Key information

• Transparency across channels – Review interactions handled by your AI Agent in chat, email, and phone from one place.
• Explainability built-in – Access transcripts of customer–AI conversations and, when needed, switch to a detailed view showing the AI’s reasoning, knowledge used, and process execution.
• Customer feedback signals – Collect and review thumbs up/down and sentiment analysis on every message to understand customer reactions at scale.
• Contact reason assignment – Automatically categorize interactions by customer-defined reasons to uncover why customers are reaching out.
• Quality at scale – Supervisor surfaces errors, failed processes, and negative sentiment automatically, making it easier to spot problems across thousands of interactions.

---

Using Supervisor in practice

Supervisor is most valuable when you want to understand how your AI Agent is performing and take action to improve it. Here are some practical use cases:

• Investigating customer complaints – Open the specific interaction to see exactly what was said, how the AI reasoned, and why it responded the way it did.
• Spotting failed processes – Supervisor highlights automation errors, helping your team pinpoint what went wrong and fix it faster.
• Tracking sentiment trends – Identify intents or processes that cause frustration and prioritize improvements.
• Collecting customer feedback – Thumbs up/down signals and optional feedback forms give direct insight into whether answers were useful.
• Understanding contact reasons – Automatically assigned categories show why customers are reaching out most often.
• Monitoring automation quality at scale – Supervisor surfaces problematic interactions so teams can focus on where it matters most.

---

Interaction List View and Interaction View

Supervisor gives you two ways to explore conversations handled by your AI Agent:

• Interaction List View – A searchable, filterable list of all AI-handled interactions across channels. Use it to quickly find conversations by channel, contact reason, sentiment, thumbs feedback, or automation status.
• Interaction View – A detailed look at a single interaction. Read the transcript, see thumbs and sentiment signals, check assigned contact reasons, and — when needed — switch to the reasoning view to understand why the AI Agent responded the way it did.

---

Filters in Supervisor

Supervisor includes filters that let you quickly narrow down interactions to review.
Filters can be combined to answer specific questions, such as:
“Show me all closed conversations about returns where the AI Agent used the refund intent.”

---

Available filters

• Status – Filter by how the interaction ended:

  • Deflected – Resolved fully by the AI Agent.
  • Ongoing – Still in progress.
  • Transferred – Handed off to a human agent.
  • Closed – Conversation ended.

• Contact reason – See only interactions assigned to a specific reason (e.g., “Order status,” “Returns”).

• Channels – Limit to a channel: chat, email, or phone.

• Interaction ID / Conversation ID – Look up a specific case if you have the ID.

• Intents – Filter by which intents were triggered during the interaction.

• Processes – Find interactions where specific processes (e.g., order lookup, refund initiation) were executed.

• Knowledge – Show interactions in which specific knowledge policy was used.

---

How filters work

• Apply one or multiple filters together (e.g., Deflected + Intents: Order status).
• The Interaction List View updates instantly with matching cases.
• Clear filters at any time to return to the full list.

---

Contact Reasons

Contact Reasons in Supervisor allow you to categorize and analyze why customers are reaching out. This makes it easier to spot trends, understand customer needs, and identify which topics are driving the highest volumes or automation challenges.

Key information

• Configurable by you – Define the categories (e.g., “Order status,” “Damaged package,” “Returns”) that reflect your business needs.
• Assigned at interaction close – Contact reasons are selected when a conversation ends.
• Filter and analyze – Filter interactions in Supervisor by contact reason to quickly find and review cases.
• Historic integrity – Past interactions retain old categories even if they are renamed or deleted.
• Combine with other signals – Use contact reasons with handoff status, thumbs up/down, or sentiment to pinpoint problem areas.

How to configure

1. Go to Supervisor.
2. In the upper-right corner, open Contact Reason settings.
3. Add new contact reasons that match your customer request categories.
4. Remove or rename old ones if your needs change (past interactions keep their old labels).
5. Save changes — new conversations will use the updated list immediately.

Practical examples

• Track how many conversations are about returns versus order tracking.
• Identify which contact reasons most often result in a handoff.
• Filter by negative sentiment + contact reason to find where automation underperforms.

---

Thumbs Up / Down

Thumbs Up / Down is the simplest way to collect direct feedback from your customers about AI Agent responses. Every message generated by the AI Agent can be rated, giving you a clear signal of what’s working well and where improvements are needed.

Key information

• Customer-driven signal – End users provide feedback directly in the conversation by clicking 👍 or 👎.
• Message-level granularity – Feedback applies to individual AI responses.
• Visible in Supervisor – Ratings appear alongside transcripts for context.
• Filterable – Filter for all 👎 responses to focus on problematic cases.
• Combined insights – Pair thumbs feedback with contact reasons, handoff status, or sentiment to identify deeper trends.
• Compatible with Widget – Thumbs Up / Down is supported only in Zowie Widget (Chat).

How to configure

1. Go to AI Agent → Channels → Zowie Chat.
2. Open the Advanced tab.
3. Ensure Thumbs Up / Down feedback is enabled.
4. Save changes — the feedback option will now appear in your website widget.
5. Monitor results in Supervisor by filtering for 👍 or 👎.

Practical examples

• Identify where 👎 feedback clusters around a specific intent and retrain.
• Validate new intents or processes by tracking 👍 rates after launch.
• Compare 👍/👎 distribution with handoff data to see if customers are more likely unhappy when automation escalates too late.

---

Sentiment

Sentiment analysis in Supervisor helps you understand how customers feel during their interactions with the AI Agent. It detects whether messages express positive, neutral, or negative sentiment, giving you an extra layer of context beyond the words themselves.

Key information

• Automatic detection – Every customer message is tagged with sentiment.
• Three levels – Positive 😊, Neutral 😐, and Negative 😡.
• Visible in Supervisor – Sentiment markers appear directly in transcripts.
• Filterable – Quickly review all negative or positive conversations.
• Trend analysis – Combine with automation outcomes to see if certain intents consistently generate negative experiences.

How to use it

• No manual configuration required.
• Open Supervisor and filter by sentiment (positive, neutral, negative).
• Review transcripts to understand why certain messages triggered the sentiment.
• Cross-check with thumbs or handoff for a fuller picture.
• Use insights to refine AI Agent responses, improve intents, or adjust escalation logic.

---

Customer Satisfaction (CSAT)

Customer Satisfaction (CSAT) in Supervisor helps you understand customer satisfaction on a scale from 1 to 5 (unsatisfied to satisfied) that is collected through Zowie's new Widget at the end of interactions. Surveys are collected when the user ends the interaction or minimizes the chat window.

Key information

• Automatic collection – Surveys are displayed to users automatically in crucial points of conversation.
• On a scale from 1-5 – Measured from 1 to 5 (unsatisfied to satisfied.)
• Additional reasons – For all scores 1 to 3 users are prompted questions for additional context for negative sentiment.
• Widget compatible – Currently satisfaction scores are only collected through Zowie Widget (new and legacy widget are supported).

How to use it

• To enable Customer Satisfaction (CSAT) collection head over to Channels > Zowie Chat > Advanced and enable the toggle next to "Chat CSAT survey."
• Open Supervisor and filter by CSAT (you can select scored from 1 to 5).
• Review transcripts to understand why certain interactions resulted in positive / negative sentiment.
• Cross-check with thumbs, sentiment, transfers for a fuller picture.
• Use insights to refine AI Agent responses, improve intents, or adjust escalation logic.