Smartbi is a significant actor in the AI agent stack, specifically within the Business Intelligence and Data Analysis layer. They are moving beyond simple Natural Language Interfaces (NLI) to ship a full ReAct-based agentic framework called BaiZe. This is relevant to the ecosystem because it demonstrates how legacy enterprise software can be reconstructed around autonomous reasoning rather than just adding a chatbot wrapper.
Their system is noteworthy for its "Skills" architecture, which effectively turns static data assets—reports, OLAP cubes, and Excel templates—into executable tools for LLM-driven agents. For developers and architects in the agent ecosystem, Smartbi represents the convergence of deterministic semantic layers with probabilistic reasoning models. They are pushing forward the concept of multi-agent collaboration in a corporate environment, where agents are given specific roles and governed access to sensitive data sources.
Smartbi, the flagship brand of Guangzhou Smartbi Software Co., Ltd., is a veteran of the Chinese data sector. Founded in 2011 and based in Guangzhou, the company has spent over a decade building a specialized niche around the unique complexities of Chinese enterprise reporting. While global tools often prioritize clean, drag-and-drop visualizations, Smartbi built its early reputation on its "Spreadsheet" product, which allows enterprises to maintain their existing Excel logic and formatting while connecting it to robust, governed data warehouses. This focus on high-fidelity, complex financial and regulatory reports made them a standard in the Chinese banking and securities sectors.
In response to the shift toward generative AI, Smartbi introduced "BaiZe," an enterprise-grade agentic platform. BaiZe is not a simple chatbot; it is built on a ReAct (Reason + Act) loop architecture. When a user asks a business question, the agent performs task decomposition, identifying the necessary metrics, data sources, and analytical tools required to fulfill the request. This system is grounded by a unified metric semantic layer, which ensures that when an agent queries "gross margin," it uses the exact calculation logic defined by the organization's finance department.
Smartbi has developed a "Skills" system that turns existing dashboards, reports, and data mining models into tools that agents can autonomously deploy. The platform supports multi-agent collaboration where different specialized agents—such as a data discovery agent and an attribution agent—work together to diagnose business anomalies. For instance, if a monitoring agent detects an abnormal dip in sales, it can trigger an attribution agent to analyze the core driving factors and output an actionable executive report. This workflow replaces traditional manual digging through dashboards with autonomous insight generation.
The company serves over 6,000 enterprise clients, with a dominant share in the domestic financial sector, including leading institutions like Bank of China and Ping An Bank. Competitively, Smartbi sits between traditional BI vendors and new-age AI startups. They leverage their massive installed base of 5,000+ localized implementations to provide a level of security and "Xinchuan" (IT innovation) compliance that foreign vendors cannot match. By offering a unified ABI (All-in-one Business Intelligence) platform that spans from ETL and data modeling to AI agents, they provide a full-stack solution for organizations that prefer to keep their data architecture within a single, controlled ecosystem. This integrated approach allows them to ground AI agents in historical data and expert knowledge, addressing the common problem of hallucinations in generic large language models.
An AI agent-based decision analysis platform that automatically decomposes tasks and calls analytical tools.
Smartbi is hiring.