G-MATRIX

A generative AI solution for data analysis

G-MATRIX is a generative AI solution for data analysis that queries, analyzes, and makes predictions.

Data analysis is a crucial aspect of modern business, but building a business system requires a lot of time and effort. G-MATRIX, on the other hand, leverages ‘Generative AI’ and ‘Low Code’ technologies to simplify the process of identifying and analyzing data without the need for expertise in query-based data extraction or coding for business system development. If you ask questions in natural language, you can retrieve data from the DB, analyze it, and visualize the results.

G-MATRIX Usage Example

When a user asks a question, G-MATRIX recognizes the natural language, automatically generates a query for data retrieving, develops a screen required for results, and provides answers with numbers, text, pivot tables, visualized screens, and more.

Integration of various scattered DB

G-MATRIX automatically determines which DB contains the desired data, supporting various DBs.

SAP HANA, Oracle, MySQL, MS SQL, IBM DB2…
Legacy DB, DW, DM…

Supports various LLM (Large Language Model)

G-MATRIX supports various LLM tailored to customer needs.

Both cloud LLM and on-premise LLM

* What is LLM?

LLM stands for ‘Large Language Model’. Learning large amounts of text data, understanding human language (natural language), generating sentences in natural language. ChatGPT developed by OpenAI is an example of LLM.

Integration with Business Intelligence functions

G-MATRIX provides BI (Business Intelligence) functions that allow not only data querying but also additional data analysis.

Presenting data results in pivot tables (Pivot, Filter, Sorting, etc.)
Presenting data results in statistical tables, various charts, and graphs
Applying these results to your business system screens (supporting the development of low code business system without coding)

Offering a variety of expansion features

Saving results data in various formats (Excel, PDF etc.)
Sharing or viewing result screens via e-mail
Repetitive tasks can be scheduled

Security

Enhancing security through ‘Private deployment AI services’ utilizing an on-premise LLM
Company DB is not shared with LLM
User, department, and group-specific data authorization settings are possible
(Unauthorized data is masked before being provided, allowing comprehensive data analysis without detailed data access)

G-MATRIX leverages AI technology to understand the meaning of words within sentences, considering the overall context. It automatically retrieves the required data from the company DB to provide users with the desired answers. This section explains how G-MATRIX understands natural language and queries data from the DB.

Definition of Meta

G-MATRIX logically structures and defines DB data to create a meta for analysis.

* META of G-MATRIX

Specifies relationships between DB objects (tables, views, etc.)

Implements a folder-based classification system (Dimension, Attribute, Measure) to help users easily find the business meaning of entered words

Teaching meta-information

AI Server of G-MATRIX teaches meta-information to LLM.

Understanding natural language

LLM analyzes the user’s natural language and transforms it into a structure that the meta engine can understand, based on what It learns with the AI server of G-MATRIX.

The keyword used to question doesn’t have to be same as the word of DB. Even if you ask questions as products, product names, goods, packages, etc., G-MATRIX generates query statements to the fields of DB.

Typos are also recognized. Even if the user accidentally enters ‘Anternet’ incorrectly, it is recognized as ‘Internet’ throughout the entire meaning of the sentence.

Even if there is no ‘quarter’ data in the DB, G-MATRIX understands the word ‘the first quarter’ on its own and expresses the value of the first quarter by adding the values of the DB’s January, February, and March data.

Automatically creating query statements for data retrieval

Based on what LLM understands, the AI engine of G-MATRIX generates query statements to extract data from the DB.

G-MATRIX comprehends sentence meaning and automatically sets up search conditions as ‘Where’ or ‘Select’ based on the sentence’s context.

Creating result screens

It generates results in various forms such as numbers, text, pivot tables, charts, etc., based on the user’s question.