At the China AI Agent Conference 2026, Xiong Jibin, CEO of Xunsun Intelligence, presented an AI-ready requirements model and a software factory. These tools enable the generation of complete systems from natural language through multi-agent collaboration.
Compared to current AI coding tools that focus on improving agents for writing code, Xunsun Intelligence has focused on a deeper problem: involving AI in the entire software development lifecycle—from defining requirements to creating a runnable system, not just generating code.
The company has developed two key products: the AI-ready Requirements Modeler and the AI-ready Software Factory. Together, they form an automated process that transforms vague requirements expressed in natural language into deployable software.
The Requirements Modeler is capable of transforming ambiguous descriptions in natural language into clear, unambiguous domain models built using the Domain-Driven Design (DDD) methodology. Xiong provided an example from the financial industry: a client provided over 300 PRD documents in various formats (PDF scans, Word documents, Visio flowcharts) with inconsistent structure, terminology, and detail. The traditional method required two or three senior engineers for two months just to understand these requirements. The AI-powered Requirements Modeler automated this translation, using DDD methodology to identify bounded contexts, entities, value objects, aggregates, and domain events from the natural language input.
The Software Factory takes these structured domain models and breaks them down into development tasks performed by multiple AI agents in parallel. Xunsun's architecture does not view a single agent as a universal coding assistant; instead, it coordinates specialized agents to perform different roles, each responsible for a specific aspect of system creation. This factory approach solves a fundamental gap in modern AI coding: although AI now writes 70-80% of the code in many projects, overall software development efficiency has not increased proportionally because understanding requirements, business modeling, domain design, task decomposition, collaboration orchestration, testing, and deployment still consume most of the project's time and resources.
Xiong emphasized that the bottleneck is not the capability of a single agent, but systematic interaction. Individual agents function as work teams, but orchestrating multiple agents into an automated factory creates a true competitive barrier. Software development is inherently a problem of collaboration, not capability. The shift from code-driven development to requirements-driven development represents a paradigm change where developers describe the necessary functionality, and AI creates the entire system, covering requirements analysis, architecture design, code generation, testing, and deployment.
Xunsun Intelligence's approach reflects a broader industry recognition that the value in AI-assisted software development lies not in accelerating code generation, but in bridging the gap between human intent and machine execution throughout the entire software lifecycle. The company's focus on the DDD methodology as a structural framework ensures that the AI-generated code maintains architectural consistency and alignment with business goals, solving the maintainability and scalability issues that limited AI code generation beyond prototyping and prevented it from moving to production-level enterprise systems.
Youzu Interactive has integrated artificial intelligence agents into 75% of its staff using Tencent Cloud WorkBuddy. This integration allows the game development process to be transformed, from coding to global marketing analysis.
Youzu Interactive, one of China's pioneering foreign publishing companies that has shown international revenue exceeding domestic revenue for the past 11 years, has reached an AI agent penetration level of over 75% of its employees. CTO Wang Junzheng describes this transformation as providing each employee with their own AI agent, while the company covers all computational costs and provides unlimited access to 47 different models.
The company's journey in AI began in 2023 with the use of Stable Diffusion for image generation, and in 2024 with GitHub Copilot for coding assistance. A turning point came in August 2025 when OpenAI demonstrated an agent capable of creating a million-line product from scratch. By early 2026, agent-based tools, including Tencent Cloud WorkBuddy, were implemented across the entire company.
Currently, AI agents operate within three key business scenarios. The first is industry analytics: agents automatically collect information from multiple sources, perform sentiment analysis using large language models, and provide structured results. Youzu developed a bot with instant broadcasting functionality that delivers important industry updates to employees daily, supplemented by a portal for collecting community information.
The second scenario relates to advertising material production. Since advertising materials are the main channel for attracting users for a global gaming company, the previous creative creation process was linear and sequential: trend identification, scriptwriting, storyboarding, modeling, post-production, and editing. A delay at any stage affected the entire cycle. Now, AI covers most of this chain, significantly reducing the time from concept to first draft. However, the final decision on compliance with regional specifics, game genre, and target audience still requires human expertise.
The third scenario is operational data analysis. Although long-running games generate continuous streams of data, obtaining user metrics traditionally required waiting for report templates or coordinating time with analysts. Youzu implemented ChatBI, which allows business teams to ask questions in natural language. Recently, the company packaged the data platform capabilities into tools that business teams can install and run directly within their agents, enabling self-service in data analysis without engineering intervention.
Youzu has completed the transformation of four internal systems using AI, with another seven or eight in progress. The company also uses the WorkBuddy expert group feature, which brings together specialists from different fields into collaborative teams. Wang tested an expert group for content creation that was able to prepare a complete video based on a text prompt in just 20 minutes. According to Youzu's 2025 financial report, revenue from overseas markets grew by 8.06% year-over-year, and the company attributes part of this growth to the integration of AI technologies, which increased gross margin by 2.46 percentage points.
