The Moment You're Lost Becomes a Learning Opportunity. Game-Based Poker Strategy Learning App 'POKER Q’z' Unveils Real-time Commentary Feature (β)

CLOViZ Inc. (Headquarters: Tokyo, CEO: Sotaro Masaki), whose philosophy is "Learning while playing," has released the "Real-time Commentary Feature (β)" for its game-based poker strategy learning app "POKER Q’z (Poker Cues)." This feature allows users to learn the thought process behind plays in real-time during matches.
Poker is a prime example of an imperfect information game, and instantly calculating the optimal strategy for each situation is itself a highly advanced technical challenge. This feature not only generates strategic decisions in real-time for each game situation but also explains the thought process behind those decisions in easy-to-understand natural language, offering an unprecedented learning experience in the industry.

■ Technical Background and Uniqueness
Unlike perfect information games such as chess or Go, poker is an imperfect information game where opponents' hands are hidden. Deriving optimal strategies amidst uncertain decision-making information requires advanced calculations based on game theory.
Traditional poker solvers (GTO strategy calculation tools) are typically designed to calculate optimal strategies over extensive pre-computation time, making it extremely difficult to output solutions in real-time during a match. Furthermore, existing analysis tools generally separate learning from gameplay, and mechanisms for automatically recognizing situations during play and providing commentary have not been established. While some tools offer in-game advice, they are limited to presenting expected values and recommended actions, lacking the ability to convey the thought process behind decisions in natural language.
The real-time commentary feature of POKER Q’z is unique in its integrated realization of three technical elements: real-time situation recognition, immediate generation of strategic solutions, and explanation of thought processes in natural language. Users can learn the thought flow of strong players—how they organize situations and make decisions—without interrupting their gameplay.

■ Development Background
In recent years, poker learning tools based on GTO (Game Theory Optimal strategy) have become widespread. However, many existing tools primarily present strategies in numerical or tabular formats. Even if the optimal solution is shown, it is not easy to replicate it in actual play without understanding the underlying reasoning.
The separation of learning and practice was also an issue. Most tools typically allow users to look up solutions beforehand or review them after playing, making it impossible to learn at the very moment of hesitation during a decision.
POKER Q’z developed this feature 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 answer, and providing an experience where users can learn in real-time during practice.

■ Overview of the Real-time Commentary Feature

When you hesitate on an action during a match, simply press a button, and POKER Q’z's character "Aria" will present the thought process for that situation. It supports both AI matches and matches with friends, allowing you to cultivate decision-making skills in an environment close to actual play.

1. Ask Questions the Moment You Hesitate
Receive advice the moment you are unsure about a decision, allowing you to absorb thinking methods in a way that closely resembles real-time gameplay, unlike traditional post-game reviews.

2. Supports Both AI Matches and Matches with Friends
Available not only for practice AI matches but also for head-to-head games with friends. You can acquire thinking methods while playing in an environment that is closer to actual competition.

3. Learn the Thought Flow, Not Just Numbers
Aria's explanations are designed to mimic how strong players organize and make decisions in various situations. Instead of presenting numerical optimal solutions, it provides reproducible thinking patterns for the user.

4. Understand "Why This Play is Good" in Natural Language
By explaining why a particular play is good in natural language, it supports the improvement of application skills rather than mere imitation.

■ Comment from Toshiki Ohtsubo, Developer/AI Engineer

(Executive Officer, CLOViZ Inc. & University of Tokyo Graduate School of Information Science and Technology, Tsuruoka Lab)
For this real-time commentary feature, we focused on how to replicate the thought process of strong human players, rather than just returning optimal solutions.
Generating high-precision solutions in real-time for imperfect information games is technically challenging. However, the real challenge of this feature lies in consistently achieving the conversion of those solutions into natural language explanations of 'why,' all in one go.
POKER Q’z has consistently aimed not just to provide a strong AI, but an AI that supports user understanding and growth. We will continue to provide experiences that allow learners to enjoy learning and apply strategies in practice.

■ Research and Development Structure and Future Outlook
The research and development of POKER Q’z is being driven by engineers primarily belonging to the University of Tokyo's Tsuruoka Lab (Specialty: Game Theory, Natural Language Processing). By integrating academic knowledge of game theory with AI technology, we are working to further advance poker strategies and evolve the learning experience. We will proceed with research and development in the following three areas:

1. Advancement of Strategy AI

・Evolution into an AI capable of practical and highly accurate decisions closer to GTO strategy
・Introduction of a deep-dive commentary function for each play
・Addition of a session analysis function that reviews the entire session and suggests improvements
・Expansion into a highly versatile AI capable of handling diverse situations

2. Construction of Diverse Playing Environments 
・Development of AI that replicates beginner-specific mistakes and particular habits
・Development of AI that guides optimal strategies against opponents with weaknesses
・Establishment of a system to replicate level-specific playing environments for beginners, intermediate, and advanced players.

3. Personalized Learning Support

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

Overview of Poker Learning App "POKER Q’z"

POKER Q’z conducts research and development on strategies for competitive poker. It provides a poker learning app that allows users to efficiently learn strategies during their spare time. It includes modes for learning both ring games and tournaments. Our goal is to become the closest partner for everyone worldwide who wants to improve their poker skills.

Download here:here

X (formerly Twitter):here

We are also recruiting for team internships!

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

info@cloviz.co.jp

CLOViZ Inc.

CEO: Sotaro Masaki

Capital: 43 million yen

Business Activities: Research on competitive poker, development of game-based learning apps

■ Inquiries Regarding This Matter

CLOViZ Inc.

info@cloviz.co.jp