A domestic AI team has placed its bet on enterprise-level AI in a direction completely different from chatbots. On the evening of July 16, Mingshi Intelligence announced the open-source release of StaffDeck—a platform for building and managing digital employees, with the code hosted at the GitHub repository github.com/OpenBMB/StaffDeck.

Unlike many conversation bots that merely wear an enterprise shell, StaffDeck targets digital labor force. It aims to transform the professional knowledge, standard operating procedures (SOP), and various decision rules within an organization into a group of digital employees that can work continuously, improve constantly, and retain organizational knowledge. According to Mingshi Intelligence, the next step for enterprise-level AI is not a more conversational chat window, but a productivity tool that can accumulate experience and remember organizational rules.

image.png

The platform is supported by a joint team spanning academia and industry. The project was jointly developed by Mingshi Intelligence, Northeastern University, the Mingshi-OpenBMB Data Intelligence Joint Laboratory, the Tsinghua University Natural Language Processing Lab (THUNLP), OpenBMB, and AI9Stars. Multiple institutions have pooled their academic research capabilities and enterprise application experiences, which means StaffDeck is not just an internal tool for a single company, but an open platform tailored for enterprise AI deployment scenarios.

From a technical perspective, the core difference of StaffDeck lies in the continuous retention of knowledge. The awkwardness of traditional chatbots is that once the conversation ends, the context disappears, making it difficult to accumulate the organization's tacit knowledge. StaffDeck aims to translate the professional judgment stored in the minds of veteran employees, written process specifications, and informal decision-making logic into digital employees that can be executed and iterated by the system over time. This means knowledge will no longer be lost due to personnel turnover but will remain within the system.

This open-source release was posted by OpenBMB on a social media platform, along with a complete GitHub repository link. For teams evaluating enterprise AI implementation but unwilling to simply layer another chatbot on top, StaffDeck provides a technical foundation that can be directly examined and further developed. While the industry is still competing on the responsiveness of chatbots, Mingshi Intelligence has chosen a path that focuses more on organizational capability accumulation.

Project address: https://github.com/OpenBMB/StaffDeck