MarkLines Co., Ltd. (Headquarters: Minato-ku, Tokyo; President & CEO: Makoto Sakai; https://www.marklines.com/) today began offering a new feature, the "MarkLines Gen-AI Beta Version," for B2B users involved in the automotive industry.

This AI-powered feature enables users to ask questions in natural language and instantly receive highly reliable, relevant data drawn from our extensive automotive database. In addition, the system analyzes this information by leveraging the rich content across our platform and delivers clear, concise, text-based insights.

By combining trusted data with advanced analytical capabilities through generative AI, this service represents a first-of-its-kind offering in Japan within the B2B space, with only a limited number of comparable examples globally. This feature is available at no additional cost to corporate members of our information platform.

■ Key Features of MarkLines Gen- AI

1. Simultaneous Provision of "Data" and "Analysis" in Search Results
   Generative AI tools that search publicly available information on the web often cannot point to accurate and detailed data, frequently failing to provide satisfactory answers for automotive industry professionals.
   With this new function, the "Data" section introduces the most relevant proprietary data from our content, while the "Analysis" section generates accurate and reliable textual explanations based on cross-referenced search results.
2. Cross-Searching Across a Broad Range of Automotive Information
   Our content spans a wide array of areas: market trends such as sales volumes and model changes; technological trends like ADAS and EV powertrains; supply relationships between automakers and suppliers; and industry trends represented by news and technical reports.
   This feature enables cross-searching across multiple content menus, allowing users to obtain comprehensive, purpose-fit answers in a one-stop service.
3. Queries and Answers Supported in Over 60 Languages
   In addition to English, Japanese, and Chinese, questions can be asked in over 60 other languages.
   Answers will also be provided in the language used for the query.

■ Development Background of MarkLines Gen-AI

Technological innovation driven by generative AI has created an environment where anyone can easily use it to obtain answers.

In response to this shift, we have been developing a generative AI that leverages LLM (Large Language Model) and RAG (Retrieval-Augmented Generation) technologies. It is designed to provide appropriate and well-fitted answers by drawing from our extensive and diverse database.

Specifically, the RAG technology enables high-precision retrieval and extraction of the most relevant information from our database in response to a query. The LLM then performs cross-sectional analysis and summarization of this information, achieving practical answers that consider context, going beyond mere information presentation.

■ Structure of MarkLines Gen-AI Responses

Analysis results are structured into a 'Data Section' and an 'Analysis Section'.

• The Data Section presents the most suitable data from within our content.
• The Analysis Section searches our latest professional information, such as news and reports, and generates accurate, reliable answers in text format.

For example, when a user asks, “Which regions are expected to see growth in the EV market going forward?”, the system uses RAG (Retrieval-Augmented Generation) to extract the most relevant information from MarkLines’ proprietary data. Leveraging MarkLines’ unique datasets—unavailable to general-purpose generative AI—the platform generates a written overview of recent trends based on the latest news and reports, and presents them in a clear and concise format.

By doing so, we aim to provide a generative AI that enables users to extract insights from the automotive industry using natural language, without requiring specialized knowledge or complex search operations.

■ Future Plans

This feature will be launched as a beta version. We intend to continuously improve it based on user feedback and work towards the release of an official version.
In addition, we will enhance the capabilities of "MarkLines Generative AI," aiming to further improve its accuracy and provide more specialized and high-quality answers.

MarkLines Co., Ltd.
Phone: +81-3-4241-3901
Email: ir@marklines.com
Persons in Charge: Kondo, Uki

End