The moment you get lost becomes a learning opportunity. Game-type poker strategy learning app "POKER Q'z" newly releases real-time commentary function (β)

CLOViZ Inc. (Headquarters: Tokyo, Representative Director: Sotaro Masaki), whose philosophy is "Learning while playing," has released the "Real-time Commentary Function (β)" for its game-type poker strategy learning app "POKER Q'z," which allows users to learn how to think about their play in real-time during a match.
Poker is a prime example of an imperfect information game, and instantly calculating the optimal strategy for each situation is a highly technical challenge. This function not only generates strategic judgments in real-time for each situation during a match but also explains the thought process behind those judgments in easy-to-understand natural language, thereby providing a learning experience unparalleled in the industry.

■ Technical Background and Uniqueness
Unlike perfect information games such as chess and Go, poker is an imperfect information game where opponents' hands are not visible. Deriving an optimal strategy when the information required for decision-making is uncertain requires advanced calculations based on game theory.
Conventional poker solvers (GTO strategy calculation tools) are primarily designed to calculate optimal strategies by spending enormous amounts of computation time in advance, making the technical hurdle of outputting solutions in real-time during a match extremely high. Furthermore, existing analysis tools generally have a structure where learning and playing are separated, and a mechanism for automatically recognizing situations during gameplay and providing commentary has not been established. While some tools offer in-game advice, they are limited to presenting expected value figures and recommended actions, and do not have the function to convey the thought process behind judgments in natural language.
The uniqueness of POKER Q'z's real-time commentary function lies in its integrated realization of three technical elements: real-time situation recognition during a match, instant generation of strategic solutions, and explanation of the thought process in natural language. Users can learn the thought process of how strong players organize and make judgments without interrupting their game.

■ Development Background
In recent years, poker learning tools based on GTO (Game Theory Optimal strategy) have become widely popular, but most existing tools primarily present strategies in numerical or tabular formats. Even if an optimal solution is shown, it is not easy to reproduce it in actual play without understanding the underlying thought process.
Another challenge was the separation of learning and practice. In many tools, it was common to research solutions beforehand or review them after playing, making it impossible to learn at the moment of indecision.
POKER Q'z developed this function to address these structural challenges of being "difficult to understand" and "difficult to use in practice," by conveying the thought process rather than just the answers, and providing a real-time learning experience during practice.

■ Overview of the Real-time Commentary Function

The moment you hesitate on an action during a match, simply press a button and POKER Q'z's character "Aria" will present the thought process appropriate for that situation. It supports both AI matches and matches with friends, allowing you to develop your judgment skills in an environment close to actual play.

1. Ask questions the moment you hesitate
Since you can receive advice at the very moment you are unsure, you can absorb ways of thinking in a practical sense, unlike traditional post-play review learning.

2. Supports both AI matches and matches with friends
It can be used not only for practice AI matches but also for human-to-human matches with friends. You can acquire ways of thinking while playing in an environment closer to actual practice.

3. Learn the flow of thought, not just numbers
Aria's commentary is designed to mimic the thought process of how strong players organize and make judgments. It provides a reproducible thinking framework for users, rather than just presenting numerical optimal solutions.

4. Understand "why that play is good" in natural language
By explaining in natural language why a particular play is good, it helps improve adaptability rather than just mere imitation.

■ Comment from Developer/AI Engineer Toshiki Ohata

(Executive Officer, CLOViZ Inc. & Tsuruoka Lab, Graduate School of Information Science and Technology, The University of Tokyo)
In this real-time commentary function, we emphasized how to reproduce the thought process of strong human players, rather than simply returning optimal solutions.
Generating highly accurate solutions in real-time in an imperfect information game is itself a technically difficult challenge, but the technical challenge of this function is to consistently achieve the conversion of those solutions into natural language in a way that explains "why" they are so.
POKER Q'z consistently aims to be an AI that supports user understanding and growth, rather than simply providing a strong AI. We will continue to provide learning experiences where learners can enjoy learning and apply strategies in practice.

■ Research and Development System and Future Outlook
Research and development for POKER Q'z is being promoted primarily by engineers belonging to the Tsuruoka Lab at the University of Tokyo (specializing in game theory and natural language processing). By fusing academic knowledge of game theory with AI technology, we are working to further advance poker strategy and evolve the learning experience. Going forward, we will pursue research and development in the following three areas:

1. Advanced Strategic AI

・Evolution to an AI that can make practical and highly accurate judgments closer to GTO strategy
・Introduction of in-depth commentary function for each play
・Addition of a session analysis function that reviews the entire session and suggests improvements
・Expansion to a highly versatile AI that can handle diverse situations

2. Construction of Diverse Playing Environments 
・Development of AI that reproduces mistakes specific to beginners and certain habits
・Development of AI that teaches optimal strategies against opponents with weaknesses
・Construction of a mechanism to reproduce playing environments for different levels, such as beginner, intermediate, and advanced

3. Personalized Learning Support

・Development of AI that analyzes play history and automatically detects tendencies for mistakes and weaknesses
・Practice problems tailored to weaknesses
・Automatic suggestion of learning content
・Construction of a learning navigation function that optimizes the growth of each individual

Overview of the Poker Learning App "POKER Q'z"

POKER Q'z conducts research and development on competitive poker strategy. It provides a poker learning app that allows users to efficiently learn strategies in their spare time. It includes modes for learning both ring games and tournaments. The goal is to be the closest partner for "those who want to improve their poker skills" worldwide.

Download herehere

X ishere

We are also recruiting for team and internship applications!

If you are interested, please contact us at the email address below:

info@cloviz.co.jp

CLOViZ Inc.

Representative Director: Sotaro Masaki

Capital: 43 million yen

Business content: Research on competitive poker and development of game-type learning apps

■ Contact for this matter

CLOViZ Inc.

info@cloviz.co.jp