Quiz: AI Nodes in Practice

1. What question should you ask before adding an AI node?

Is this the most cost-effective approach? Does this step require judgement, or does it require transformation? Has the AI been trained on this type of data? Will the AI node slow down the workflow?

2. Why is 'return only the label' critical for AI classification?

Long responses cost more tokens The Switch node needs a clean output to route on, not a paragraph of explanation Shorter responses are more accurate n8n has a character limit on node outputs

3. When should you use a local Ollama model in n8n?

For all AI tasks to save money Only for testing For classification tasks where data is sensitive and should never leave the machine Only when the internet is unreliable

Check Your Understanding

Check Your Understanding