Release Date
Compatible FineBI Version
Component Image and Plugin Version
2026-09-18
V7.0.8 and later versions
FineAI and FineChatBI semantic parsing SLM component image: upgraded to v2026.1.3.0.
Data Agent plugin: upgraded to v2026.1.3.0
After this update, during chats with an agent, the agent will guide you to further clarify your analysis intent through card-based options, thereby enhancing the interactive experience and optimizing the operation path. The following figures show the effects.
During chats with an agent, you can upload local images and select a model with image understanding capabilities for Q&A and analysis.
If only one or more specific rounds of content in a chat are selected for sharing, only the chat products from the selected rounds will be shared. This prevents products in unselected rounds from being shared together, achieving more granular control over product sharing.
In addition, after this update, the Download, Save to Resources, and Share operations for chat products under Agent Supervision> Chat Records > Chat Details will be removed, with only the product preview function kept.
When a dashboard with a related agent published is shared via a preview link on the FineBI platform, a floating agent button will be displayed on the preview page. After clicking this button, you can quickly open the side agent panel for conversational analysis.
For example, after the Chain Operation Analysis Assistant agent in Data Agent is published to the Chain Operation Analysis dashboard on the FineBI platform, you can click the floating button in the preview link for dashboard sharing to open the agent and perform conversational analysis. The following figures show the effects.
After this update, in scenarios such as data Q&A and report generation based on model metric sets, you can further query the extended fields from the model metric sets.
Under Admin Console > Model, you can add a multimodal large language model (LLM). Once the model connection test is complete, this model can be used for the agent's image recognition and image understanding scenarios.
After this update, a permission control has been added for Admin Console > Data, allowing public data and my data to be differentiated by permission.
Developers can create or delete public data only when they are assigned the public management permissions on data, further achieving data isolation and reducing the risk of data leaks.
After this update, the User label has been added for the Data, Knowledge, Skill, and Model pages in the admin console. In this way, super administrators can view and edit data, knowledge, skills, and models by user, improving management efficiency.
Taking User Data as an example, the following figure shows the effect.
After this update, if an agent's visibility modified by a super administrator has exceeded the permission scope of the agent's current editor, the current editor can unpublish the agent, adjust the visibility, and then republish the agent.
(1) The current editors of the agent are its first-order administrators.
(2) The current editors of the agent include super administrators, the agent creator, and agent collaborators (to be introduced in future versions).
A function to preview added data assets has been added to the agent editing page. This allows you to easily view data content immediately after configuring data assets, further enhancing the data adding and previewing experience.
The Disable All button has been added under Agent Supervision > Scheduled Task to allow all scheduled tasks to be quickly disabled. This helps administrators promptly control the execution status of scheduled tasks, strictly manage token consumption, and prevent wasted computing power caused by the continuous execution of invalid tasks.
The Skill and Data options have been added to the management permissions for developers under Admin Console > Permission Management. This is used to control whether developers have the permissions to manage public skills and data, further improving permission assignment.