Baidu has initiated an internal integration of its office AI agent, dodo, with the Baidu Dazi team, as reported on Tuesday morning. The development and resources for dodo will be absorbed into Baidu Dazi, creating a fully unified internal and external office AI agent product. This move signals a shift from a multi-track approach to a centralized strategy for Baidu's office Agent business, as the company consolidates its strengths to compete in the AI office sector.
Dodo is an internal office AI agent used by Baidu employees. It can be tasked via communication tools like Liu, web interfaces, or client apps to organize meetings, process documents, and search emails. It also accesses knowledge bases and development platforms for code submissions and requirement management. Baidu Dazi, launched in March, is a professional office AI agent platform for both personal and enterprise use. Users simply state their work goals, and Baidu Dazi handles task breakdown, data retrieval, document processing, web operations, and result delivery, executing tasks across files and applications continuously.
Product capabilities from Baidu Search, Cloud Drive, Maps, Miaoda, and Famou are integrated via Skill, covering research, content creation, PPT generation, and enterprise collaboration. In July, the enterprise version of Baidu Dazi was released, supporting shared Skills, project contexts, and work outcomes, with unified management of knowledge assets, business processes, and execution. It has been deployed across multiple industries. Since its launch, daily queries on Baidu Dazi have grown 20-fold, and it has topped both the PinchBench and DeepResearchBench AI agent evaluations.
Post-merger, Baidu Dazi will serve as a unified office Agent product for Baidu's internal operations, personal productivity, and enterprise collaboration. Internal practices from dodo in areas like meetings, documents, knowledge bases, development management, and team Skill reuse will be gradually integrated into the enterprise version of Baidu Dazi, enhancing its capabilities in organizational knowledge accumulation, workflow execution, and team collaboration.
Comments