n8n AI Deep Research Agent Workflow

This n8n workflow enables you to host your own AI deep research agent, leveraging n8n's capabilities to orchestrate AI models and other services. It's designed to process research requests, generate detailed output, and manage the flow of information.

What it does

This workflow automates the following steps:

1. Receives Research Requests: It starts by listening for incoming research requests, likely via an n8n form submission or another workflow execution.
2. Initial Data Processing: It processes the incoming data, potentially setting or modifying fields to prepare for the AI research phase.
3. Conditional AI Model Selection: It intelligently decides which AI model to use (e.g., OpenAI Chat Model or Google Gemini Chat Model) based on predefined conditions in the input data.
4. Executes AI Research Chain: It utilizes a "Basic LLM Chain" to perform the core AI research, interacting with the selected AI chat model.
5. Parses AI Output: The output from the AI model is then processed by a "Structured Output Parser" to extract and format the relevant information.
6. Handles Successful Research: If the AI research is successful, the output is formatted into Markdown and can be stored or further processed (e.g., saved to Notion).
7. Handles Research Failures: If the AI research encounters an error, the workflow captures the error and stops, providing clear indication of failure.
8. Aggregates and Splits Data: It includes nodes for aggregating and splitting data, suggesting that it can handle multiple research items or complex data structures, processing them in batches if necessary.
9. Sub-workflow Execution: The presence of "Execute Sub-workflow" and "Execute Workflow Trigger" nodes indicates that this workflow can be part of a larger, modular system, allowing for specialized sub-tasks to be delegated to other workflows.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running instance of n8n.
• OpenAI API Key: For using the OpenAI Chat Model.
• Google Gemini API Key: For using the Google Gemini Chat Model.
• Notion Account (Optional): If you intend to save the research output to Notion, you'll need a Notion API integration and database set up.
• Credentials: Appropriate credentials configured in n8n for OpenAI, Google Gemini, and Notion (if used).

Setup/Usage

1. Import the Workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Set up your OpenAI API Key credentials.
   • Set up your Google Gemini API Key credentials.
   • (Optional) Set up your Notion API credentials if you plan to use the Notion node.
3. Configure Nodes:
   • "On form submission" (Trigger): Customize the form fields to match the input you expect for your research requests.
   • "Edit Fields" (Set): Adjust any initial data manipulation as needed.
   • "Switch": Review and modify the conditions that determine which AI model (OpenAI or Google Gemini) will be used based on your input data.
   • "Basic LLM Chain": Configure the chain's prompts and parameters according to your research needs.
   • "OpenAI Chat Model" / "Google Gemini Chat Model": Select the specific model versions and any other relevant settings.
   • "Structured Output Parser": Define the schema for the expected output from your AI model to ensure correct parsing.
   • "Markdown": Customize the Markdown formatting for the final output.
   • "Notion" (Optional): Configure the Notion database ID and properties where the research results should be stored.
   • "Stop and Error": Review the error handling logic, especially if you want custom error notifications.
4. Activate the Workflow: Once configured, activate the workflow to start processing research requests.
5. Trigger the Workflow: Submit data to the "On form submission" trigger, or execute the workflow manually with test data to verify its functionality.