Question and Answer Chain node common issues

Here are some common errors and issues with the Question and Answer Chain node and steps to resolve or troubleshoot them.

No prompt specified error

This error displays when the Prompt is empty or invalid.

You might see this in one of two scenarios:

1. When you've set the Prompt to Define below and have an expression in your Text that isn't generating a value.
   • To resolve, enter a valid prompt in the Text field.
   • Make sure any expressions reference valid fields and that they resolve to valid input rather than null.
2. When you've set the Prompt to Connected Chat Trigger Node and the incoming data has null values.
   • To resolve, make sure your input contains a chatInput field. Add an Edit Fields (Set) node to edit an incoming field name to chatInput.
   • Remove any null values from the chatInput field of the input node.

A Retriever sub-node must be connected error

This error displays when n8n tries to execute the node without having a Retriever connected.

To resolve this, click the + Retriever button at the bottom of your screen when the node is open, or click the Retriever + connector when the node isn't open. n8n will then open a selection of possible Retrievers to pick from.

Can't produce longer responses

If you need to generate longer responses than the Question and Answer Chain node produces by default, you can try one or more of the following techniques:

• Connect a more verbose model: Some AI models produce more terse results than others. Swapping your model for one with a larger context window and more verbose output can increase the word length of your responses.
• Increase the maximum number of tokens: Many model nodes (for example the OpenAI Chat Model) include a Maximum Number of Tokens option. You can set this to increase the maximum number of tokens the model can use to produce a response.
• Build larger responses in stages: For more detailed answers, you may want to construct replies in stages using a variety of AI nodes. You can use AI split up a single question into multiple prompts and create responses for each. You can then compose a final reply by combining the responses again. Though the details are different, you can find a good example of the general idea in this template for writing a WordPress post with AI.