This n8n workflow is designed to facilitate an AI-powered voice travel agent, enabling users to receive personalized tour and activity recommendations through voice commands. The workflow integrates several nodes, including webhooks, AI language models (GPT-4 variants), embedding models, and a vector database (Pinecone), to analyze user input, retrieve relevant tour packages, and generate detailed itineraries.

The process begins with a webhook that captures voice input or requests from users, which are then processed to extract travel details such as destination, trip duration, and passenger count. The workflow employs a tour recommendation AI agent equipped with a system prompt that considers tour timing, proximity, and meal schedules, generating suitable tour options.

Embedded tour data and activities are stored in and retrieved from a Pinecone vector database, enabling efficient similarity searches and fast responses. The AI models further analyze retrieved tours, combine data, and respond to queries, providing personalized travel itineraries. These are then sent back to the user via the webhook response, completing the voice-driven interaction.

This workflow is particularly useful for travel agencies or tour operators looking to automate personalized travel planning and enhance customer experience through intelligent, voice-activated services.

Node Count	6 – 10 Nodes
Nodes Used	@n8n/n8n-nodes-langchain.agent, @n8n/n8n-nodes-langchain.embeddingsOpenAi, @n8n/n8n-nodes-langchain.lmChatOpenAi, @n8n/n8n-nodes-langchain.toolVectorStore, @n8n/n8n-nodes-langchain.vectorStorePinecone, respondToWebhook, stickyNote, webhook