Enterprise Manager logoEM: AI > AI Settings

Description

The AI Settings screen is where you configure the large language model (LLM) providers that power the artificial intelligence features in the Enterprise Manager. From this single screen you can enter provider credentials, tune each provider's behavior, choose which models are available, and select the provider and model that should be used by default whenever an AI feature does not specify one.

Location

Enterprise Manager logoEM Navigator → AI → AI Settings

How This Screen Fits Into the Enterprise Manager

The provider credentials, models, and defaults which configured here are not limited to this screen. They are shared by every AI-powered feature in the Enterprise Manager, including:

  • The AI Query Builder, which turns natural language questions into SQL statements (see the AI Query Builder documentation page).

  • The AI chat panel available from the File System > Auditing results screen, which lets you ask questions about audit log search results in plain language.

Both of those features allow user to pick which of the providers/models configured here to use for that particular request, but neither of them lets you edit provider credentials directly — all provider configuration is managed from this AI Settings screen.

All of the configuration on this screen is also available through the Admin API, not just through the Enterprise Manager. The AI configuration is represented by the BBjAdminAIConfig class and is read and saved through the BBjAdminAI interface (obtained by calling getAI() on your admin connection). This means administrators can also read or script AI provider configuration programmatically wherever the Admin API is available.

Supported Providers

The Enterprise Manager supports four AI providers, each configured in its own expandable section:

  • Anthropic

  • Gemini(Google)

  • OpenAI

  • GitHub Copilot

Click a provider's section header to expand or collapse it. The provider currently marked as the system default is always shown at the top of the list, with (Default Provider) appended to its name.

Provider Settings

Each provider section exposes the following fields. A few fields are specific to a single provider, as noted below.

Settings Applies To Description
API Key / GitHub Token (PAT) All providers The credential used to authenticate with the provider. Anthropic, Gemini, and OpenAI use an API key; GitHub Copilot uses a personal access token (PAT). This value is masked on screen for security.
Temperature All providers Controls how deterministic or creative the model's responses are. Lower values (closer to 0) produce more consistent, predictable output — important for reliable SQL generation. Higher values allow more varied responses.
Timeout (sec) All providers The number of seconds the Enterprise Manager will wait for the provider to respond before giving up on a request.
Max Tokens All providers The maximum number of tokens the model may generate in a single response. Enter 0 to use the provider's own default limit.
Base URL (optional) OpenAI only Lets you point requests at a custom or self-hosted OpenAI-compatible endpoint instead of the standard OpenAI service. Leave blank to use the default OpenAI endpoint.
Inference Endpoint GitHub Copilot only The inference endpoint URL that GitHub Copilot requests are sent.
Models All providers The list of model identifiers (for example, claude-haiku-4-5) available for that provider. See Managing Models below.

Managing Models

Each provider maintains its own list of models. Use the buttons next to the model list to manage it:

Button Function
Add... Opens a prompt where you can type in a new model identifier to add to this provider's list.
Remove Removes the currently selected model from the list. Every provider must always have at least one model configured, so Remove has no effect if only one model remains.
Set As Default Model Marks the currently selected model as this provider's default. The default model is labeled (default) in the list and is the model used for that provider whenever a request does not explicitly choose a different one.

Setting the Default Provider

Exactly one provider is designated the system-wide default at any given time. This is the provider (and its default model) that AI features use when no other provider/model is explicitly selected.

To change the default provider, expand that provider's section and click Set As Default Provider. The provider you choose moves to the top of the list and its section header is updated to read <Provider Name> (Default Provider). Its Set As Default Provider button is disabled, since a provider that is already the default cannot be set as the default again.

System Prompts

Two text areas at the bottom of the screen let you customize the system-level instructions given to the AI model for different kinds of requests:

Settings Description
SQL Generation Prompt The system prompt used when the AI Query Builder converts a natural language question into a SQL statement. Use this field to give the AI additional guidance about how you want generated SQL to be written (for example, preferred formatting, dialect conventions, or business rules to keep in mind).
General Assistant Prompt The system prompt used for general-purpose AI assistant interactions elsewhere in the Enterprise Manager.

Saving Your Changes

Changes made on this screen are not applied until you save the editor (for example, using the Save toolbar button or the Ctrl+S keyboard shortcut). The editor's tab label indicates unsaved changes in the usual way. Once saved, the updated provider configuration is immediately available to the AI Query Builder and the auditing chat panel.