Creating a Business Utilization Platform for the Generative AI
and Improving the Accuracy of RAG
Started Providing “Generative AI Operation Quality Improvement Support Service”

Including Support from the Standardization of Data and Business Processes to the Establishment in an Organization,
Achieved Productivity Improvement in Advance Introduction in Amagasaki-shi, Hyogo Prefecture and Major Japanese Companies

SHIFT will begin providing “Generative AI Operation Quality Improvement Support Service,” a solution that provides wide-ranging support from data maintenance to business utilization and organizational establishment of generative AI to increase the accuracy of RAG (Retrieval-Augmented Generation) *.

By combining SHIFT’s expertise in standardization and quality analysis cultivated in quality assurance and software testing with the know-how of improving operations accumulated in customer support (CS), we will strongly support the standardization of data and business processes, development of knowledge, and improvement of output accuracy that are essential for the utilization of generative AI, and realize the establishment of the generative AI in companies and the improvement of productivity through these efforts. It can be used regardless of the language-model of the generative AI introduced.

* RAG is a mechanism to extract highly reliable information by searching for information such as external databases, and to generate responses to LLM (large-scale linguistic models) based on this.

-Background of the Service Launch
The introduction of generative AI by companies and local governments is becoming more common, although the current situation is that it is still difficult to establish generative AI in the organizational system after the introduction. For the introduced generative AI to correspond to the operations and rules specific to the organization and to achieve high-precision output, it is necessary to link data such as business documents to build a RAG. However, there are issues in many organizations, such as the lack of such data, inconsistent formats, and the fact that the generative AI system is difficult to search, which is one reason why the full-fledged use of business for generative AI is not advancing.

To develop the high-quality data required to build a high-precision RAG that contributes to business utilization, it is essential to have in-depth knowledge of AI and an understanding of the operations of the introducing organization.

SHIFT has supported organizations in a variety of industries, primarily in quality assurance and CS, with respect to business standardization, quality analysis, and operational process-improvement. In addition, by internally developing and operating AI products, SHIFT is accumulating knowledge on the operation of generative AI that contributes to the high-precision utilization of its operations and the preparation of the required data. By leveraging the wealth of experience and knowledge that SHIFT has cultivated, it has decided to launch a new provision of “Generative AI Operation Quality Improvement Support Service,” a solution specializing in improvement of building and operating RAG, to promote the use of generative AI in more organizations, thereby contributing to the improvement of productivity and the promotion of DX.

-Support for Improving the Operation Quality of Generative AI
Generative AI Operation Quality Improvement Support is a solution that supports RAG construction planning to improvement in an organization that implements generative AI with SHIFT’s own standardized processes. This will improve the breadth and quality of RAG utilization in the organization by swiftly turning the cycle of building, operating, and improving RAG.

Arrangement of Applicable Business
SHIFT will identify the operations to which the generative AI will be applied through thorough business disassembling. It decomposes jobs into tasks that require human judgment and those that do not and identifies the applicable scope and application method of the generative AI.

Building Environment
SHIFT will develop data linked to generative AI and construct RAG. Drawing on its expertise in documenting in the quality assurance business, it will create structured document data.

Improving RAG and Business Processes
SHIFT evaluates and analyzes the accumulated output from multiple perspectives and proposes effective improvement measures to further improve the accuracy of RAG, such as improving document data and utilizing the accumulated output results. By devising its expertise in improving business processes in CS, it will promote business process transformation by documenting business flow knowledge, which has become implicit knowledge, and expanding data.

Service-related inquiries: https://service.shiftinc.jp/en/contact/

-Examples of Advance Introduction
Establishment and Operation of RAG Specializing in Responding to Inquiries
Based on analyzed 40,000 VOC (Voice of Customer) data from about 110 companies that SHIFT PLUS, SHIFT’s group company that specializes in customer support, owns, SHIFT proceeded with the scrutiny and preparation of email templates that are easy for generative AI to learn and improved the accuracy of RAG. It reduced the number of searches for email templates.
Introduced company: Major Japanese entertainment companies (customer support operations)

Establishment and Operation of RAG specializing in IT Help Desk
By systematically documenting the knowledge of IT help desk and decomposing human tasks to those for AI, SHIFT has improved the accuracy of RAG, improved response, and reduced downtime at sites. Knowledge has also been visualized, which has resulted in the standardization of business processes and the correction of imbalances in internal information.
Introduced company: Major Japanese enterprise companies (Microsoft365 related IT help desk operations)

Establishment and Operation of RAG Specializing in Local Operations
By repeatedly tuning the accuracy of RAG by organizing and analyzing government documents, including collection of rules and regulations data published and municipal meeting minutes, as well as administrative work, SHIFT realized to specialize in and increase the precision of city-specific operations.
Introduced municipality: Amagasaki-shi, Hyogo Prefecture

In addition, there are more examples of advance introduction in various industries. Please contact us for details.

-AI Related Services Provided by SHIFT
SHIFT provides a diverse range of solutions covering every process from strategic formulation of AI utilization to organizational establishment.

AI Specific Quality Assurance
Using its own quality assurance framework and testing standards, SHIFT assist in everything from AI product consulting to test implementation.

TD AI Assistant
Implemented recommendation generation for test cases in TD, the test design tool. It will prevent omissions from the viewpoints and achieve uniformity in the level of descriptions.

Librari AI
Supports the introduction and customization of generative AI that are secure and easy to incorporate into daily operations.

AI Document Reverse
Decodes the black-boxed source code of the legacy system and generates various documents such as high-precision class diagrams and sequence diagrams in natural Japanese.

Tensai-kun
It is an AI platform that enables employees to mass-produce business-specific AI tools that are available at no-prompt without knowing AI.

###

Contact
Motoya Kobayashi
Director
ir_info@shiftinc.jp