Skip to main content

Approval Mode

What is Approval Mode?

Approval Mode lets you decide when your AI agents can act on their own and when they need to check with you first. Think of it like setting permission levels for your digital assistant.

Why Use Approval Mode?

Approval Mode helps you find the perfect balance between:

  • Getting things done quickly (automation)
  • Keeping control over important decisions (oversight)

This means you can confidently use AI across your organization while maintaining the right level of human involvement for each task.

How It Works

Approval Mode gives you two simple options for each tool your agent uses:

✈️ Autopilot Mode

Your agent can use the tool without asking for permission first.

Best for:

  • Routine tasks with minimal risk
  • Well-tested processes with predictable results
  • When speed matters more than double-checking
  • Tasks you’ve already verified work correctly

👤 Human-in-the-Loop Mode

Your agent must get your explicit approval before using the tool.

Best for:

  • Important decisions with significant impact
  • Actions involving sensitive information
  • New workflows you haven’t fully tested yet
  • Tasks that need to meet compliance requirements
  • During training periods when you’re still evaluating performance

Key Benefits

  • Flexible Control: Adjust how much freedom your agent has based on what it’s doing
  • Reduced Risk: Prevent potential mistakes in critical operations
  • Step-by-Step Automation: Start with more oversight and gradually give more independence
  • Compliance Ready: Meet regulations that require human review
  • Optimized Workflow: Balance speed and accuracy based on your specific needs

Setting Up Approval Mode

  1. Go to “Edit Agent” in your Relevance AI dashboard
  2. Navigate to the “Tools” section under “Connected Resources” in the sidebar
  3. Select the specific tool you want to configure
  4. Find the “Approval Mode” toggle
  5. Choose either “Autopilot” or “Human in the Loop”
  6. Save your changes

Best Practices

  • Evaluate Risk: Consider what could go wrong with each tool to decide the right approval level
  • Start Cautiously: Use Human-in-the-Loop for new tools until you trust them
  • Use with Escalation: Combine with the “Escalate to Humans” feature for complete oversight
  • Review Regularly: Check how tools are performing and adjust settings as needed
  • Keep Records: Document which tools require approval and why

Frequently asked questions (FAQs)

Can I change approval settings after setup?

Yes! You can change the approval mode for any tool at any time by editing your agent.

How will I know when my approval is needed?

You’ll receive a notification in the agent interface when your agent needs permission to use a tool.

Can different team members handle approvals?

Yes, approval requests can be directed to specific team members based on your permissions setup.

Does Human-in-the-Loop mode slow things down?

It adds some delay since the agent waits for your approval, while Autopilot mode runs immediately. Choose based on your speed vs. control needs.

Can I mix approval modes within one agent?

Absolutely! You can set different approval modes for each individual tool, giving you precise control over what needs supervision.