Skip to main content
Overview The Human Interaction node is used when part of a workflow cannot be fully automated and requires human judgment or manual input. When the workflow reaches this node, browser control is temporarily handed over to a human; after the user completes the required actions and confirms, the workflow resumes automatically. Gemini_Generated_Image_9kiorn9kiorn9kio.png Use Cases
  • Completing login flows that require username, password, and verification codes (e.g., email/SMS OTP).
  • Handling captcha-protected pages that must be solved by a human.
  • Manually reviewing content or making a decision before continuing (e.g., approve/reject, choose a category).
  • Performing sensitive or high-risk operations that should not be fully automated.
Core Capabilities
  • Pauses the workflow and temporarily transfers browser control to a human operator.
  • Supports multiple notification options: in-app notification and optional email notification.
  • Allows execution either inside the BrowserAct interface or via a standalone web interface opened from email.
  • Automatically resumes the workflow once the human step is completed and confirmed.
  • Supports a configurable time limit for how long the workflow waits for human intervention.
Configuration Steps
  1. Add a Human Interaction node to the workflow.
  2. In Instruction for human, clearly describe what the human should do on the page. For example:
    • “Please enter your username, password, and verification code to complete the login process.”
    • “Please solve the captcha and ensure the page successfully loads the dashboard.”
  3. Set Time Limit to define how long the workflow should wait for human intervention. When the time limit expires, this step ends immediately according to your workflow design.
  4. Configure Notify when this step starts:
    • Checked: Enter a valid email address. An email notification will be sent, and the recipient can open a standalone web interface to perform the task—useful when the operator is away from the main console.
    • Unchecked: The human performs all actions directly inside the BrowserAct interface without email notification.
  5. When execution reaches the Human Interaction node, the operator follows the instructions, completes the required actions in the browser, and clicks Resume to continue the workflow.
PixPin_2025-12-05_21-11-08.png Recommendations
  • Keep instructions short, specific, and action-oriented so the human knows exactly what to do.
  • Choose a reasonable time limit based on how long the manual step typically takes.
  • Use email notifications when the operator may not be in front of the BrowserAct interface.
  • Place Human Interaction nodes only at steps that truly require manual actions to avoid unnecessary interruptions.