⚠️ 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 required. All responsibility for their use lies solely with the user. It 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 4, we established three relationships between the six information types. We structured how strategic categories, target products, economic size, HTS codes, tariff rates, and implementation periods were interconnected. Now, it was time to actually collect the information.
We decided to start by collecting strategic categories. This is because strategic classification is a superordinate concept that encompasses target items, and the tariff rate range is determined by the strategy. I asked Claude, "Find the strategies used in the US semiconductor tariff policy on China.”
The search results came back, but there was a problem. Past data was mixed in, including policies implemented in 2018, investigations published in 2019, and reviews completed in 2024. I needed policies that will be implemented in the future. I didn't need past policies that had already been implemented.
I pondered how to find "future implementation policies." The simplest approach was to filter by date. "Why not just find documents published in the last few months?”
I added a time filter to web_search. I included "after:2025-09-01" in the search query, meaning I wanted to find only documents published after September 2025. The results came back, and the most recent documents were returned.
Note that search engines apply filters differently depending on the search scope. When searching broadly ('AI'), relevance is prioritized, while when searching specifically ('site:federalregister.gov Section 232'), the date filter is strictly applied. For precise filtering, using search commands like site: or more detailed search terms enclosed in quotes is effective.
However, even after filtering as specifically as possible, I felt something was missing. "Is this all there is?" I wondered, "What if there's a survey launched in April 2025? If that survey is scheduled to be conducted in 2026, it's clearly a future implementation policy. However, since the announcement date is before September, it will be missed in my search.