add comment on how to do retrieval

main.go

@@ -115,3 +115,10 @@ func singlePromptInteraction(systemPrompt, prompt string, retries int) (openai.C
 
 	return resp, nil
 }
+
+// TODO: anything to be stored in the database should be chunked to sizes between 512 and 1024 tokens
+// it should also overlap with the previous chunk by 100-200 tokens
+// When the LLM asks for more context, it should be able to use the database to find the most relevant chunks here is how:
+// We will get the embeddings for each prompt and use those embeddings to search for the closest 6 chunks
+// we will use a separate LLM prompt to make an attempt to select and sort the chunks based on the user's input
+// then we will add the best matched chunks to the main prompt as further context for the given prompt