Skip to main content

Semantic vector store

Semantic vector store stores meta-data that are essential for most of AI operations. It is used by the following modules/features:

Technical requirements​

  • For ALVAO SaaS subscription:

    • no additional requirements

  • For on-premises setup:

On-premises deployment​

Go to Administration - Settings - Web services and enable the Azure Open AI service.

The Azure AI Search service is required only for on-premises deployments using SQL Server 2022 or earlier.

To configure semantic vector store in such an environment:

  1. Open Azure Portal and create a new Azure AI Search service in your Azure subscription.

  2. Complete the json template below with values from your Azure AI Search service, and then add it to the appsettings.json file of each of the following Alvao applications:

    • Alvao Service
    • Alvao WebApp
    • Alvao REST API
    • Alvao Custom Apps WebService
    "AzureAISearch": {
        "Endpoint": "", // endpoint of the service
        "Key": "", // secret API key
        "IndexName": "" // name of the index to be created in the Azure AI Search service
    },

Settings​

On the Administration - Settings – Advanced page, you may tune the following parameters.

  • Vectors.General.AtHours - hours of the day during which new vectors are calculated.

    Default value: 20, 21, 22, 23, 24, 0, 1, 2, 3, 4, 5, 6

  • Vectors.General.MaxCountPerPeriod - maximum number of records that are processed in a single batch during vector calculation.

    Default value: 500

  • Vectors.Tickets.MaxMonths - vectors are initially calculated only for tickets created within the specified number of months counting back from today.

    Default value: 12

  • Vectors.Tickets.MaxCount - vectors are initially calculated only for the specified number of most recently created tickets.

    Default value: 10000