⚠️ This column is a free educational sharing of the prompt creation process and is not any investment, legal, or financial advice. AI execution results may contain errors, and independent verification is essential. All responsibility for use is entirely attributed to the user themselves. It cannot be used directly for business decision-making. [View all required confirmation items below the 30-week timeline before use[]](https://www.notion.so/30-29f84ddcec3c80a1a87eda9883d81453?pvs=21)
Before moving on to Week 13, there is something I need to tell you.
As I mentioned in Week 1, this series covers only the 30-week process of planning prompts to collect and select policy information with future implementation possibilities among US-China semiconductor tariff policies based on a specific search date using Claude, but the actual execution was done using Gemini API (Paid Tier) in Google Colab.
Since we only cover the prompt design process until Week 30, I felt that if not now, there would be no opportunity to tell you how much Google Infrastructure helped with data accuracy, so I prepared this break corner.
As I mentioned, I originally intended to use Claude for execution as well, but after using it nearly 12 hours a day, even with the Max Plan subscription, I reached the weekly limit around the time the planning was completed.
At that time, I urgently needed results, so while looking for a solution, I discovered Google's Vertex AI. It is Google's enterprise AI platform that operates with Python code and seemed capable of large-scale work, so I converted the 15,000-line prompt to fit Gemini 3 Pro, and after conversion, I requested AI to convert the converted prompt into Vertex AI Python code.
However, when I requested Vertex AI Python code from Gemini 3 Pro (Paid Tier), it first completed and provided me with Gemini API-specific Python code optimized for fast execution.
When it said I could convert this to Vertex AI, I asked why it didn't make Vertex AI-specific code from the beginning, and it recommended testing with the completed Python code first, warning that converting to Vertex AI-specific Python code would become more complex, and even using itself, Gemini 3 Pro, would require more considerations and work time.
If converted to Vertex AI, execution time would be shortened, but I would have to write and test code for longer than the time saved,