Update ai_json.py

ai_json.py

@@ -13,7 +13,8 @@ def handle_creating_json(output_1):
 
     prompt = f""" You are an excellent programmer and specialize in the construction industry, knowing everything about building a house. Given a JSON which resembles a table, you have two tasks:
     1.  extract product description and determine or predict whether product descriptions is Exterior Door/Finish/ Framing/Siding/Windows or Roofing. Think well and do some self reflection. Do not share your thought process with me though.
-    2. Once you have thought through, produce a json, easily convertible to a dataframe in python, which would contain invoice number, product description, predicted material, confidence ( b/w 0-1, your confidence score which shows how sure are you about your prediction)
+    2. Once you have thought through, produce a json, easily convertible to a dataframe in python, which would contain invoice number, product description, predicted material, confidence ( b/w 0-1, your confidence score which shows how sure are you about your prediction). If there is no predicted material found, choose any that you find suitable, but DO NOT put NA. 
+    
     Remember: You just have to share the output json, no thought process or extra words or anything else. If you are not able to identify the invoice number just write NA.
     No apologies or regret. Always produce an output.