AI-Powered Email Assistant with Human Review

Unlock Efficient Email Management with this AI Agent

This n8n workflow acts as an intelligent email assistant. It connects to your IMAP email server, fetches new emails, and converts them to markdown for optimal AI processing. The AI Agent then uses a Langchain summarization chain (powered by models like DeepSeek) to create a concise summary. Based on this summary and the original email, a Langchain agent leveraging OpenAI's GPT models (like GPT-4o-mini) drafts a professional reply. Crucially, this draft is first sent to a designated human reviewer via email. Only upon approval is the AI-generated response automatically sent to the original sender. This AI Agent empowers you to manage email overload efficiently while maintaining full control over outbound communications.

Key Features & Benefits

• AI-Driven Summarization: Automatically condenses incoming emails using Langchain and LLMs (e.g., DeepSeek), saving you reading time.
• Intelligent Reply Drafting: Generates context-aware, professional email replies using OpenAI's advanced models (e.g., GPT-4o-mini) through a Langchain agent.
• Human-in-the-Loop Approval: Ensures all AI-generated replies are reviewed and approved by a human before sending, maintaining quality and brand voice.
• Automated Email Handling: Fetches emails via IMAP, processes them, and sends replies via SMTP, reducing manual effort.
• Optimized for AI: Converts HTML emails to Markdown for improved understanding by LLMs.
• Customizable AI Behavior: System prompts for both summarization and reply generation can be tailored to specific needs.
• Increased Productivity: Significantly speeds up email response times and helps manage high email volumes.

• An n8n instance (Cloud or self-hosted).
• OpenAI API Key (e.g., for GPT-4o-mini).
• DeepSeek API Key (for the deepseek-chat model used in summarization).
• IMAP and SMTP credentials for the email account to be monitored and used for sending.
• Email address for receiving approval requests.

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure the 'Email Trigger (IMAP)' node with your IMAP server details and credentials.
4. In the 'Email Summarization Chain' node, ensure the connected 'OpenAI Chat Model' (this node might be generically named but is configured for DeepSeek) is set up with your DeepSeek API Key and the deepseek-chat model identifier.
5. In the 'Write email' Langchain Agent node, ensure the connected 'OpenAI' Chat Model node (this node might be generically named 'OpenAI' or similar) is configured with your OpenAI API Key and the desired model (e.g., gpt-4o-mini).
6. Customize the system prompts within the 'Email Summarization Chain' node (specifically its prompt parameters for prompt and combineMapPrompt) and the 'Write email' Langchain Agent node (systemMessage parameter) to tailor the AI's behavior.
7. Update the 'To Email' field in the 'Approve Email' node to your email address (or the reviewer's email address) for receiving approval requests.
8. Configure the 'Send Email' node (connected after the 'Approved?' node) with your SMTP server details and credentials for sending the final AI-assisted reply.
9. Review and adjust expressions in the final 'Send Email' node for subject (e.g., Re: {{ $('Email Trigger (IMAP)').item.json.subject }}), recipient ({{ $('Email Trigger (IMAP)').item.json.from }}), and sender ({{ $('Email Trigger (IMAP)').item.json.to }}) to ensure they match your email setup (e.g., the sender email might be a fixed address).
10. Activate the workflow and test with a sample email.