What's Provided

Over 7 months, this methodology for creating prompts (planned with Claude) and executing (executed with Gemini API, controlled and cross-verified with Python) to select policies with future implementation potential from AI-based US-China semiconductor tariff policies will be delivered over 30 weeks, helping to strengthen your company's AI utilization capabilities.

30-Week Timeline

Week 1: Why You Need Both AIs

• Planning a 15,000-line prompt using Claude's MCP Sequential thinking and text-editor
• Executing selection of policies with future implementation potential with Gemini API
• Control and cross-verification with Python, log recording

Week 2: The Limitations of AI and the Need for Verification

• 5 fundamental incompleteness of AI
• User's continuous verification and planning ability that determines the completeness of AI output

Week 3: Types of Information Handled by Various US Government Agencies

• Diversity of content published by multiple government agencies including Federal Register, CBP, USITC, USTR, etc.

Week 4: Top-Down Information Mapping

• 3 relationships of 6 information types - 1. Inclusion relationship, 2. Intertwined relationship, 3. Set relationship

Week 5: Defining Strategic Categories

• Reason for identifying various strategies first, including National Security Strategy (Section 232), Economic Security Strategy (IEEPA), General Trade Strategy (Section 301), etc.

Week 6: Strategy Classification Prompt Structure

• Search strategy for selecting policies under review or scheduled for implementation, not past policies