Mentat Launches 'Mentat Discovery,' a Book Data Search Service for AI Companies
TECHWORLD ·
✦ AI Summary
Mentat announced on the 13th the launch of "Mentat Discovery (Mentat Discovery Console)." This service provides book data search and contract processing services for AI companies and shows scale and estimated cost based on tokens. Mentat has built searchable book data totaling about 8 billion tokens and provides it in two formats: training datasets and a reference/citation API.
Mentat announced the launch of "Mentat Discovery Console" on the 13th. Mentat is a licensing platform connecting publishers and AI companies, and it has handled functions that support AI training and citation of book content. Founded in 2025, Mentat has signed contracts with more than 90 domestic publishers and secured about 34,000 book titles.
The newly launched "Mentat Discovery" provides AI companies with book data search and contract processing services. Based on the book data Mentat has secured, the service offers functions for book data exploration, evaluation, grouping by category, and contract processing.
Previously, AI companies had to check each publisher's holdings individually and negotiate separately. Mentat streamlined the process by unifying these search and review steps into a single workflow.
The service displays the scale of book data in terms of "tokens" rather than "copies." A "token" is the unit AI uses to process text. It also provides category-by-category data size and estimated cost information so AI companies can gauge the amount of data they need and the cost in advance.
Mentat has built searchable book data totaling about 8 billion tokens, and it organized the data into a searchable format by carrying out Mentat's book data preprocessing, then generating Mentat's metadata, and then classifying the content by category.
AI companies can select the data they need, check the estimated cost of the selected data, and request a detailed book catalog. They then receive the data after completing a contract.
There are two delivery formats: training datasets and a reference/citation API. The training dataset includes bibliographic information such as title, author, publisher, and ISBN, and the delivery format uses file formats for AI development applications, including JSONL. The reference/citation API supports searching relevant book passages when generating AI responses and presenting them together with their sources.
The books available for search are those for which Mentat has completed formal licensing agreements with publishers. As a result, AI companies can immediately review data with confirmed rights relationships and reduce the burden of copyright verification during the process of securing book data.
Mentat CEO Jeon Byung-wook explained that the company presented data based on tokens and cost with the aim of making it easier for AI companies to search and use books. He added that Mentat plans to promote broader AI use of books and expand the structure for returning resulting revenue to publishers and authors as well.
Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408124
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.