Break through labor bottlenecks, achieve unlimited expansion of testing capacity, and enable 24/7 uninterrupted automated testing.
Accumulate testing experience and expertise into reusable assets, and build an exclusive game knowledge graph.
AI Agents deeply interpret game scenarios, simulate real player behaviors, and support complex task workflows.
Cover the entire process from R&D, testing to launch and operation, realizing a smart quality assurance closed loop.
Based on game design documents and knowledge graphs, LLMs automatically decompose functional points and generate high-coverage test cases.
- Support natural language input for requirements and automatically decompose task operation chains. - Automatically generate test cases for various paths and validate execution results at each step. - Test cases are reusable and editable, with support for data-driven expansion.
Integrates visual recognition with in-game data collection, enabling autonomous adaptation of scripts.
- No hard-coded element positioning required; UI style and resolution changes do not impact test execution. - Built-in intelligent retries and exception handling automatically address loading delays and pop-up interruptions. - Supports DSL instruction-driven execution, with execution efficiency comparable to traditional automation scripts.
Collects full-dimensional test data through video stream collection + game plugin collection.
- Visual Perception: Identifies UI controls, character status, and scene visuals. - Data Perception: Acquires game memory data, performance metrics, and log information. - Behavior Perception: Records keyboard/mouse operations, task progress, and resource changes.
Integrate AI large models and log analysis capabilities to automatically cluster test logs, identify root causes of defects, and generate standardized defect reports ready for direct submission.
- Automatically identify defect types and accurately label them as functional/performance/compatibility/logical defects. - Intelligently associate historical similar defects, automatically flag duplicate tickets, and eliminate duplicate submissions and redundant work. - Generate a visualized reproduction path with one click, including video recordings, step-by-step operation instructions, and data snapshots, to accurately restore the issue scenario.
· Achievement: Accumulate testing knowledge and experience (e.g., defect patterns, test cases) to form structured, reusable assets. · Core Value: Shift testing knowledge and experience from individual-dependent to organizational assets, eliminate redundant work, and build a valuable knowledge repository for future projects.
· Achievement: Support test task execution across multiple machines, accounts, and platforms in parallel; automatically generate test cases, execution paths, and validation rules based on testing requirements. · Core Value: Unlike traditional testing where capacity expansion relies on increased manual effort, AI Agents enable rapid, elastic scaling of testing throughput through automation and parallelization.
· Achievement: Build a structured game knowledge graph from project data, merging static (game rules, workflows, character attributes) and dynamic (player behavior, defect history) information. · Core Value: Leverage knowledge graphs to enhance AI decision accuracy and reliability, providing accurate, traceable answers for users to quickly obtain game information and improve decision efficiency.
