Human-in-the-Loop

How Kadoa’s human-in-the-loop system ensures data quality and handles edge cases

Kadoa’s human-in-the-loop system combines AI automation with human oversight to ensure high-quality data extraction.

Preview State

After creating a workflow, Kadoa automatically generates a data sample for your review:

1. Preview generates automatically → You’ll get an email when ready
2. Review the data → Check that fields are extracted correctly
3. Edit and rerun → Adjust your schema or prompt and rerun until it looks right
4. Approve → The workflow processes the full dataset

Preview mode is free — no quota is used until you approve. Edit your schema or prompt and rerun as many times as you need.

Error Handling and Recovery

Kadoa uses a multi-layer approach to handle issues automatically:

1. Self-Healing (Automatic)

Our AI adapts to changes without intervention:

• Detects page structure changes and updates selectors
• Adjusts to new layouts and UI elements
• Retries with different strategies when extraction fails

Most issues are resolved at this stage without you knowing.

2. Error Detection & Recovery

If self-healing can’t resolve the issue:

• System enters error state
• Automatic retries with expanded strategies
• You’re notified via email and dashboard

3. Support Investigation

For complex issues that can’t auto-recover:

• Kadoa support team is automatically alerted
• Engineers investigate with detailed error logs
• Manual fixes are applied when needed

Common issues we handle: authentication problems, site maintenance, anti-bot challenges, and network failures.

For all workflow states, see States.