Human-in-the-Loop

Human-in-the-Loop is Envole’s way of ensuring you stay in complete control when your assistants want to take actions that affect your systems, data, or communications. Instead of assistants acting autonomously, they ask for your explicit approval before executing important tasks. Think of it as having a thoughtful assistant who checks with you before sending emails, creating calendar events, updating project management tools, or making changes to documents. You get the efficiency of AI automation with the safety and control of human oversight.

​

Why Human-in-the-Loop Matters

Safety First: Prevents assistants from taking unintended actions that could affect your work, relationships, or data. Maintain Control: You decide what actions are appropriate for each situation, maintaining authority over your tools and communications. Learn and Trust: As you see what assistants want to do, you build trust in their judgment and can provide feedback to improve future suggestions. Compliance: Ensures sensitive actions meet your organization’s approval requirements and policies.

​

How Tool Approval Works

When your assistant determines that a task requires taking an action—like sending an email, creating a calendar event, or updating a project status—it will pause and ask for your approval instead of acting immediately.

​

Step 1: Assistant Identifies Actions Needed

Your assistant analyzes your request and determines what specific actions it needs to take. For example, if you ask it to “schedule a follow-up meeting with the client,” it identifies that it needs to:

• Create a calendar event
• Send invitation emails to attendees
• Possibly update your CRM with the meeting details

​

Step 2: Approval Request Appears

Instead of taking these actions immediately, your assistant presents them to you for review and approval. ![Tool Approval Interface - Image 1] What you’ll see:

• Clear description of each action the assistant wants to take
• Specific details like email recipients, calendar event details, or document changes
• Preview of exactly what will happen if you approve
• Options to approve, deny, or provide feedback

​

Step 3: Review and Decide

You have three options for each action: Approve: The assistant will execute the action exactly as proposed. Deny: The assistant will skip this action but continue with others you approve. Provide Feedback: Stop the process and give the assistant feedback about why the action isn’t quite right, allowing it to revise its approach. ![Approval Options - Image 2]

​

Step 4: Batch Decision Making

Often, assistants will propose multiple related actions at once. Envole presents these as a group so you can:

• See how actions work together
• Make decisions about the entire workflow
• Ensure all actions align with your intentions

![Batch Approval Interface - Image 3] Key insight: Related actions are grouped together because they often depend on each other. For example, if you’re organizing a project kickoff, the assistant might want to create the meeting, send invitations, and update the project timeline all at once.

​

Step 5: Execution and Confirmation

Once you approve actions, your assistant executes them and provides confirmation of what was completed. ![Execution Confirmation - Image 4] You’ll receive:

• Confirmation that approved actions were successfully completed
• Details of what was accomplished (emails sent, events created, etc.)
• Information about any actions that couldn’t be completed and why

​

Common Human-in-the-Loop Scenarios

​

Email and Communication

• Scenario: “Send a project update to the team”
• Approval needed: Review recipient list, email content, attachments
• Why important: Ensures the right people get the right message at the right time

​

Calendar Management

• Scenario: “Schedule meetings with our key clients”
• Approval needed: Meeting times, attendee lists, meeting locations or video links
• Why important: Prevents scheduling conflicts and ensures appropriate attendees

​

Project Management

• Scenario: “Update project status based on our team standup”
• Approval needed: Status changes, task assignments, deadline updates
• Why important: Maintains accurate project tracking and team alignment

​

Document Creation and Updates

• Scenario: “Create a proposal based on our discussion”
• Approval needed: Document content, sharing permissions, storage location
• Why important: Ensures documents meet your standards and are shared appropriately

​

Tool Integrations

• Scenario: “Create Jira tickets for the bugs we discussed”
• Approval needed: Ticket details, assignees, priority levels
• Why important: Ensures accurate issue tracking and proper workflow management

​

Making Approval Decisions

​

What to Look For

Accuracy: Are the details correct? Check recipients, dates, amounts, and other specifics. Appropriateness: Is this the right action for the situation? Consider timing, audience, and context. Completeness: Does the action accomplish what you intended? Are any important details missing? Impact: What are the consequences of this action? Who will be affected and how?

​

When to Approve

• Details are accurate and appropriate
• The action aligns with your goals
• Timing is right for the intended outcome
• You’re comfortable with the potential impact

​

When to Deny

• Some details need adjustment but the overall approach is wrong
• You want the assistant to try a different approach
• The timing isn’t right for this action
• You prefer to handle this manually

​

When to Provide Feedback

• The approach is close but needs refinement
• Important context or constraints were missed
• You want to guide the assistant toward a better solution
• The action needs significant modification

​

Best Practices for Human-in-the-Loop

​

Be Specific in Your Feedback

When providing feedback, help your assistant understand not just what to change, but why:

• ❌ “This email doesn’t look right”
• ✅ “The email tone is too formal for this client relationship, and we should include the budget discussion we had last week”

​

Consider Batch Dependencies

When reviewing multiple actions, think about how they work together:

• If you approve the meeting but deny the email invitation, how will attendees know about it?
• If you approve updating the project status but deny notifying stakeholders, will important people be left out of the loop?

​

Use Approval Patterns

As you work with your assistants, you’ll develop patterns for what you typically approve or modify:

• Always review external communications before sending
• Auto-approve internal calendar bookings during specific hours
• Always check budget-related updates before execution

​

Provide Context for Better Future Suggestions

The more context you provide in your feedback, the better your assistant becomes at making appropriate suggestions:

• Explain your preferences for different types of communications
• Share information about team dynamics or client relationships
• Clarify organizational policies or constraints

​

Benefits of Human-in-the-Loop

Builds Trust: You see exactly what your assistant wants to do before it happens, building confidence in AI assistance. Improves Over Time: Your feedback helps assistants learn your preferences and make better suggestions. Prevents Mistakes: Catches potential issues before they impact your work or relationships. Maintains Standards: Ensures all AI-generated actions meet your quality and appropriateness standards. Enables Learning: You understand what tasks assistants can handle and how they approach different challenges.

​

What’s Next?

Congratulations! You now understand all the key concepts that make Envole powerful: Team Personal Assistants, Sub-Assistants, Multi-Assistant Collaboration, and Human-in-the-Loop approval. You’re ready to start transforming how your team works.

Get Started with Envole

Start your free trial and experience intelligent team assistance

​

Dive Deeper