Integrate TotalAgility with AI Knowledge Base
Use the following procedure to integrate TotalAgility with AI Knowledge Base. You can add, modify, or delete the AI knowledge base integration.
-
Navigate to
.
The AI Knowledge Base page appears. By default, the Tungsten knowledge base provider is displayed.
-
Click
New to configure a new provider.
The New AI Knowledge Base configuration dialog box is displayed.
-
Configure the following
General properties.
Do not include the API version in the URLs, as TotalAgility handles this internally.
Property
Description
ID
An ID to identify the AI knowledge base integration.
Display name
Name of the AI knowledge base configuration.
Description
Optional. A brief description or additional information about the AI knowledge base Integration.
Document Chunking
Azure AI Document Intelligence URL
The URL contains the document intelligence model to use. Azure Document Intelligence service intelligently converts documents from customer cases into chunks/sections.
Azure AI Document Intelligence API key
The API key required for accessing the Azure Document Intelligence service. This key is encrypted to ensure security.
Storage and Search
Azure AI Search URL
The URL of the AI Search service that will be integrated.
Azure AI Search API key
The API key for accessing the AI Search service. This key is encrypted to ensure security.
Embedding URL
The URL in the Document Chunk / Vector Storage and Search settings.
For example, https://kfx-pd-kta.openai.azure.com/openai/deployments/TextEmbeddings/embeddingsEmbedding API key
The API key in the Document Chunk / Vector Storage and Search settings for accessing the embedding model. This key is encrypted to ensure security.
Indexes
Create indexes by defining dimensions and adding index fields.
Dimensions: Different models may support different dimensions (numeric value) for optimal performance. Therefore, define dimensions for the embedding model you intend to use. Use the default value or provide an inline value for dimensions. If you specify an inline value, the configured embedded model uses the given index when adding documents to the knowledge base. Once the index is created, you cannot modify the dimension setting.
Index fields: Add index fields as needed by providing a Display name and ID. The index field ID must contain only upper or lowercase letters, digits, or underscores. It cannot start with a number and is limited to 1-100 characters.
Once you save the AI knowledge base configuration, all specified indexes are created automatically in Azure AI Search. Any existing indexes are validated to ensure they contain the same internal fields as required by TotalAgility.
LLM Analysis and Insights
Azure Open AI provider URL
The URL for the Azure Open AI provider.
Azure Open AI provider API key
The API key for the Azure Open AI provider. This key is encrypted to ensure security.
Query type
The query type based on which TotalAgility specifies the default Semantic configuration created previously for the Azure AI Search index. The following query types are available:
-
Simple (default): Represents the simple query parser.
-
Semantic: Represents the semantic query parser for advanced semantic modeling.
-
Vector: Represents vector search over computed data.
-
Vector simple hybrid: Represents a combination of the simple query strategy with vector data.
-
Vector semantic hybrid: Represents a combination of semantic search and vector data querying.
Max number of matches
The maximum number of documents configured for the query. (Default: 5, Minimum: 3, and Maximum: 20)
Strictness
The strictness level for search relevance filtering. A higher strictness means higher precision but lower recall. (Default: 3, Minimum: 1, and Maximum: 5)
Timeout in seconds The timeout period (in seconds) for each provider. (Default: 300, Minimum: 120, and Maximum: 3600)
When you send a request such as text generation or image processing to an AI model, the timeout period ensures the request does not become unresponsive indefinitely. This is specifically helpful to providers who are slower than others and require time adjustment to optimize performance.
For example, if you set a timeout period of 120 seconds for a text generation request, the system waits up to 120 seconds for the AI to generate the response. By then, if the response is not generated, the system terminates the request and returns an error.
Retry count The number of times a system retries a failed operation for each provider before giving up and returning a failure response. (Default: 5, Minimum: 0, and Maximum: 100).
For example, if you set a retry count of 3, the system attempts to resend a failed request up to three times, and if the request continues to fail, it returns an error.
When you use the AI models, the timeout and retry count settings are applied for Copilot, Copilot for extraction, dashboard insights, Copilot insights, and vision usage.
-
- Configure the search access permissions (Full control or Has access) on the Access control tab. The access to search using an Azure AI knowledge base configuration at runtime is controlled by the permissions defined against each AI knowledge base configuration. The AI knowledge base search is based on the security permissions of the current user. By default, everyone has full control.
-
To save the configuration, click
Save.
The configuration is saved and listed on the AI Knowledge Base page. You can update or delete an existing AI knowledge base integration.
Tungsten Knowledge Base provider
Tungsten knowledge base provider is an out-of-the-box AI provider for use with an AI knowledge base. By default, the knowledge base provider, "Tungsten" is available with a license. The Tungsten knowledge base provider configuration is same as the new AI Knowledge Base configuration. However, you cannot delete the Tungsten Knowledge Base provider.
On a clean installation or on upgrading TotalAgility to version 2025.2, the Tungsten knowledge base provider is available by default on the AI Knowledge Base page, with the following settings.
|
Property |
Description |
|---|---|
|
Display name |
Read-only field. Name of the Tungsten knowledge base provider. |
|
Indexes |
Create indexes by adding index fields. Index fields: Add index fields as needed by providing a Display name and ID. The index field ID must contain only upper or lowercase letters, digits, or underscores. It cannot start with a number and is limited to 1-100 characters. If using a shared license, you can only see indexes created on this instance of TotalAgility as these indexes are saved in the database. Once you save the AI knowledge base configuration, all specified indexes are created automatically in Azure AI Search. Any existing indexes are validated to ensure they contain the same internal fields as required by TotalAgility. |
|
Query type |
The query type based on which TotalAgility specifies the default Semantic configuration created previously for the Azure AI Search index. The following query types are available: Simple (default), Semantic, Vector, Vector simple hybrid, and Vector semantic hybrid. |
|
Max number of matches |
The maximum number of documents configured for the query. (Default: 5, Minimum: 3, and Maximum: 20) |
|
Strictness |
The strictness level for search relevance filtering. A higher strictness means higher precision but lower recall. (Default: 3, Minimum: 1, and Maximum: 5) |
| Timeout in seconds |
The timeout period (in seconds) for each provider. (Default: 300, Minimum: 120, and Maximum: 3600) |
| Retry count |
The number of times a system retries a failed operation for each provider before giving up and returning a failure response. (Default: 5, Minimum: 0, and Maximum: 100). |
In a TotalAgility on-premise multi-tenant environment, a tenant can only view indexes created specifically for that tenant, as these are stored in the tenant's dedicated database. However, because the indexing service is shared across all tenants, index names must be unique. If a tenant attempts to create an index with a name that already exists, an error message is displayed indicating the conflict.
View associations
You can view the list of items associated with the Tungsten knowledge base provider.
-
Navigate to the AI knowledge base list page.
The AI Knowledge Base page displays the Tungsten knowledge base provider by default.
-
On the context menu of the
Tungsten
knowledge base provider, click
Associations .
The list of activities using the Tungsten knowledge base provider are displayed.