factory implemented by Getech, scattered equipment-related documents on employee computers, enterprise networks, and various systems—such as异常 reports, operation manuals,异常 records, maintenance records in PPT, PDF, EXCEL, WORD formats—were centrally managed. Combined with a vector database and the OctopusGPT base, an intelligent equipment fault inquiry assistant was built. By providing interfaces integrated with the equipment health management system, when equipment fails, it offers intelligent solutions and helps technicians with knowledge queries, fault localization, and decision guidance on-site, effectively improving fault handling efficiency and creating tens of millions in value for the factory annually.

However, the value of OctopusGPT lies not only in the depth of application innovation but also in the breadth of application coverage. It encourages us to look to the future, using the Octopus industrial base model as a foundation to沉淀 more vertical domain models with our customers and ecosystem partners, building a more complex, cross-domain interconnected industrial intelligent operating system.

In this era full of opportunities, we will together witness OctopusGPT leading a new chapter in industrial intelligent innovation. Let us join hands, harness the power of intelligence, and create a better future.