Automatic contact sentiment

Omnifox can tag each contact with a mood (happy, satisfied, neutral, frustrated, or angry) based on their conversations, so you can prioritize the ones who need attention.

Jun 26, 2026

Sentiment is a mood label Omnifox attaches to each contact based on the tone of their conversations. It lets you see at a glance who's happy and who's upset, so you can prioritize at-risk customers.

The available values

A contact can have one of five sentiments, each with its own face:

  • Happy
  • Satisfied
  • Neutral
  • Frustrated
  • Angry

Frustrated and Angry are the ones that need attention - your early warning to prevent a churn or a bad review.

How it's assigned automatically

Sentiment is inferred by the AI agent as it handles or summarizes a contact's conversations. When the AI detects the dominant tone, it updates the contact's sentiment field without any action from you. For it to fill in automatically you need:

  1. An AI agent active on the channel the contact writes through.
  2. At least one conversation with enough messages to evaluate the tone.

As new conversations come in, the value updates to reflect the most recent state.

How to see it and filter by it

  1. Go to Contacts.
  2. Each contact card shows a colored badge with its current sentiment (it only appears once one is assigned).
  3. To focus on a group, use the Sentiment dropdown in the filter bar and pick, say, Angry. The All sentiments option clears the filter.

This way you can, for example, list every Frustrated contact and follow up with them first.

Verify it

Apply the Sentiment -> Angry filter. If the list returns contacts, the tagging is working. Open one and review its timeline: the tone of the latest messages should match the label.

If it doesn't work

  • No contact has a sentiment: there's probably no active AI agent, or the conversations are too short to evaluate the tone. Connect and enable an AI agent on the channel.
  • The label looks stale: sentiment reflects conversations that have already been processed; right after a very recent chat it may take a moment to recompute.
  • It doesn't match your expectation: it's an AI estimate - use it as a guide, not an absolute truth.

Tips

  • Combine the Sentiment filter with tags or lifecycle stages to spot, for instance, customers in the renewal stage who are Frustrated.
  • Build a workflow that notifies a human agent whenever a contact turns Angry.

Related articles

Was this helpful?

Related articles

Message us on WhatsApp