I think QARetrievalChain and GraphCypherChain both output runnable classes that aren't directly compatible with standard LLM Chain nodes in Flowise.

Possible Solution

Try using a Custom Function node to merge the outputs from both RAG flows:

1. Create a Custom Function node that accepts both runnable outputs

2. Extract the actual response data from each runnable class using their respective methods (like .invoke() or .run())

3. Combine the responses in your custom logic

4. Pass the merged result to an LLM Chain node for final response generation