AI Assistant

Feature Description

Artificial intelligence technology is deeply integrating into various workflows, becoming a new engine for efficiency and intelligence, and is widely used in intelligent office, development, services, and creative assistance. Geographic Information System (GIS) software is often complex and has a high learning curve, creating an urgent need for AI to bring innovation in interaction methods and efficiency improvements.

The AI Assistant feature launched by SuperMap iDesktopX, based on advanced large language models, aims to lower the barrier to use and enhance operational efficiency through natural language interaction, automated processes, and intelligent guidance, ushering in a new era of intelligent interaction.

The AI Assistant provides intelligent support for different task types. The current version offers three core functionalities:

  • Intelligent question and answer: Focuses on Q&A and executing operational commands;
  • Intelligent mapping: Allows users to describe mapping goals in natural language to modify map properties, modify layer styles, and create and modify thematic maps;
  • Intelligent modeling: Understands user intent, automatically filters tools, pre-fills parameters, and intelligently constructs processing and analysis models.

Feature Requirements and Conditions

  • Network Requirement: Requires internet access to connect to AI model services.
  • Permission Requirement: Using third-party models requires a valid API key.
  • Environment Configuration: To use the AI Assistant feature, you need to download and configure the JDK 17 environment. For details, please refer to: AI Assistant Environment Configuration.
  • After configuring the environment, you need to start SuperMap iDesktopX_AI.exe or startup_AI.bat.

Supported starting from SuperMap iDesktopX 2025 version.

Function Entry

  • View Tab -> View Group -> Check AI Assistant (After checking, the AI Assistant panel will be displayed on the right side of the interface)

Operation Steps

1. In the AI Assistant panel, select a task type. Currently, three task types are provided: Intelligent question and answer, Intelligent mapping, and Intelligent modeling. Choose the appropriate  task type based on your needs. 

Task Type Capability Description Example Questions
Intelligent question and answer
  • Performs Q&A based on the native large language model.
  • Supports executing common operations: including opening workspace, opening map, and importing SHP files.
  • Open the ../specific path/world.smwu workspace.
  • Open the world map.
  • Import the ../specific path/Point.shp file.
Intelligent mapping
  • Modifies map property parameters via natural language, supporting changes to map background and scale.
  • Modifies layer styles via natural language.
  • Creates and modifies thematic maps via natural language, including: unique value thematic maps, label thematic maps, graduated color thematic maps, statistical thematic maps, graduated symbol thematic maps, dot density thematic maps, etc.
  • Change the map background to white.
  • Change the foreground color of the current layer to blue.
  • Create a label thematic map for the current text layer.
Intelligent modeling
  • Screens modeling tools and constructs models based on business intent, supporting parameter pre-filling.
  • How to quickly build a GPA model?
  • How to generate a cultivated land slope classification map using a GPA model?
  • How to repair minor overlaps between land polygons?

 2. Initiate a session:  Click the " Initiate a session" button to pop up a new conversation window.  Alternatively, double-click a task type directly to enter its corresponding  session interface.      The session interface will display example questions relevant to the selected task type.     

3. Model switching: Click the model name drop-down menu to switch the currently used model. If no model or key has been set, the AI Assistant function will not work properly.

4. Click the Model Settings button to enter the configuration interface, which supports adding, modifying, and deleting models.

  • Model Service Name: Currently supported service providers include: DeepSeek, Tong Yi Qian Wen, Tencent Hun Yuan, OpenAI, Anthropic, Google Gemini, Ollama (for locally deployed models). Select the corresponding service provider from the list.
  • Model Name: The specific model name corresponding to the selected service provider. For example, for DeepSeek service, you can enter deepseek-chat.
  • Server Address: The service URL address for the model API.
  • Key: The access key obtained from the model service platform.
Model Service Name Example Model Name Server Address Note
DeepSeek
  • deepseek-chat (The deepseek-chat model points to DeepSeek-V3-0324)
  • https://api.deepseek.com
  • https://api.deepseek.com/v1
 For calling DeepSeek model services, please refer to the official documentation.
Tong Yi Qian Wen

Supports the qwen-max series and qwen-plus series. Example model names:

  • qwen-max-latest
  • qwen-max
  • qwen-plus
  • qwen-plus-latest
  • https://dashscope.aliyuncs.com/compatible-mode/v1
 For calling Tong Yi Qian Wen model services, please refer to the official documentation.
Tencent Hun Yuan
  • hunyuan-t1-latest
  • hunyuan-t1-20250529
  • https://api.hunyuan.cloud.tencent.com/v1
 For calling Tencent Hun Yuan model services, please refer to the official documentation.

5. Click the Task switching button, to switch between Intelligent question and answer, Intelligent mapping, and Intelligent modeling task types. Switching task types will create a new session.

  • Intelligent mapping
    • Intelligent mapping involves calling related mapping tools based on the selected layer. Therefore, before sending a mapping request, you need to select the target layer in the layer manager.
    • Enter mapping requirements in the input box. The AI Assistant can automatically match the mapping task type, call the mapping tools, and complete the layer modification based on intent understanding.
    • Before the AI Assistant completes the mapping operation and returns the result, please do not switch the currently selected target layer in the layer manager, otherwise, it may cause the modification operation to fail or be applied to the wrong layer.

  • Intelligent modeling
    • By describing business scenario needs for modeling in natural language, the AI Assistant can understand business goals based on semantics, reducing reliance on expert experience.
    • The AI Assistant will output the modeling thought process in the conversation panel, including tool screening, model construction, parameter pre-filling, and process validation, ultimately building a GPA model.
    • Click Add tool to model window, and a new model window will automatically open to view the generated model.
    • The AI Assistant will pre-fill some parameters based on the modeling requirements. Parameters that are not filled need to be manually completed by the user before clicking Execute.
    • To improve the accuracy of model construction, it is recommended to describe the required data processing operations as clearly as possible in the description. Clearer and more accurate instructions will lead to models that better meet business needs.

6. Input a question and send:

  • Type your question in the input box according to the selected task type.

  • Click the Send button to initiate the AI response.

  • If you wish to interrupt generation while sending, click the Stop the build button.

7. Click the History button, to pop up the History panel. It supports auto-saving history, allowing you to easily view and review the complete conversation process at any time, and continue previous workflows without repeating questions, thereby improving efficiency.

  • Historical Conversation List: Conversation records are displayed in reverse chronological order, with the most recent one on the first line. Each task displays the following:

    • Conversation Title: A title automatically generated by the large language model based on the question, supporting renaming to modify the title.
    • Conversation Time: The specific time of the conversation, including year, month, day, and time, e.g., 2025/07/30 17:00.
  • Click on a specific historical mission, and the AI Assistant panel will display the historical Q&A information for the current task. The history session panel will disappear.

  • Hover the mouse over the More button, click to pop up the operation list, supporting editing the current conversation title and deleting the current conversation record.

  • Supports using the search toolbar to search for historical missions. Also supports clicking the Clear button to clear all historical conversations at once.

8. Click the Create a new session button to clear the panel, end the current session, and start a new session.

9. Click the Home button in the upper-right corner of the panel to return to the AI Assistant homepage.

Notice:

  • Select the corresponding task type (Intelligent question and answer, Intelligent mapping, Intelligent modeling) to ask questions based on your specific task objectives to obtain the most accurate and effective response. Cross-type questioning may lead to unexpected results or failed responses.

  • The API address and access key for the model service must be correctly configured; otherwise, the AI Assistant function will not work properly.

Related Topics

AI Assistant Environment Configuration