Multilingual Metadata Generation for Content

**Email Categorization (Example #1):

• The On new email to nathan's inbox (Gmail Trigger) node can be enabled to receive emails.
• The OpenAI Chat Model1 (GPT-4o) is used by the Assign label with AI node to classify the email text.
• Based on the classification, Add automation label or Add music label (Gmail nodes) apply the relevant labels.
• The Webhook and Whether email contains n8n nodes demonstrate a separate path for handling incoming webhook data, potentially related to email processing or other triggers.

**PDF Question Answering (Example #2):

• The PDFs to download node serves as a placeholder for PDF data.
• Download PDF (HTTP Request) fetches the PDF.
• Recursive Character Text Splitter breaks the PDF text into manageable chunks.
• Embeddings OpenAI generates embeddings for the text chunks.
• Insert into Pinecone vector store stores these embeddings and associated metadata (document title, publish year, author) in Pinecone.
• Read Pinecone Vector Store retrieves relevant information based on a query.
• OpenAI Chat Model (GPT-4o) and Vector Store Retriever are used by Question and Answer Chain to answer user queries based on the PDF content.

**Appointment Booking (Example #3):

• The When chat message received node (Chat Trigger) initiates the appointment booking process.
• The Appointment booking agent orchestrates the interaction.
• It uses Anthropic Chat Model for conversational AI and Window Buffer Memory to maintain context.
• Get calendar availability (Google Calendar API Tool) checks for free slots.
• Book appointment (Google Calendar API Tool) creates the event if a slot is available.
• The agent is configured with a systemMessage providing instructions and context.