AI Core Server Installation and Configuration (Versions: 14.2+)

Versions: 8.0+

Overview

QuerySurge AI Core is a dedicated server installation that empowers organizations to securely host and integrate local large language models (LLMs) directly into their QuerySurge environments. By bringing AI capabilities on-premises, QuerySurge AI Core enables advanced natural language interactions, automated test generation, and intelligent data validation — all without relying on external services.

Because it runs entirely inside your network, data never leaves your environment. Sensitive information stays behind your firewall, ensuring that proprietary datasets, test queries, and metadata are fully protected. This architecture gives organizations complete control over their AI infrastructure and compliance posture, while still unlocking the power and flexibility of modern LLMs.

Installation

To get started, download the QuerySurge AI installer specific to your operating system. Be sure to verify you meet our recommended system requirements for your desired configuration.

System Requirements

Processing	Minimum Requirements	Recommended Requirements
CPU	8 cores with 32+ GB of RAM	16 cores with 64+ GB of RAM
GPU (Nvidia Only)	12+ GB of VRAM	24+ GB of VRAM

Please note, your CPU must support AVX and AVX2 instruction sets. The vast majority of modern CPUs have AVX/AVX2 support; however, you may need to explicitly enable this if you're hosting QuerySurge AI Core on a virtual machine.

To verify that you have an available Nvidia GPU with the required drivers correctly installed, run the following command (in PowerShell on Windows or in the terminal on Linux).

nvidia-smi

If your Nvidia GPU is detected and the drivers are correctly installed, you should see something like this.

If you don't see something very similar to the above output, contact your system administrator to verify the availability of an Nvidia GPU and correct installation of the associated drivers.

Installing on Windows

1. Right-click the installer file and choose Run as administrator.
2. Under Setup - QuerySurgeAICore, click the Next > button.
   
3. Read the License Agreement, and if you choose to accept it, click the Next > button.
   
4. Choose an Installation Directory (or accept the default), and click the Next > button.
   
5. Ensure that the AI Core checkbox is selected, and click the Next > button.
   
6. Choose a Port (or accept the default), and click the Next > button.
   
7. QuerySurge AI Core needs to connect to your QuerySurge application server. Enter the Hostname and Port for your application server. If your application server is configured with SSL, select the Use SSL checkbox. Then, click the Next > button.
   
8. Under Ready to Install, click the Next > button.
   
9. First, the installer will download the model file.
10. Then, it'll install the server components.
11. When the installer finishes, click the Finish button.
    

Installing on Linux

Please verify that you're installing QuerySurge AI Core on a supported Linux distribution.

• Red Hat Enterprise Linux (RHEL), versions 8-9
• Fedora Core, versions 36-38

For more details, please review our QuerySurge System Requirements.

1. Verify that the installer file is executable.
   chmod +x querysurge-ai-installer-14.2.20-linux-x64.run
2. Run the installer.
   sudo ./querysurge-ai-installer-14.2.20-linux-x64.run
3. First, you'll be prompted to read the License Agreement. Press Enter to advance to each section of the agreement. When you've reached the end of the agreement, you'll be asked if you accept the agreement. Enter y to indicate your agreement and press Enter.
   Do you accept this license? [y/n]:
4. Choose an installation directory (or accept the default). Enter your desired directory, and press Enter (or, just press Enter to accept the default).
   Installation Directory [/opt/QuerySurgeAICore]:
5. There is only one component (AI Core) to install. Enter Y to confirm and press Enter.
   AI Core [Y/n]:
6. Verify your selections. Enter Y to confirm and press Enter.
   Is the selection above correct? [Y/n]:
7. Install QuerySurge AI Core as a service. Enter Y to confirm and press Enter.
   Install as a service? [Y/n]:
8. Choose a port (or accept the default). Enter your desired port number, and press Enter (or, just press Enter to accept the default).
   Port [80]:
9. Enter your QuerySurge application hostname, and press Enter.
   Hostname []:
10. Enter your QuerySurge application port, and press Enter.
    Port [80]:
11. Enter y if your QuerySurge application server uses SSL (or N if it doesn't), and press Enter.
    Use SSL [y/N]:
12. Enter Y and press Enter to continue.
    Do you want to continue? [Y/n]:

13. Wait for the installer to download the model file and install the service (note, this may take several minutes depending on your connection speed).

    Please wait while Setup installs QuerySurgeAICore on your computer.
    
    Installing...
    0% ______________ 50% ______________ 100%
    ###########################

14. When the installer finishes, you

    Please wait while Setup installs QuerySurgeAICore on your computer.
    
    Installing...
    0% ______________ 50% ______________ 100%
    #########################################
    
    ----------------------------------------------------------------------------
    For help configuring QuerySurgeAICore, please visit the QuerySurge
    Knowledge base.

15. The installer will exit automatically.

Configuration

Everything related to QuerySurge AI Core is easily configurable and customizable through the bundled configuration file. Navigate to your server installation directory and open the core-server.xml file (in the config folder).

By default, it'll look something like this. Remember, your QuerySurge AI Core server settings must match your QuerySurge AI Core application settings.

<config>
   <ai>
      <!-- set to HTTPS if you want the QuerySurge AI Core server to use SSL -->
      <protocol>HTTP</protocol>
      <!-- the port on which the QuerySurge AI Core server will listen -->
      <port>8888</port>
      <model>
         <!-- the directory where your LLMs are located -->
         <path><Installation Directory>/QuerySurgeAICore/ai/models</path>
         <!-- the name (i.e., filename with extension) of the LLM to use -->
         <name>core-l3-8b-INST-Q4-K-M.gguf</name>
      </model>
      <ssl>
         <!-- your keystore file (incl. path, file, and extension) -->
         <keystore>path\to\keystore.jks</keystore>
         <!-- your keystore password -->
         <password>your-password</password>
      </ssl>
      <gpu enabled="off">
         <!-- the directory where the GPU native libraries are located -->
         <binaries><Installation Directory>/QuerySurgeAICore/ai/gpu_lib</binaries>
         <!-- the number of layers to offload to the GPU -->
         <layers>0</layers>
      </gpu>
      <!-- set enabled="on" for enhanced logging -->
      <logs enabled="off" />
   </ai>
   <querysurge>
      <!-- set to HTTPS if your QuerySurge instance has SSL configured -->
      <protocol>HTTP</protocol>
      <!-- the hostname or IP address of your QuerySurge instance -->
      <host>localhost</host>
      <!-- the port on which your QuerySurge instance is running -->
      <port>80</port>
   </querysurge>
</config>

Verify that the default configuration values, which have been defined based on the installation process, are correct and as expected. If not, you can modify them directly in this file. Be sure to stop the QuerySurge AI Core server before making any changes, and restart it when you're finished.

Adding SSL Support

SSL support requires a certificate (either a self-signed or CA-signed one). You can generate your own or use an existing certificate.

To enable SSL support, modify the <ssl> element of the core-server.xml by entering your keystore (path, filename, and extension) and keystore password.

<ssl>
   <!-- your keystore file (incl. path, file, and extension) -->
   <keystore>path\to\keystore.jks</keystore>
   <!-- your keystore password -->
   <password>your-password</password>
</ssl>

Please contact your system administrator for help generating or retrieving the correct certificate and/or keystore file.

Adding GPU Support

GPU support can significantly improve the speed and efficiency with which QuerySurge AI Core can process your requests. If you have an Nvidia GPU at your disposal and would like to use it, there are a few additional prerequisites and configuration steps involved.

First, modify the core-server.xml file to enable GPU processing. Locate the <gpu> element and change its enabled attribute from off to on.

<gpu enabled="on">
   <!-- the directory where the GPU native libraries are located -->
   <binaries><Installation Directory>/QuerySurgeAICore/ai/gpu_lib</binaries>
   <!-- the number of layers to offload to the GPU -->
   <layers>0</layers>
</gpu>

Next, specify a number of layers to offload to the GPU.

<gpu enabled="on">
   <!-- the directory where the GPU native libraries are located -->
   <binaries><Installation Directory>QuerySurgeAICore/ai/gpu_lib</binaries>
   <!-- the number of layers to offload to the GPU -->
   <layers>32</layers>
</gpu>

Note, if you use 999 for layers, QuerySurge AI Core will attempt to offload as many layers as possible. However, for longer chat sessions with increasing context windows, this runs the risk of exceeding your VRAM. If this is an issue or concern, you should choose a set value to offload that meets your performance needs without exceeding available memory.

Starting and Stopping the Server

To start and stop the QuerySurge AI Core server on a Windows installation, use the start.bat and stop.bat files in the ai folder of your installation directory (right-click the file and choose Run as administrator). Alternatively, you can start and stop it like any other Windows service.

For Linux installations, use the sudo systemctl QuerySurgeAICore start and sudo systemctl QuerySurgeAICore stop commands to start and stop the QuerySurge AI Core server.