OpenClawに仕事を丸投げしたら、何回呼び戻される？｜実際に試した

This video demonstrates a practical test of OpenClaw's autonomous task handling by delegating the selection of meeting rooms based on 7 documents with specific conditions. The experiment measures how many times the AI agent needs to call back for additional information or clarification before completing the task. The test reveals the iterative nature of AI agent workflows and their ability to handle complex, multi-document decision-making scenarios.

Key Points

• •OpenClaw can process multiple documents (7 in this test) to make informed decisions autonomously
• •AI agents may require multiple callback iterations to gather missing information or clarify ambiguous requirements
• •Meeting room selection involves evaluating multiple criteria across different documents simultaneously
• •Understanding callback frequency helps optimize prompt engineering and task decomposition strategies
• •Real-world testing reveals gaps between expected autonomous behavior and actual agent performance
• •Document-based decision-making requires agents to cross-reference information across multiple sources
• •Iterative refinement cycles are a natural part of complex AI agent workflows
• •Task completion metrics should include callback count as a measure of efficiency

Workflow Diagram

Concepts