⚠️ This column shares the process of creating prompts for free educational purposes and does not constitute investment, legal, or financial advice.** AI execution results may contain errors, so independent verification is essential. All risks associated with their use lie solely with the user. They cannot be directly used for business decisions. [See all essential checklists below the 30-week timeline before use[]](https://www.notion.so/30-29f84ddcec3c80a1a87eda9883d81453?pvs=21)
In Week 2, we learned about the limitations of AI and the importance of verification. However, as I mentioned, when we started collecting policy information, we knew nothing about it and didn't even know what to collect. In particular, the statements made by key figures in international politics and the policy information from government agencies I had previously collected differed significantly in terms of their sources and scope, making it even more difficult to grasp the methods for collecting policy information.
Before discussing the various government agencies announcing tariff policies, I'd like to first explain how I monitor international political trends, compare them, and explain the differences.
During my 50 months of overseas futures trading, I've closely observed international political trends. I believe international politics is the most important investment tool, so information like Trump's remarks, interviews with key politicians, and White House briefings move the market, making it essential to quickly grasp this information.
When I decided to utilize AI, my first priority was to automate this process of monitoring international political trends. While I could have connected to platforms like X (Twitter) and APIs to quickly identify statements from key politicians, I'm not a developer, so I wanted to proceed with the simplest approach. Therefore, I decided to utilize web_search, a search tool already provided by Claude, as a simpler approach.
When I request a search, Claude instantly selects appropriate keywords, searches various sites, and summarizes the results. Typically, it reviews the titles and content of 10 articles at a time and organizes them. If there's no significant content, it might check fewer than 10 articles. However, at peak times, it might perform five searches of 10 articles each, for a total of 50 articles.
This method was perfect. Specifically, the combination of Sequential Thinking and web_search allowed Claude to automatically research the information I needed, performing additional searches if necessary. Simple search commands were sufficient, without complex technology, and I was extremely satisfied.
As mentioned in Week 2, I gradually developed the system, analyzing the summarized content into seven hubs, strengthening the insights, and even generating and validating predictions. Having successfully utilized AI to understand international political trends, I was confident that policy information collection would also be easy, and I was ready to move on to the next step.