【AI前沿】Kunlun Tech Launches TianGong 3.1: Introduces New Canvas Design and Multi-Agent Parallel Workflow

AI NEWSLatest AI NewsArticleKunlun Tech Launches TianGong 3.1: Introduces New Canvas Design and Multi-Agent Parallel WorkflowPublished in Latest AI NewsTime :Jun 17, 2026Read :3minuteKunlun Tech officially launched Skywork 3.1, which is now available in both domestic and international markets.This iteration introduces two core capabilities: Skywork Design, which uses a canvas to support design work, and Dynamic Workflows, which can coordinate multiple Agents in parallel. These features aim to drive AI from the “Demo phase” to a high-completion product stage.Within a month of the release of Skywork 3.0, the Skywork Super Agent achieved a threefold increase in revenue by leveraging high-engagement website scenarios. Data shows that deep users have an average of over 40 interaction rounds per project, indicating a shift from simple Q&A to long-term, complex collaborative workflows.To meet this deeper demand, Skywork Design breaks through the limitations of traditional linear dialog boxes, moving the design process onto an infinitely expandable canvas. It supports scenarios such as full website design, App prototypes, content-to-design conversion, and reference image replication, while ensuring brand consistency across multi-page products through unified design standards.At the same time, for large-scale tasks, Dynamic Workflows can dynamically plan and schedule dozens to hundreds of sub-Agents working in parallel. It introduces a cross-validation mechanism to intercept errors and supports checkpoint resuming to handle large projects spanning multiple days.Currently, these two new capabilities are fully integrated into the Skywork main platform ecosystem, supporting exports as images, Figma layered files, or code compressed packages. The release of Skywork 3.1 marks the acceleration of large model applications from single-text interaction to complex, multi-Agent collaborative graphical workflows, offering a new paradigm for the industry to explore the practical implementation of AI-native applications.