DataAgent (OnPremise)
DataAgent (OnPremise)
- Documentation
DataAgent (OnPremise)Intelligent Analysis Agent User GuideCreation and Configuration for Intelligent Analysis AgentOverview of the Creation and Configuration Process
Copy page
Download PDFCopy pageDownload PDFThe Intelligent Analysis Agent supports two application types: web version and Lark version. There are some differences in the creation and configuration processes. This article provides a brief overview of the operation procedures and guidance links for both types. You can review this article before proceeding to quickly understand the process and key configuration points.
Currently, the Intelligent Analysis Agent supports two versions: web version and Lark version. The comparison is as follows.
Comparison items | Web version | Lark version |
|---|---|---|
Main differences | After creating the Agent and configuring the dataset and other related settings, you can obtain the web version usage link and open the URL directly in your browser. | Interaction is based on the Lark bot, so you first need to create a Lark bot for the Agent, and then associate the Lark bot with the Agent; after configuration, the Lark version Agent can also be used directly on the web version. |
Usage process |
Caution and prerequisites
- Before creating and configuring an agent, you must first complete the purchase and activation of the Intelligent Analysis Agent, create a project, and prepare data.
- For detailed instructions on creating a project, see: Project Information, User Management.
- For instructions on preparing data, see: Data Preparation Overview.
- Only Intelligent Analysis Agent administrator users have permission to create, configure, and manage agents.
Create and configure the Intelligent Analysis Agent
Operation entry
- Go to the configuration page of the Intelligent Analysis Agent . Then, click "Create Intelligent Agent" in the upper right corner to access the agent configuration page.
- You can click the "Basic Information" tab in the left navigation bar to configure the basic information of the agent, including application name, application description, and administrator. After completing the configuration, click "Save".
- At this point, you have created an agent. Further configuration and publishing are required before it can be used. You can view the agent in the "Pending Publish" tab of the agent list.
Configure the agent
Agents are typically configured by data teams, IT teams, or professional analysts. For most business or management personnel, it is only necessary to use the agent, without needing to perform configuration.
The configuration of agents includes the following scenarios. Details are as follows.
Supported configuration directions | Configuration purpose | Detailed guidance documentation |
|---|---|---|
Basic information | You can set the Agent's avatar, name, and users on the basic information configuration page. | |
Dataset configuration | You can associate relevant datasets for data queries with the intelligent agent on the dataset configuration page, and configure semantic, knowledge, and other settings for fields within the dataset. By further describing dataset fields, adding synonyms, and providing explanations for industry jargon, you enable the large model to more accurately understand dataset fields, improving the intelligent agent's ability to accurately query corresponding data in the dataset based on user questions during subsequent use. | |
Content configuration | On the content configuration page, you can configure the content layout and partial result styles that users will see on the intelligent agent interaction page after the agent is published and launched. For example, you can configure frequently asked questions as recommended questions to be displayed directly on the page, or set a welcome message for users when they first use the intelligent agent. | |
Intent clarification configuration | On the intent clarification configuration page, you can set intent clarification rules for the intelligent agent. These rules enhance the intelligent analysis Agent's interaction capabilities and task execution accuracy when handling complex, ambiguous, or incomplete user inputs. When users enter questions, the intelligent analysis Agent automatically identifies whether the user's intent is clear based on the match between the input and the intent clarification rules, and will prompt for clarification when the question is ambiguous. | |
Prompt configuration | The intelligent analysis Agent system comes with some built-in prompts and also allows you to add custom Prompt based on data query and analysis requirements, providing the large model with certain prerequisites to enable more accurate answers. | |
Audit | The intelligent agent configuration interface includes an audit feature, allowing users to view records of intelligent agent usage and questions asked. | |
Knowledge configuration | You can upload and manage knowledge specific to data query scenarios on the knowledge configuration page to improve the accuracy of data queries. |
Publish and manage agents
After completing the creation and configuration of the Agent, you need to publish the Agent before you can further test and use it; for Agents that are temporarily not in use, you can also take them offline. For detailed guidance on Agent publishing, taking offline, and other management operations, see Publish and Manage Intelligent Analysis agents.
Testing and using the Agent
- Testing the Agent
- Web version
- After logging in to the BytePlus DataAgent-Analysis Agent page with your account and password, click "Agent" to enter the Agent list page.
- Click a published Agent to enter the conversation page and use the Analysis Agent.
- After entering the Agent conversation interface, click the switch button in the dialog box to switch between the "Intelligent Query" and "In-depth Research" features.
- Lark version
For Lark Agents, after publishing, you can search for the Agent name in Lark and click to use it.
- Web version
- Scenario applications
- For Intelligent Query, see Intelligent Data Query Agent.
- For In-depth Research Reports, see Deep Research Agent.
Agent usage optimization and best practices
Case 1: Best practices for Lark version usage
For details, see Lark Platform Agent Configuration and Testing.
Case 2: Agent question suggestions and examples
For details, see Best Practices: Recommendations and Examples for Intelligent Data Query Questions.
