General
The General tab contains the core configuration for a Custom Model. Navigate to Workspace → Models → Custom Models, select your model, and open the General tab.
Fields
Name
The display name of the model as shown in the Lakehousecat interface. Choose a clear, descriptive name that communicates the model's purpose.
Model ID
A unique identifier auto-generated from the name. The Model ID cannot be changed after the model has been created.
Provider Model
The underlying Provider Model (Chat type) that this Custom Model uses for all AI interactions. Only Chat-type Provider Models that have been shared with you are available for selection. If the list is empty, ask your Administrator to share a Provider Model with you.
System Prompt
Instructions that guide the model's behavior, tone, and responses. Use the System Prompt to define the model's persona, restrict its scope, or provide domain-specific context.
Voice input is available via the microphone button for convenient dictation.
Description
An optional text field visible to users when browsing models. Use it to explain the model's purpose and what kind of questions it is best suited for.
Model State
Two switches control the operational state of the model:
| Switch | Description |
|---|---|
| Enabled | If disabled, the model is no longer available for use. |
| Locked | If enabled, the model cannot be edited or deleted. Use to protect a finalized configuration. |
Agent Mode
Enables the conversational AI Agent for this model. When active, users interact with an intelligent assistant that can generate charts, navigate your data, and manage platform resources through natural language — instead of a standard chat completion.
| Switch | Description |
|---|---|
| Enable Agent Mode | Activates the AI Agent for sessions using this model. |
The AI Agent is role-aware: what it can do depends on the role of the user who is currently chatting. A standard User can generate charts; a Builder can also manage datasources and models; an Admin has full platform access.
See Agent Mode for a detailed overview of capabilities and behavior.
Materialization
Enables performance optimization for this model's data layer. When active, the model's analytical views are pre-computed and stored — so chart and dashboard queries read from pre-built data instead of recalculating joins on every request.
| Switch | Description |
|---|---|
| Enable Materialization | Activates pre-computed analytical views for this model. Faster chart queries at the cost of additional storage. |
Materialized views are refreshed automatically after each datasource load. Charts and dashboards remain fully functional during and after the switch.
See Materialization for details on when to enable it and what to expect.
Prompt Suggestion Visibility
Controls which types of prompt suggestions are shown to users when they select this model:
| Switch | Description |
|---|---|
| Show Global Suggestions | Displays globally configured prompt suggestions. |
| Show Model-Specific Suggestions | Displays suggestions defined specifically for this model (see the Suggestions tab). |
| Show Datasource Suggestions | Displays suggestions derived from connected data sources. |
Profile Image
An optional image used to visually identify the model in lists. Supported formats: JPEG, PNG, GIF, WebP.
Saving
Click Save to apply changes. The model stays open in the editor after saving.
Best Practices
- Write a System Prompt that is precise and focused on the model's intended use case.
- Use the Locked switch to prevent accidental changes to production models.