Escalate angry Gorgias support tickets to Slack

When a new ticket is created in Gorgias, this trigger captures the ticket details and passes them to the next step for message retrieval. The trigger checks your Gorgias inbox hourly for new tickets, though this can be adjusted if you need more frequent monitoring. No configuration is needed—it automatically connects to your Gorgias account once you authorize MESA.

The step pulls all messages associated with the ticket ID from Step 1, including the initial customer message and any follow-up replies. It automatically extracts the message text in a format the AI can process in the next step. This is completely auto-configured and uses the ticket ID from the trigger to fetch the correct messages from Gorgias.

The AI receives the customer's message text and analyzes specific language patterns, word choices, and sentence structures that indicate anger or frustration. It outputs a confidence score from 1 to 5 (where 5 means highly confident the customer is angry) along with a brief explanation of what indicators it detected, like aggressive phrasing, urgent demands, or negative sentiment. The AI temperature is set to 0 for consistent, reproducible analysis without creative interpretation.

This code takes the AI's response text and converts it into a clean JSON object with separate fields for the confidence score and explanation. It parses the response string into a JavaScript object so the filter step can access the numeric confidence score directly. Here's the conversion code:

const Mesa = require('vendor/Mesa.js');

module.exports = new class {
  script = (prevResponse, context) => {
    // Retrieve the Variables Available to this step
    const vars = context.steps;
    // Get AI response
    let aiResponse = vars.ai.response;
    // Convert to JSON array
    let responseArray = JSON.parse(aiResponse);

    // Call the next step in this workflow
    // response will be the Variables Available from this step
    Mesa.output.next({ai_response: responseArray});
  }
}

The filter compares the AI's confidence score against your chosen threshold (default is 3 or higher). If the score meets or exceeds the threshold, the ticket continues to Slack; if not, the workflow stops here and the ticket stays in your normal support queue. You can adjust the threshold between 1 and 5 when setting up the workflow—set it higher (4-5) if you only want extremely angry messages escalated, or lower (2-3) if you want to catch more potentially upset customers.

The step pulls your Shopify shop information, specifically your store domain, which gets used to construct the Gorgias ticket URL in the next step. For example, if your Shopify store is "yourstore.myshopify.com", this step captures "yourstore" so the workflow can build the correct Gorgias link format. This is auto-configured and runs in the background.

You'll configure two things: which Slack channel receives the alerts (like #support-urgent or #customer-escalations), and your Gorgias Inbox ID. The Gorgias Inbox ID is found in your Gorgias URL when viewing your inbox—it's the numbers after "/apps/views/". The message format includes a clickable link to the ticket, the AI's confidence score, the AI's explanation of why it detected anger, and the customer's original message text. Make sure to invite the MESA Slack app to your chosen channel before selecting it.