Baidu launched GenFlow 4.0 at its AI Day event on April 27, turning its cloud storage platform into what it calls the first fully functional “AI workspace” jointly developed by Baidu Wenku and Baidu Drive. The headline feature: deep integration with OpenClaw, enabling users to deploy agents with one click from Baidu Drive on both PC and mobile devices.

Scale Numbers

The GenFlow agent series now exceeds 100 million monthly active users and processes 200 million tasks per month, according to AIBase. The platform has attracted over 240,000 developers across more than 6,000 enterprises. Those are not pilot numbers. That is production-scale agent adoption in China’s productivity software market.

What Changed in 4.0

The upgrade centers on three areas. First, a revamped Office Agent that supports parallel invocation of PowerPoint, Excel, and Word agents from a single prompt. Users can trigger multimodal content generation and element-level editing across all three applications simultaneously, according to Insider Monkey.

Second, a new “Memory Center” and “Intention Architecture” that enable GenFlow to handle autonomous thinking and task planning across an entire workflow. Baidu is positioning the shift explicitly: GenFlow 4.0 moves from passive tool to active “digital employee” that plans and executes multi-step tasks without continuous prompting.

Third, a “Team Space” feature in Baidu Drive that lets agents inherit organizational permissions and collaboration history natively. This addresses a real pain point in enterprise agent deployment: agents that operate in isolation from the team’s existing access controls and shared context.

OpenClaw as Distribution Channel

The OpenClaw integration is the most strategically interesting piece. GenFlow 4.0 is “deeply compatible” with OpenClaw and other open-source agent frameworks, per AIBase. One-click deployment from Baidu Drive means OpenClaw agents gain distribution through a productivity platform that already has 100 million monthly active users. That is a go-to-market channel that bypasses the typical developer-first adoption path and puts agents directly in front of mainstream users managing documents, spreadsheets, and cloud storage.

This mirrors a pattern emerging across major cloud vendors: embedding agent deployment into existing productivity surfaces rather than requiring users to adopt new standalone platforms. Microsoft is doing it with Agent 365 and Copilot. Google is doing it with the Gemini Enterprise Agent Platform. Baidu is now doing it through Drive and Wenku.

Coming Next

Baidu plans to release a video editing agent and a team version of its “Agent Collaboration Group” in May. The collaboration group concept suggests multi-agent orchestration is moving from developer tooling into mainstream productivity software, where teams of agents coordinate on shared projects within a single workspace.

Baidu’s full AI stack, spanning cloud infrastructure, the PaddlePaddle deep learning framework, ERNIE foundation models, and consumer applications, gives it vertical integration that most Western agent platforms lack. Whether that integration advantage translates outside China’s market remains an open question, but within it, the 100 million MAU figure suggests the personal AI workspace category is already at scale.