Over the past couple of years, we have all seen the rise of AI tools. In this article, we will explore how to better use AI to find trade opportunities. Notice that this article is initially written in December 2024, reflecting the current level of AI tools. This is an important disclaimer, since the AI landscape is changing by the day. Many considerations and shortcomings may no longer be applicable soon. It is worth confirming that this article is not written by AI, unless otherwise indicated, in order to document the output of AI systems as needed. The main point here is that using better prompts, for the current AI technology, can help you find better trade opportunities.
Why Prompts and Generative AI?
This paragraph may be a bit technical, but it is important to have a basic understanding of modern AI, which can have some confusing terms and concepts. Current AI tools use what is known as “Large Language Models” (LLMs). These models are basically statistical approaches to help the computer generate a response by predicting the next word.
That’s why these modern AI approaches are referred to as “Generative AI”, a term that also encompasses AI ability to generate text, images, audio, video and so forth. The “Large” part of LLMs is that each word, or even part of a word, is represented as a huge matrix of numbers (which are called parameters, and they are organized in what we call the transformer). One case of these models is the GPT model, which is the basis of ChatGPT. GPT means “Generative Pre-trained Transformer”. In the training part, the transformer learns the numbers of each word or part of a word, just by the context. It knows that a certain word usually comes before another word. It also learns the parameters of each word, hence it understands the meaning of the word to some extent.
How can knowing the prompts help us?
As you have seen above, generative AI basically “predicts” the next word. It makes the prediction based on the previous words. Therefore, if we ask an AI system “what is the capital of France?”, it will supposedly respond “Paris”. This is because it has seen sentences such as “The capital of France is … Paris“.
Suppose we ask ChatGPT the same question as the one in the title of this article “How to Find Trade Opportunities Using AI”. We will find the answer listed at the end of this article under this prompt (prompt 1 in the annex). This answer can be quite generic or formulaic. It may give some useful tips or pointers, but it is not exactly what we want. We want to get specific directions on basically how to find importers.
How to Improve your prompts?
So this is what we do. A very simple starting point here is to actually ask ChatGPT to provide you the prompts you can use. Something like this:
“What are some useful prompts that I can use to discover companies which import a certain product?”
Or: “I need prompts that I can use here on ChatGPT or other AI systems”
Interestingly, you can find responses such as these:
“List companies that import to [country]. Include relevant industries and trade partners”
“Identify major importers of in [region]. Provide examples of companies, if possible.”
So let’s try it with this example:
“List companies that import onions to Germany. Include relevant industries and trade partners”
And here we can actually see the results (these are generated on ChatGPT, latest model on 19 December 2024)
[Beginning of AI generated content – not fact checked]
Germany is a significant importer of onions, sourcing both fresh and dried varieties from various countries. In 2022, Germany imported approximately $350 million worth of onions, making it the fourth-largest onion importer globally.
Key Importing Companies:
Several companies play pivotal roles in importing onions into Germany:
- TRANSFOOD GMBH: Accounts for 28% of onion imports, with 132 shipments.
- H G ASIEN OBST UND GEMUSE GMBH: Holds a 26% share, corresponding to 122 shipments.
- MSTRANS FOOD GMBH: Contributes 12% of imports, totaling 54 shipments.
Volza
…
[End of AI generated content – full response in annex-prompt 4]
While generating this response, ChatGPT indicated that it performed web-search. It also included references to sources. These can be very useful to make sure that the system is not hallucinating or making up results. However, notice that not all AI systems or models do the exact same thing all the time, so it is useful to always check the data produced by AI systems.
Conclusion
We learned in this article briefly about using AI prompts to get trade information. This information can unlock export opportunities and give you a better perspective. You can ask the AI system to suggest prompts you can use to get this information. When you have more focused prompts, you will have relevant results. It is important to notice that AI systems are not a search engine. We have discussed improving using search engines in another article. We can always use the interactive chat features, such as “give me more information”, or “no, I wanted information about X”, and so forth. This way, you can get the maximum value of AI systems to find better export, import, and trade opportunities in general.
Annex – ChatGPT output
We are dealing primarily here with ChatGPT, although we should note that we have a multitude of other AI platforms, systems and models. The underlying concerns are largely the same for all modern AI tools. You will find in the links below the full responses by ChatGPT to each prompt.
Prompt 1:
If I am exporting a product, how can I use AI to find importers of this product?
Prompt 2:
What are some useful prompts that I can use to discover companies which import a certain product?
Prompt 3:
no, I need prompts that I can use here on ChatGPT or other AI systems
Prompt 4:
List companies that import onions to Germany. Include relevant industries and trade partners