How to structure a RAG system with multiple Neo4j databases in Python?

Question

I am a beginner working with LLMs. I am trying to create a RAG system to work with multiple knowledge graphs. My code is working when I query one graph database, but I need to extend it to work with multiple Neo4j databases. In other words, I want the LLM to

1. generate Cypher queries for each of the databases,
2. execute those queries, and
3. use the query results to generate a unified answer. I somehow done the first two steps, the same way for using one database. But I have no idea about the third step.

Any ideas or hints on how to approach this?

Thank you.

What I've Done So Far? Here’s a part of my current code that works with a single database:

from langchain_community.graphs import Neo4jGraph
from langchain.prompts.prompt import PromptTemplate
from langchain.chains import GraphCypherQAChain
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    openai_api_key=api_key,temperature=0,model='gpt-4'
)
kg = Neo4jGraph(
    url=uri, 
    username=username, 
    password=password
)

CYPHER_GENERATION_TEMPLATE = upload_cypher_prompt() #a function for uploading Query prompt
kg.refresh_schema()
CYPHER_GENERATION_PROMPT = PromptTemplate(
    input_variables=["schema"], 
    template=CYPHER_GENERATION_TEMPLATE
)


QA_PROMPT_TEMPLATE = upload_qa_prompt() #a function for uploading QA prompt
QA_PROMPT = PromptTemplate(
    input_variables=["context", "question"], 
    template=QA_PROMPT_TEMPLATE
)

question= upload_question() #a function for uploading question
cypherChain = GraphCypherQAChain.from_llm(
            llm,
            graph=kg,
            verbose=True,
            cypher_prompt=CYPHER_GENERATION_PROMPT,
            qa_prompt=QA_PROMPT
        )   
input_variables = {
    "query": question
}

response = cypherChain.invoke(input_variables)
print(response['result'])