AI Settings

Overview

AI Settings has two parts: tenant-level AI configuration (managed by admins in Admin settings) and per-user chat display preferences (managed on the AI > Settings page).

Chat Display Settings

Access these from AI > Settings in the sidebar. These settings are per-user and stored locally in the browser.

Density

Controls spacing between messages and elements:

• Compact — tighter spacing, more content visible on screen.
• Comfortable — balanced spacing (the default).
• Spacious — more breathing room between elements.

Font Size

Adjust the text size in the chat area:

• Small — smaller text for information-dense viewing.
• Medium — standard text size (the default).
• Large — larger text for easier reading.

Layout

Configure the history panel:

• History Panel — toggle the conversation history sidebar on or off.
• History Position — place the sidebar on the left or right side of the screen.
• History Width — slider to adjust the panel width between 240px and 400px.

Custom Suggestions

Add your own quick prompts that appear alongside the default suggestions on the empty chat screen. Useful for frequently asked questions specific to your fleet.

To add a custom suggestion:

1. Type the prompt text in the input field.
2. Click Add or press Enter.
3. The suggestion appears on the empty chat screen.

Remove suggestions by clicking the X button next to each one.

Reset

Click Reset to restore all chat display settings to their defaults.

Tenant AI Configuration (Admin)

These settings are managed by admins through the Admin section and affect all users in the tenant.

Enable/Disable AI

The AI feature can be turned on or off at the tenant level. When disabled, the AI sidebar section is hidden for all users.

Provider Configuration

Gemini

• Requires an API key (stored encrypted in the database).
• Provides access to Google's Gemini model family.

Ollama

• Connects to a self-hosted Ollama instance.
• Configure the Base URL (default: http://localhost:11434).
• MineSync auto-discovers installed models from the Ollama server, falling back to a common models list if discovery fails.

OpenRouter

• Requires an API key (stored encrypted).
• Provides access to a wide range of third-party models.
• Admins can add Custom Models by specifying a model ID and display name.

Default Provider and Model

Set which provider and model are selected by default when users open a new chat.

System Prompt

An optional system prompt that is prepended to every conversation. Use this to give the AI context about your organization, fleet conventions, or preferred response formats.

Generation Parameters

• Max Tokens — maximum tokens per AI response (256 to 32,768).
• Temperature — controls response randomness from 0 (deterministic) to 2 (creative). Default is 0.7.

Feature Toggles

• Enable Tools — allow the AI to call internal tools for database queries and structured operations.
• Enable Attachments — allow users to attach files to conversations.

Model Management

• Disabled Models — hide specific models from the model selector without removing them entirely.
• Custom Models — add OpenRouter model IDs that are not in the default list.

Permissions

• Chat display settings (density, font, layout) require no special role — any user can adjust them.
• Tenant AI configuration (providers, API keys, system prompt) requires the Admin role.

Common Questions

Are API keys visible to other users?

No. API keys are encrypted before storage. The settings page only shows whether a key is configured, not its value.

Can different users use different models?

Yes. Each user can select their preferred model from the model picker in the chat. The admin setting only controls the default.

What happens if Ollama is not running?

The provider status check will show Ollama as unavailable. Users can still use other enabled providers. Ollama models will not appear in the model selector until the server is reachable.