Skip to main content
System Settings provide organization-wide configuration options including authentication methods, AI/ML model selection, data retention, and observability settings.

​
Settings Architecture

Support Bot uses a versioned settings system with segment-based configuration:

Segments

Settings are organized into logical segments (AI/ML, Auth, etc.) that can be updated independently.

Versioning

Every settings change creates a new version with full audit trail and rollback capability.

Append-Only

Old versions are never deleted, enabling complete change history and forensics.

Atomic Updates

Each segment update is atomic - either all changes succeed or none do.

​
AI/ML Configuration

​
Model Selection

Choose which AI model powers your Support Bot responses.
1

Navigate to AI/ML Settings

Go to Settings > AI / ML Configuration.
2

Select Model

In the Model & Generation section, choose from available models.
Available models come from your configured LLM providers. Add providers first to see more options.
3

Set Temperature

Adjust the temperature slider (0.0 to 1.0):
  • Lower (0.0-0.3) - More focused, consistent, deterministic
  • Medium (0.4-0.7) - Balanced creativity and consistency
  • Higher (0.8-1.0) - More creative, varied, less predictable
For support use cases, values between 0.2-0.4 work best.
4

User Model Selection

Toggle Let Users Choose Model to allow users to select their preferred model in chat sessions.When disabled, all users are locked to the admin-selected model.
5

Save

Click Save to apply your model configuration.

​
Safety Filters

Protect your organization from inappropriate content.
1

Configure Deny List

In the Safety & Observability section, add words to block:
  1. Type words separated by commas in the input field
  2. Click Add word
  3. Repeat for all words you want to filter
Deny list words are filtered from both user prompts and AI responses.
2

View All Deny Words

Click View All to see the complete list in a table format.
3

Remove Words

In the table view, click the × button next to any word to remove it.
4

Save Changes

Click Save to apply deny word updates.

​
Observability

1

Enable Langfuse Tracing

Toggle Enable Langfuse Tracing to send all AI interactions to Langfuse for monitoring.

Configure Langfuse

Set up your Langfuse account and configure environment variables:
  • LANGFUSE_PUBLIC_KEY
  • LANGFUSE_SECRET_KEY
  • LANGFUSE_HOST
3

Save

Click Save to enable tracing.
Langfuse provides insights into response quality, latency, token usage, and costs.

​
Authentication Settings

Control which authentication methods are available to users.

​
Configure Auth Providers

1

Navigate to User Management

Go to Settings > User Management.
2

Toggle Providers

In the Authentication Methods section, enable or disable:
  • Basic Authentication - Email/password login
  • Google Authentication - Sign in with Google
  • GitHub Authentication - Sign in with GitHub
  • Microsoft Authentication - Sign in with Microsoft
3

Save Changes

Click Save to apply authentication updates.
Disabling a provider prevents new login sessions but doesn’t terminate existing sessions.

​
Provider Setup

Before enabling OAuth providers, configure them in your environment:
  1. Create OAuth credentials in Google Cloud Console
  2. Add authorized redirect URIs
  3. Set environment variables: 
    GOOGLE_CLIENT_ID=your_client_id
GOOGLE_CLIENT_SECRET=your_client_secret

​
Configuration History

View and rollback to previous settings versions.

​
View History

1

Navigate to History

Go to Settings > Configuration History.
2

Review Changes

The history table shows:
  • Timestamp - When the change was made
  • User - Who made the change
  • Change Type - Create, Update, or Rollback
  • Changes - Summary of what changed
  • Settings Values - Current values in that version
3

Filter by Segment

Use the segment filter to show only changes to specific settings (AI/ML, Auth, etc.).

​
Rollback to Previous Version

Rolling back creates a new version with the old values. The rollback itself becomes part of the audit trail.
1

Find Version

In the history table, locate the version you want to restore.
2

Click Rollback

Click the Rollback button for that version.
3

Add Reason (Optional)

Enter a reason for the rollback (helps with audit trail).
4

Confirm

Confirm the rollback. A new version is created with the old values.
Rollback is instantaneous - settings take effect immediately for all users.

​
Data Retention

Configure data retention policies to manage stored conversations and incident data.
Data retention features include:
  • Conversation Archiving - Archive old conversations to reduce active database size
  • Manual Deletion - Delete specific conversations or incidents through the admin interface
  • Version Cleanup - Remove old dataset versions to free up storage space

​
API Access

All settings can be managed programmatically via API.

​
Get Current Settings

GET /api/settings/segment/{segment}
Returns the current values for a specific segment (aiml, auth).

​
Update Settings

PUT /api/settings/segment/{segment}
Updates settings for a specific segment. Creates a new version.

​
View History

GET /api/settings/history?limit=50&segment=aiml
Retrieves version history with optional segment filter.

​
Rollback

POST /api/settings/rollback/{version_id}
Rolls back to a specific version. See the Settings API documentation for full details.

​
Required Permissions

ActionPermission Required
View AI/ML settingsaiml.view
Edit AI/ML settingsaiml.edit
View auth settingsauth.view
Edit auth settingsauth.edit
View historyhistory.view
Rollback settingshistory.rollback

​
Best Practices

Test Before Production

Test configuration changes in a staging environment before applying to production.

Document Changes

Use the rollback reason field to document why settings were changed.

Monitor Performance

After changing AI models or temperature, monitor response quality and user feedback.

Regular Reviews

Periodically review settings history to audit configuration changes.

​
Troubleshooting

​
Settings Not Saving

Verify you have the required permission for that settings segment (aiml.edit, auth.edit).
Check that all required fields are filled and values are in valid ranges (e.g., temperature 0-1).
Another admin may have saved changes at the same time. Refresh and try again.

​
Model Not Available

  1. Verify LLM providers are configured and active
  2. Check provider health status is “success”
  3. Try discovering models again from the provider
  4. Ensure the provider has a valid API key

​
Rollback Failed

  1. Verify you have history.rollback permission
  2. Check that the version ID is valid
  3. Ensure the version hasn’t been deleted (versions are never deleted in normal operation)

LLM Providers

Configure AI model providers before selecting models.

User Management

Manage authentication and user access control.

Roles & Permissions

Configure who can modify settings.

API Reference

Programmatic settings management.

Build docs developers (and LLMs) love

Get started for freeTalk to us