n8n Workflow: Basic Langchain AI Agent with OpenAI

This n8n workflow demonstrates a foundational setup for an AI agent using Langchain nodes, integrating with OpenAI for language modeling and embeddings. It includes components for data loading, text splitting, and a simple in-memory vector store, providing a starting point for more complex AI applications.

What it does

This workflow showcases a basic Langchain AI agent configuration, primarily focusing on the AI components rather than a full end-to-end process. It includes:

1. Scheduled Trigger: Initiates the workflow on a defined schedule.
2. Default Data Loader: A Langchain node for loading data, likely from a specified source (configuration not visible in JSON, but implied by node type).
3. Recursive Character Text Splitter: Processes the loaded data by splitting it into smaller, manageable chunks based on character count.
4. Embeddings OpenAI: Generates vector embeddings for the text chunks using the OpenAI API.
5. Simple Vector Store: Stores the generated embeddings in an in-memory vector database.
6. OpenAI Chat Model: Configures an OpenAI chat language model (e.g., GPT-3.5, GPT-4).
7. Simple Memory: Provides a basic memory buffer for the AI agent to maintain context within a conversation or task.
8. Structured Output Parser: A Langchain node designed to parse the output from the AI model into a structured format.
9. AI Agent: The core Langchain agent that orchestrates the use of the language model, memory, and tools (like the vector store) to perform tasks.
10. Edit Fields (Set): A utility node to modify or add fields to the data, likely for preparing data for subsequent steps or for output.
11. HTTP Request: A generic node for making HTTP calls, potentially to an external API or service.
12. Google Sheets: A node for interacting with Google Sheets, which could be used for inputting data, logging results, or managing research paper metadata.
13. If: A conditional node for branching the workflow based on certain criteria.
14. Merge: A utility node to combine data streams from different branches of the workflow.
15. Wait: A utility node to pause the workflow for a specified duration.
16. Sticky Note: A documentation node for adding comments or notes within the workflow.

While the directory name suggests "draft and manage academic research papers with GPT-4 and Pinecone," the provided JSON primarily outlines the AI agent components (OpenAI, Langchain tools, in-memory vector store) and general utility nodes (Google Sheets, HTTP Request, If, Merge, Wait, Set). It does not explicitly define the logic for paper drafting, management, or Pinecone integration, focusing more on the underlying AI setup.

Prerequisites/Requirements

• n8n Instance: A running n8n instance (self-hosted or cloud).
• OpenAI API Key: Required for the Embeddings OpenAI and OpenAI Chat Model nodes.
• Google Sheets Account: If the Google Sheets node is intended for active use.
• Langchain Nodes: Ensure the @n8n/n8n-nodes-langchain package is installed in your n8n instance.

Setup/Usage

1. Import the Workflow:
   • Copy the provided JSON content.
   • In your n8n instance, go to "Workflows" and click "New".
   • Click the three dots menu (...) in the top right and select "Import from JSON".
   • Paste the JSON content and click "Import".
2. Configure Credentials:
   • Locate the Embeddings OpenAI and OpenAI Chat Model nodes.
   • Configure your OpenAI API Key credentials for these nodes.
   • If using the Google Sheets node, configure your Google Sheets credentials.
3. Customize Nodes:
   • Schedule Trigger: Adjust the schedule as needed for your use case.
   • Default Data Loader: Configure this node to load your specific research paper data (e.g., from a file, URL, database).
   • Recursive Character Text Splitter: Adjust chunk size and overlap parameters if necessary for optimal text processing.
   • OpenAI Chat Model: Select the desired OpenAI model (e.g., gpt-4, gpt-3.5-turbo) and adjust temperature or other parameters.
   • AI Agent: Define the agent's prompt, tools, and memory based on the specific research paper tasks you want it to perform.
   • Edit Fields (Set): Modify this node to transform data as required.
   • HTTP Request: Configure the URL, method, and body for any external API calls.
   • Google Sheets: Specify the spreadsheet ID, sheet name, and operations (e.g., append, read, update) if you intend to use it.
   • If/Merge/Wait: Adjust the logic of these nodes according to your workflow's branching and timing requirements.
4. Activate the Workflow: Once configured, activate the workflow by toggling the "Active" switch in the top right corner.