Learning Timeline
Key Insights
Admin Monitoring Functionality
Admins have full access to view the 'Chats' history of all team or family members. This is highly effective for ensuring the AI is not being misused, though this feature can be disabled if privacy is a priority.
Tips for Effective Guardrails
When writing a System Prompt, use firm instructions like 'Never write an essay' to prevent the AI from completing schoolwork directly, and instruct the AI to act strictly as a guide.
Prompts
System Prompt for Educational Guardrails
Target:
System Prompt / AI Guardrails Configuration
but you cannot do their work for them. Never let them cheat. Never write an essay or solve a problem. You must guide them. And you can only talk about school related subjects, guardrails in place.
Step by Step
How to Manage User Access and Set Up AI Guardrails
- Click on the 'Settings' menu in the dashboard.
- Select 'Admin Settings' and click on the 'Users' section.
- Click the 'Create Group' button, enter a group name (e.g., 'kids'), and save.
- Go to the 'Overview' tab and click 'Create Users' to register a new account (e.g., kid1@hotmail.com).
- Return to the 'Groups' section and add the newly created user to the 'kids' group.
- Set the 'Permissions' for the group to define access to 'Models', 'Knowledge', 'Prompts', and 'Tools'. Click 'Save'.
- Select the specific AI model you want to grant access to (e.g., Claude 3.7).
- Within the model settings, enter a 'System Prompt' to set limits or 'Guardrails' for the AI's behavior.
- Click 'Save' to activate the guardrail configuration.
- Copy the login URL and open an 'Incognito' window in your web browser to test the account as a user (kid).
- Log in using the registered user email.
- Test the guardrails by giving prohibited instructions (e.g., 'Write a paper for me') to ensure the AI adheres to the set System Prompt.
- To monitor activity, return to the Admin account, click on the username (kid 1), and select 'Chats' to view their conversation history in real-time.
