AI Support Triage: Auto-Tag and Prioritize Front Inbox Emails

This workflow turns your Front support inbox into a self-organizing queue. Every inbound email is read by an OpenAI model that infers the topic (billing, bug, how-to, churn risk), the urgency, and the customer's sentiment. The workflow then writes those labels back to the Front conversation as tags, sets a priority, assigns it to the correct teammate or team inbox, and posts a one-line summary so an agent knows what they're opening before they click. High-severity or angry messages get escalated to a Slack channel immediately so nothing critical sits unseen.

Use it when your shared support inbox has outgrown manual triage — agents are cherry-picking easy tickets while urgent ones age, SLAs slip, and no one has a reliable read on what's flooding in. It's ideal for teams on Front who want consistent, rules-free categorization (the model adapts to new product language without you maintaining keyword filters) and want VIP or at-risk customers surfaced the moment they write in, not at the next standup.

1. 1A new inbound message in a Front inbox fires the trigger via Front's webhook, passing the subject, body, and sender.
2. 2The email text is sent to OpenAI with a structured prompt that returns JSON: `topic`, `priority` (low/medium/high/urgent), `sentiment`, and a one-sentence summary.
3. 3A logic step maps the model's output to your taxonomy and decides routing — e.g. `urgent` + negative sentiment routes to the on-call team and flags for escalation.
4. 4The workflow applies the tags, sets the priority, assigns the conversation, and adds the summary as an internal comment in Front.
5. 5If the message was flagged urgent or churn-risk, it posts an alert with the summary and a deep link to the Slack escalation channel.