Monitor Customer Risk and AI Feedback using PostgreSQL, Gmail, and Discord

This n8n workflow automates the process of monitoring customer risk by fetching data from a PostgreSQL database, analyzing it, and then sending alerts via Gmail and Discord based on predefined risk criteria. It also incorporates a feedback loop for human review.

What it does

This workflow performs the following key steps:

1. Scheduled Trigger: The workflow is initiated on a schedule (e.g., daily, hourly) to regularly check for new customer data.
2. Fetch Customer Data from PostgreSQL: It queries a PostgreSQL database to retrieve customer information.
3. Process and Transform Data: The fetched data is then processed and transformed using a "Set" node, likely to format it for subsequent analysis or to extract relevant fields.
4. Loop Through Customer Records: The workflow iterates through each customer record, allowing individual processing.
5. Analyze Customer Risk (HTTP Request): For each customer, an HTTP request is made, likely to an external AI service or an internal API, to assess customer risk based on the provided data.
6. Conditional Logic for Risk Assessment:
   • High Risk: If the AI feedback indicates a "High Risk" customer, the workflow proceeds to send an alert.
   • Low Risk: If the AI feedback indicates "Low Risk", the workflow merges these items and potentially logs them or takes no further action in this specific flow.
7. Send High-Risk Alert via Gmail: For high-risk customers, an email alert is sent via Gmail, likely to a risk management team.
8. Send High-Risk Alert via Discord: Simultaneously, a notification is posted to a Discord channel, ensuring immediate visibility for the relevant team.
9. Human-in-the-Loop (Wait Node): After sending alerts, the workflow includes a "Wait" node, which could be used to pause the workflow, awaiting manual review or acknowledgment of the high-risk customer.
10. Record AI Feedback in Google Sheets: The AI feedback for each customer (both high and low risk) is recorded in a Google Sheet, providing a centralized log for review and auditing.
11. Code Execution for Further Processing (Optional): A "Code" node is included, which can be used for custom JavaScript logic, such as further data manipulation, logging, or integration with other systems based on the AI feedback.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance.
• PostgreSQL Database: Access to a PostgreSQL database containing customer data. You will need the necessary credentials (host, port, database, user, password).
• Gmail Account: A configured Gmail credential in n8n to send email alerts.
• Discord Account/Webhook: A Discord credential or webhook URL configured in n8n to post messages to a Discord channel.
• Google Sheets Account: A Google Sheets credential in n8n to write AI feedback data.
• External AI Service/API: An endpoint for an HTTP Request node that can analyze customer data and return a risk assessment (e.g., "High Risk", "Low Risk").

Setup/Usage

1. Import the workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Set up your PostgreSQL credentials in n8n.
   • Set up your Gmail credentials in n8n.
   • Set up your Discord credentials (or webhook) in n8n.
   • Set up your Google Sheets credentials in n8n.
3. Update Node Configurations:
   • Postgres Node: Configure the "Postgres" node to connect to your database and specify the query to fetch customer data.
   • HTTP Request Node: Update the URL and any necessary headers/body for your AI risk assessment service.
   • If Node: Adjust the conditions in the "If" node if your AI service returns different risk classifications.
   • Gmail Node: Customize the recipient, subject, and body of the email alert for high-risk customers.
   • Discord Node: Customize the message content and target channel for Discord alerts.
   • Google Sheets Node: Specify the Spreadsheet ID and sheet name where AI feedback should be recorded.
   • Schedule Trigger: Adjust the schedule to your desired frequency for monitoring.
   • Wait Node: Configure the duration for the "Wait" node if human intervention is required.
   • Code Node: Implement any custom JavaScript logic as needed.
4. Activate the Workflow: Once all configurations are complete, activate the workflow.