Category: Developer Setup / Data Automation Last tested: January 2025
In a projects i was working on, i wanted to find and gather specific data from a specific city in Asia; My goal was simply gathering the total number of cafes in one city - with names , addresses, price range, instagram pages and other data.
What I tried first:
I simply I went to AI chats to get this information, and this by specifically:
I saw what happened as interesting: some LLM (Chat GPT & Gemini) returned very partial or not so correct data.
Particularly with Chat GPT, the data was in part missing leaving the clue of difficulties in finding the data structured in a CSV file ready to go. The number of cafe were very high.
I tried giving a CSV as input as an example of what I wanted as output, but rather than receiving new data i got the same data that i gave as input as for example, back as output of final work done. The time of actually creating the data in most Chat GPT chat was around 15 minutes;
The AI models could give me some results, but not thousands of cafes present and the details I needed.
Then I rembered the concept of API i heard about, more remembering a restaurant analogy.
The restaurant analogy (how I understood it):