Baidu Wenku Cloud releases GenFlow 3.0, with over 20 million active users, becoming the world’s largest universal intelligent agent

On November 13th, Baidu World 2025 was held in Beijing. At the conference, Baidu Wenku and Baidu Netdisk officially released GenFlow 3.0, with over 20 million active users, making it the “world’s largest universal intelligent agent”. At present, GenFlow 3.0 has been officially launched on both the library and cloud storage platforms, helping users become “super individuals” in their work, study, and life.

The upgrade of GenFlow 3.0 cannot be achieved without the launch of two key intelligent agents. Office Agent is the first intelligent agent that truly understands office suites, enabling GenFlow 3.0 to perform professional “generation+editing” operations for Excel, Word, and PPT with simple commands.

The industry’s first fully modal universal intelligent agent “GenX” launched by Wenku Netdisk unifies the entire process of understanding, generating, and editing graphics, images, and videos, with built-in capabilities for task decomposition, path planning, and workflow orchestration. At present, GenX Agent has been officially launched on Baidu Wenku and Baidu Netdisk, and users can experience it through the GenFlow 3.0 portal.

In addition, GenFlow 3.0 has heavily upgraded its “Memory Center” with stronger global memory capabilities, which can remember users’ past communication information, usage data, and preference needs, and call them up autonomously when needed; At the same time, GenFlow 3.0 fully integrates free canvas, allowing users to easily drag and drop materials, freely doodle, and input commands to generate multimodal content such as documents, posters, videos, and more. It also supports multi person collaborative operations and real-time sharing and editing. GenFlow 3.0 truly achieves all-round collaboration of multi-agent collaboration, multi person team collaboration, and human-machine collaboration.

As the world’s first fully universal intelligent agent, GenFlow3.0 has stronger universal multimodal capabilities, enabling input, processing, and output of all modal content. It adopts a centralized native Multi Agent architecture, constructs contextual engineering based on industry knowledge graphs, and independently develops a “multimodal fusion editor” to create industry-leading characteristic technical capabilities.