AI Central
AI Central
Breadcrumbs

Basic Agent Creation

Select Agent Type

On the AI Studio page, click "+ Create" in the upper right corner to create a basic agent;

image-20251204-085423.png

Creation Steps

  • Enter the assistant name, select assistant avatar, select model group, select assistant category, add Agent description:

    ① Agent Name: Enter the name of the assistant as its identifier (within 50 characters).

    ② Agent Avatar: Choose one from the system default avatars (custom avatar upload is not supported yet).

    ③ Model Group: Configure a suitable model group for the assistant.

    ④ Agent Category: Select the group(s) where the new assistant belongs (up to 5 categories).

    ⑤ Agent Description: Enter a brief description explaining the assistant's functions and purposes (within 200 characters).

  • Click "Create". After the assistant is created, it will enter the basic orchestration assistant configuration page. Configure and publish it to put it into use.

💡 Tip: The system interface supports the following languages: Simplified Chinese, Traditional Chinese, Japanese, English.

image-20251204-085937.png

Agent Configuration

image-20251204-090846.png

  • There are two ways to enter assistant configuration:

    • Directly enter the assistant configuration page after creating the assistant;

    • Hover the mouse over the assistant card to see the "✏️" icon, click it to enter the configuration page.

    ① Prompt: Enter the assistant prompt, also supports intelligent generation based on existing prompts (prompt limit: 2000 characters).

    ② Prologue: Enter the Agent's opening statement, also supports intelligent generation based on the prompt or existing opening statement (limit: 2000 characters).

    ③ Model Group: Click "+" to add model groups, multiple selectable models are supported.

Note: Model groups need to be added by administrators first in system management by adding multiple different models into the same model group, then configuring the model group to the assistant.

image-20251204-090941.png

Add Model Group

  • Path: Admin → Model → Model Group → Create Model Group (only administrators can add models)

  • Steps to Add:

    • Click "Create Model Group"

    • Complete the following configurations:

      • Enter the model group name

      • Select models to add to the group, multiple selections allowed

      • Choose whether to enable adaptive model deployment

      • Choose whether to enable deep thinking model

    • Click "Save"

  • Adaptive Model Deployment: Automatically adjusts computing resources based on traffic to ensure stable and smooth service;

  • Deep Thinking Model: Intelligently calls more powerful AI for complex problems, significantly improving answer quality.

image-20251204-091057.png
image-20251217-062606.png
image-20251217-062658.png

④ Skills

Click "+" to add one or more skills, recommended skills can also be added (up to 20 skills can be added)

image-20260106-094752.png


There are 3 default skills: Google Search , Text-to-Image , Webpage Reading.

  • Google Search: Obtain real-time, accurate web information through the Google search engine, supporting global webpage content retrieval.

  • Webpage Reading: Extracts webpage text, data, and other content, parsing webpage information.

  • Text-to-Image: Automatically generates corresponding images based on text descriptions, turning textual creativity into visual presentation.

Note: Additional skills can be appended, which requires administrator operation and configuration.

⑤ MCP

The MCP service manages the connection permissions between AI assistants and external tools or data sources within the system.

  • Capability Extension: Enables AI assistants with practical functions such as search, calculation, visualization, etc.

  • Rich Ecosystem: Continuously integrates various tool services to meet diversified needs.

  • Standardized Access: Integrate internal system resources through personal MCP.

Note: When the number of configured tools reaches or exceeds 5, the system will issue a prompt. Too many MCP tools may cause the prompt length to exceed the model's context limit, affecting assistant performance.

image-20251204-091732.png

⑥ Conversation Experience

  • Attached messages count: Set the number of historical conversation turns the assistant can remember, from 1 to 10. It is recommended to set it to 5 to balance conversation coherence and performance.

  • Conversation Settings: You can enable settings such as "User Question Suggestions, Question Guidance, Chat Records, Conversation Feedback, Keyword Review".

    • User Question Suggestions: After the assistant answers, provide some question suggestions to the user based on the previous context.

    • Question Guide: During user-assistant conversations, related question guidance will appear, using model capabilities to infer possible user questions and complete user queries.

    • Chat History: Whether to retain the assistant's chat records. If turned off, chat records cannot be retrieved.

    • User Feedback: Users can like or dislike the assistant's answers to optimize responses.

    • Keyword Filter: At least one of input content review or output content review must be enabled. Once enabled, sensitive word detection will be performed on prompts or AI feedback results, and sensitive words can be maintained in advance.

image-20251204-092025.png

⑦ Knowledge Base

  • Knowledge Source: Click "+" to add knowledge bases (up to 5 knowledge bases can be added as knowledge sources).

    • Allow File Upload:

      • If file upload is enabled, you cannot add knowledge base content as a knowledge source.

      • If file upload is disabled, you can selectively add knowledge bases from personal or enterprise space as knowledge sources.

  • Knowledge Base Configuration: You can modify detailed settings such as "Retrieval Strategy, Private Domain Q&A, Retrieval Method".

    1. Retrieval Strategy: Hybrid Search, Embedding Search, Text Search.

      • Hybrid Search: Combines vector retrieval and full-text retrieval results, returning reranked results.

      • Embedding Search: Finds fragments by similarity, with some cross-language generalization ability.

      • Text Search: Finds fragments by keywords, suitable for retrieval containing specific keywords or noun fragments.

    2. Maximum Recall Count: Range 1–10, not recommended to set too high or too low, recommended value is 3–5.

    3. MetaData: None, Filter, Weight.

    4. Force Private File Q&A: When enabled, network search and other skills will not be used; the assistant's answers only target knowledge base content.

    5. Document Match Similarity: Range 0–1, the higher the similarity, the more similar the recalled document content is. Recommended value is about 0.8 (i.e., 80%).

    6. QnA Match Similarity: Range 0–1, similar to document content similarity matching, recommended value is about 0.9 (i.e., 90%).

    7. Show References: When enabled, the assistant will list referenced documents in answers to improve credibility.

💡 Tip: Whether it is maximum recall count, document matching similarity, or QnA matching similarity, higher or lower is not always better. It is recommended to set according to actual needs. If no special requirements, keep the default values.

image-20251204-092135.png

⑧ Data Source

  • Data Source: Click "+" to add data sources as the assistant's Q&A data sources (up to 5 data sources can be added).

  • Template Enabled: Whether to enable preset mapping templates between natural language and SQL.

    • When a user inputs a natural language question (e.g., "What was the sales last month?"), the system will first try to match a preset template.

    • If a matching template is found (e.g., a general question like "Query sales for a certain period"), the existing SQL structure in the template will be used as a reference, combined with specific fields/table names to generate the final SQL statement.

  • Enable Parameterized Template: When enabled, parameterized queries are enabled on the template basis to enhance query flexibility and security.

  • Question Rewriting: When enabled, the user's input question will be automatically optimized to ensure accurate data queries.

    • Original user question: Check sales (incomplete information).

    • After rewriting: Query total sales of all products in July 2024 (added time and scope).

  • Step-by-Step Reasoning: When this feature is enabled, before generating the final query result, the system will output detailed thinking steps explaining how it analyzes the question and constructs the SQL query.

    • Step 1: Identify keywords "July 2024" and "sales".

    • Step 2: Determine data table Orders, fields order_date and sales_amount.

    • Step 3: Construct date range condition 2024-07-01 to 2024-07-31.

    • Step 4: Generate SQL.

image-20251204-092242.png