How to use Chromadb Rag with OpenAi

from openai import OpenAI
from langchain_openai import ChatOpenAI
from langchain.prompts import ChatPromptTemplate
import win32com.client
import chromadb
from chromadb.utils.embedding_functions import OpenAIEmbeddingFunction

client_chromadb = chromadb.PersistentClient("C:\\Users\dnishimoto\\python_files\\python-deep-learning-master")

collection_name = "emails"
has_collection = [collection for collection in client_chromadb.list_collections() if collection.name==collection_name]

if has_collection:
   collection=client_chromadb.get_collection(collection_name,embedding_function=OpenAIEmbeddingFunction(api_key=key))
else:
    collection= client_chromadb.create_collection(
    name="emails",
    embedding_function=OpenAIEmbeddingFunction(api_key=key)
    
)

session = win32com.client.gencache.EnsureDispatch ("Outlook.Application")
namespace = session.GetNamespace("MAPI")
inbox = namespace.GetDefaultFolder(6)
messages = inbox.Items

subjectLines=[]
for message in messages:
  
        print("Subject:", message.Subject)
        subjectLines.append(message.Subject)
        collection.upsert(ids=[message.EntryID],documents=[message.Body[:7000]])

instructions =f"summarize in detail {', '.join(subjectLines)}" 
result= collection.query(
    query_texts=[instructions],
    n_results=10
)

llm=ChatOpenAI(model="gpt-4o-mini",temperature=0.7, openai_api_key=key)

input=""
for document in result["documents"]:
    for i in range(len(document)):
        input+=document[i]
    
prompt=ChatPromptTemplate.from_template(f" perform the following user task instruction: {instructions} from the follow input:{input}")
chain=prompt| llm
response=chain.invoke({"input":input})
output=response.content
print(output)
        

The emails are stored as embedded documents in the chromadb sqllite database for query by key word later. langchain summarizes the last 10 chromadb documents