Skip to main content

What is the Human Review Block?

The Human Review block pauses workflow execution and waits for human input before continuing. Use it when automated decisions need human oversight, approval, or manual data entry. Common use cases:
  • Content approval before publishing
  • Quality control checkpoints
  • Manual data verification
  • Editorial review of AI-generated content
  • Decision-making that requires human judgment
  • Compliance and legal reviews

How Does the Human Review Block Work?

The Human Review block works in three stages:

1. Workflow Pause

When execution reaches the Human Review block, the workflow pauses and sends a notification to designated reviewers.

2. Human Input

Reviewers see the content and provide input based on the format you configured:
  • Text: Free-form text response
  • Single Selection: Choose one option from a list
  • Multi Selection: Choose multiple options from a list

3. Workflow Resume

After the reviewer submits their response, the workflow automatically resumes and continues to the next step. The review response becomes available as output data.

Configuration

Review Content

You can add multiple review items to a single Human Review block. Each review item has its own configuration.

Title (Required)

The title displayed to the reviewer. Example:
Tips:
  • Use clear, descriptive titles
  • Indicate what action is needed
  • Keep it concise (under 10 words)

Instructions (Optional)

Detailed instructions explaining what the reviewer should do. Example:
Tips:
  • Be specific about what to review
  • Include evaluation criteria
  • Mention what happens after approval/rejection
  • Add context if needed

Format

Choose how the reviewer will respond. Options: Text (Default) Free-form text input. Reviewer can type any response. Use when:
  • Collecting detailed feedback
  • Revision notes needed
  • Open-ended questions
  • Editorial comments
Single Selection Reviewer chooses one option from a predefined list. Use when:
  • Approval/rejection decisions
  • Choosing between alternatives
  • Status updates
  • Simple yes/no questions
Note: When using Single Selection, the “Value to Review” must be a list (array) of options. Multi Selection Reviewer can choose multiple options from a list. Use when:
  • Checking off completed items
  • Selecting applicable categories
  • Multiple issues to flag
  • Feature selection
Note: When using Multi Selection, the “Value to Review” must be a list (array) of options.

Value to Review (Required)

The content the reviewer will see and evaluate. Reference output from previous workflow steps. Examples: Review AI-generated content (Text format):
Review formatted data (Text format):
Review multiple fields (Text format):
Single Selection - List of options:
Multi Selection - List of options:
Single Selection - From previous step:
Tips:
  • For Text format: Reference the specific content that needs review
  • For Single/Multi Selection: Provide a list (array) of options
  • Format for readability
  • Include relevant context
  • Can combine multiple step outputs

Adding Multiple Review Items

Click Add Content For Review to add additional review items to the same block. Use cases:
  • Review multiple content pieces simultaneously
  • Separate approval for different aspects
  • Collect different types of feedback
  • Multi-step approval process
Example configuration:
Each review item gets a separate output field accessible via:

Error Handling

Define what happens if the review process encounters an error. Options:
  1. Terminate Workflow: Stop execution immediately if error occurs
  2. Continue Execution: Proceed to next step despite error
Recommendation: Use Terminate Workflow to ensure critical approvals are not skipped.

Output Data

The Human Review block returns an object containing responses for each review item.

Output Structure

Accessing Review Responses

Single review item:
Multiple review items:

How Do I Use Review Responses?

The review responses can be used in subsequent workflow steps.

Example: Incorporating Feedback

Human Review output:
LLM block using feedback:
Use cases:
  • Pass review feedback to LLM blocks for content revision
  • Include review responses in final output text
  • Reference approval status in Google Sheets for record keeping
  • Display review comments in subsequent workflow steps

Best Practices

Review Configuration

  • Clear instructions: Explain exactly what to review and what response is needed
  • Provide context: Include enough information for informed decisions
  • Set expectations: Indicate urgency and what happens after review
  • Format wisely: Use Single/Multi Selection for standardized decisions

Workflow Design

  • Place strategically: Put review blocks at critical decision points
  • Minimize reviews: Don’t overuse - too many reviews slow workflows
  • Group related reviews: Multiple items in one block instead of multiple blocks
  • Plan for delays: Remember workflows will pause until review completes

Response Handling

  • Store decisions: Save review responses to sheets for audit trail
  • Use feedback effectively: Pass review comments to subsequent LLM blocks for revisions
  • Document responses: Include review data in final output for record keeping

Common Patterns

Pattern 1: Content Review and Revision

Pattern 2: Multi-Stage Approval

Pattern 3: Batch Content Review


Workflow Behavior During Review

What Happens When Review Starts

  1. Workflow execution pauses at the Human Review block
  2. Reviewer receives notification (if configured)
  3. Workflow status shows “Awaiting Human Review”
  4. Other workflow steps do not execute until review completes

What Reviewers See

  • The title you configured
  • Instructions (if provided)
  • The value to review (formatted content)
  • Input field based on format selected
  • Submit button to complete review

After Review Submission

  1. Reviewer’s response is captured
  2. Workflow automatically resumes
  3. Output becomes available to subsequent steps
  4. Workflow continues to completion

Tips for Effective Reviews

For Review Configuration:
  • Include examples of good/bad responses in instructions
  • Use selections when possible for faster reviews
  • Limit text reviews to cases needing detailed feedback
  • Show only relevant context - not everything
For Reviewers:
  • Keep responses consistent with format expectations
  • Provide actionable feedback in text responses
  • Use selection options as defined
  • Complete reviews promptly to avoid workflow delays
For Workflow Design:
  • Use review feedback in subsequent LLM blocks for improvements
  • Plan for workflow delays during review periods

What’s Next

Now that you understand the Human Review block: