AiConnection Configuration Item Object
AiConnection is a Common module configuration item object used to connect Lowgile to Large Language Model (LLM) providers.
It stores the runtime configuration used when workflows, services, scripts, and integrations communicate with LLMs.
Configure AiConnection before implementing AI features such as structured extraction, PDF analysis, tool calling, or AI-assisted workflow automation.
Configuring AiConnection
Section titled “Configuring AiConnection”The AiConnection object structure is defined under:
- Common → ConfigurationItem → AiConnection
Runtime values are configured from:
- Configure → Settings
Insight
Configured values are managed in the development environment as part of the application configuration and are compiled into the final application build for production deployment.
To review or update the configured values:
- Open the Lowgile application.
- Select Configure from the left navigation.
- Open Settings.
- Look for Module Common.
- Locate
AiConnection. - Expand it using the chevron icon.
- Review the configured settings.
- If a value is missing or incorrect, click the pencil icon to edit the setting.
-
Save your changes before continuing.
Properties
Section titled “Properties”The following properties are used to configure AI provider connections.
If a property supports a platform default, leaving it unset allows Lowgile to use the default value.
Attention
Never commit real API keys or secrets into source control or shared documentation.
| Property | Example value | Description |
|---|---|---|
| aiType | AzureOpenAi | Selects the AI provider type used by the connection, such as AzureOpenAi, Anthropic, or another supported provider. |
| instanceName | ai-instance-name | The provider instance, resource, or service identifier used by the selected AI provider. |
| deploymentName | gpt-4o | The model or deployment name used for chat or text-generation calls. |
| embeddingsDeploymentName | text-embedding-3-small | The deployment name used for embeddings calls. Required only when the application uses embeddings. |
| apiKey | <YOUR_API_KEY> | Secret API key used to authenticate with the AI provider. |
| apiVersion | 2024-06-01 | API version used by the provider, when required by the selected provider. |
| tracingEnabled | true | Enables or disables tracing for AI calls. |
| tracingEndpoint | https://api.smith.langchain.com | Endpoint used for tracing, typically LangSmith. |
| tracingApiKey | <YOUR_TRACING_API_KEY> | Secret API key used for tracing. Required only when tracing is enabled. |
| usdPer1000InputTokens | 0.005 | Optional cost value used to estimate or track input-token cost. |
| usdPer1000OutputTokens | 0.015 | Optional cost value used to estimate or track output-token cost. |
LangSmith tracing
Section titled “LangSmith tracing”AiConnection supports optional LangSmith tracing for monitoring and analyzing AI requests.
The default tracing endpoint is:
https://api.smith.langchain.comFor detailed LangSmith setup instructions, tracing concepts, and project configuration, see the official LangSmith documentation.