Front and Slack: Negative-Sentiment Conversation Alert Template

A webhook-triggered Spojit workflow that reads an incoming Front conversation, judges its tone, and pings a Slack supervisor channel whenever a customer sounds frustrated.

What It Builds

A Webhook trigger fires when Front posts a conversation event to your workflow URL. An Agent-mode Connector node on the front connector reads the full conversation with get-conversation, weighs the customer's tone, and produces a structured verdict. A Condition node checks that verdict, and when the sentiment is negative, a Connector node on the slack connector uses send-message to alert your supervisor channel so a senior teammate can step in before the customer escalates.

The Prompt

Paste this into Miraxa and it builds the workflow, connecting the tools for you:

Build a workflow triggered by a Front webhook. Use an Agent-mode Connector node on the Front connector to read the incoming conversation by its id and return a structured verdict with a sentiment label of positive, neutral, or negative plus a one-line reason. Add a Condition node that, when the sentiment is negative, posts a message to the #support-escalations Slack channel with the customer name, the conversation subject, the reason, and a link to the conversation.

Connectors Used

  • Webhook trigger - Front posts a conversation event to your workflow URL, verified by a signing connection.
  • front - an Agent-mode Connector node reads the thread with get-conversation and judges its tone.
  • slack - a Direct-mode Connector node posts the alert with send-message.

Customize It

Swap #support-escalations for the channel your supervisors watch, and adjust the sentiment labels if you want a finer scale such as adding "urgent". You can change which fields land in the Slack message (for example add the inbox name or the customer's email), or tighten the trigger so only certain Front inboxes fire it. To raise the bar before alerting, ask the agent to only flag conversations where frustration is strong, not merely lukewarm.

Tips

  • Give the Agent-mode node a Response Schema so the verdict comes back as predictable JSON (a sentiment string and a reason string), which makes the Condition node easy to wire.
  • Keep the Slack step in Direct mode: posting one message is a deterministic single-tool call, so it spends no AI credits, unlike the reading-and-judging step.
  • Use Spojit's Slack signing scheme on the webhook's signing connection only if Front is proxied through Slack; otherwise pick the Custom scheme to verify Front's own signature.

Related

Did this answer your question? Thanks for the feedback There was a problem submitting your feedback. Please try again later.