Skip to main content

AI Reports, Agent Summaries & Chat Audit Logs

Overview & AI Intelligence Architecture​

Artificial Intelligence transforms massive volumes of unstructured WhatsApp conversation transcripts into structured, high-value business intelligence. Rather than requiring operations managers or quality analysts to manually read hundreds of chat logs, Wabixo's native AI Intelligence Engine continuously synthesizes customer intent, detects service friction points, grades agent communication quality, and produces executive briefings.

The AI & Reports Suite (/dashboard/analytics?tab=ai-reports) contains four specialized capabilities:

  1. AI Chat Analysis (ai-reports): Macro-level conversational intelligence, sentiment distribution, topic clustering, and upselling opportunities.
  2. Dedicated Report Drilldown (/dashboard/analytics/report/:id): Full-screen, split-pane investigation workspace for deep-diving into AI-analyzed conversation themes with embedded chat preview.
  3. Agent Summary (agent-summary): AI-generated performance evaluations and coaching scorecards for individual staff members.
  4. Report History (agent-report-history): A permanent, searchable archive of previously compiled agent performance reports.
  5. Chat Summaries Log (chat-summaries): A line-by-line audit trail of per-conversation AI summaries with 3 KPI telemetry cards and 1-click links to the live Team Inbox.

1. AI Chat Analysis (ai-reports)​

The AI Chat Analysis engine processes aggregate conversation histories across a designated timeframe (e.g. past week or month) to identify top-level business trends and operational blindspots.

┌────────────────────────────────────────────────────────────────────────────────────────┐
│ EXECUTIVE AI CHAT ANALYSIS (PAST 7 DAYS)                                              │
├────────────────────────────────────────────────────────────────────────────────────────┤
│ • Total Analyzed Conversations: 1,420 transcripts (18,450 customer/agent messages)     │
│ • Dominant Customer Sentiment:  84.2% Positive  • 11.5% Neutral  • 4.3% Escalated      │
│ • Primary Service Drivers:      Festive Hair Spa Packages, Keratin Treatments          │
│ • Top Customer Friction Point:  Sunday afternoon booking slot unavailability           │
│ • AI Operational Guidance:      Open 2 additional stylist chairs on Sunday peak shifts │
└────────────────────────────────────────────────────────────────────────────────────────┘

Structured Intelligence Sections​

When the AI engine completes its synthesis, the report renders into six interactive sections:

  1. Executive Briefing: A high-level briefing summarizing customer interactions, campaign engagement, and general outcome trends.
  2. Sentiment Distribution: Visual sentiment scoring (Positive, Neutral, Critical/Escalated) across customer demographics.
  3. Hot Topics (with Linked Chats):
    • Automatically clusters conversations into business topics (e.g., "Keratin Pricing Confusion", "Bridal Package Inquiry").
    • Shows the exact chat count per topic.
    • Clicking any topic reveals the underlying conversations with customer names and direct links to open the chat.
  4. Key Customer Issues & Severity Ratings:
    • Highlights service failures, billing disputes, or delivery delays.
    • Categorized by severity badges: Critical (Red), High (Orange), Medium (Yellow), or Low (Blue).
    • Includes actual customer quotes illustrating the friction point.
  5. Upselling Opportunities:
    • Discovers conversations where customers expressed interest in additional services or retail products that agents failed to close.
  6. Actionable Recommendations:
    • Concrete, data-backed operational improvements for management (e.g., "Update the automated FAQ bot with weekend pricing rates").

Generating a New AI Analysis Report​

  1. Navigate to Analytics > AI & Reports > AI Chat Analysis.
  2. Select the Start Date and End Date for the analysis period.
  3. Click Generate Analysis.
  4. The backend streams progress through four status states: pending → processing → completed (or failed).

2. Dedicated AI Report Drilldown (/dashboard/analytics/report/:id)​

When an executive or supervisor clicks on any historical or newly completed AI Chat Analysis report, the system opens a dedicated full-screen investigation workspace at /dashboard/analytics/report/:id:

┌────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ AI REPORT DRILLDOWN: 2026-09-20 to 2026-09-27 (1,420 Chats Analyzed)                                  │
├──────────────────────────────┬──────────────────────────────┬──────────────────────────────────────────┤
│ THEMES & TOPIC CLUSTERS      │ STRUCTURED EXECUTIVE SUMMARY │ LIVE CONVERSATION INSPECTION             │
│ • Keratin Inquiry (142 chats)│ • Customer Intent Summary    │ Customer: Priya Sharma (+91 98765 *****) │
│ • Slot Unavailable (84 chats)│ • Key Discussion Points      │ ──────────────────────────────────────── │
│ • Pricing Confusion (45 chats│ • Friction Points Identified │ [14:22] Customer: Does the package       │
│ • Bridal Packages (32 chats) │ • Operational Recommendations│         include hair wash?               │
│                              │ • Next Steps for Supervisors │ [14:23] Agent: Yes, premium shampoo &   │
│ [Click topic to load chats]  │                              │         blow-dry included.               │
└──────────────────────────────┴──────────────────────────────┴──────────────────────────────────────────┘

Three-Pane Investigation Workflow​

  1. Left Pane (Themes & Clusters): Lists all categorized topics and customer friction issues, each annotated with exact conversation counts. Clicking a topic expands the underlying customer threads.
  2. Center Pane (Structured Summary): Formatted Markdown view breaking down Customer Intent, Key Discussion Points, Friction Points, and Next Actions.
  3. Right Pane (Embedded Chatbox Preview): Selecting any customer thread immediately embeds the live conversational thread, allowing supervisors to audit the exact messages without navigating away from the analysis.

3. Agent Summary Reports (agent-summary)​

The Agent Summary tool conducts objective, automated performance reviews of individual team agents by analyzing their conversation transcripts.

┌────────────────────────────────────────────────────────────────────────────────────────┐
│ AI PERFORMANCE REVIEW: ANANYA SHARMA                                                  │
├────────────────────────────────────────────────────────────────────────────────────────┤
│ • Evaluated Period:  Last 7 Days (85 chats analyzed)                                  │
│ • Overall Score:     9.2 / 10 (Exceptional Performance)                               │
│                                                                                        │
│ Core Competency Ratings:                                                              │
│ • Tone, Empathy & Politeness:  [██████████░]  9.5 / 10                                 │
│ • Product & Pricing Accuracy:  [█████████░░]  9.0 / 10                                 │
│ • Adherence to Company SOPs:   [██████████░]  9.4 / 10                                 │
│ • Upselling & Sales Drive:     [███████░░░░]  7.2 / 10 (Coaching Area)                 │
│                                                                                        │
│ Key Strength:   Swift, warm greeting; expertly defused an appointment delay complaint. │
│ Coaching Note:  Prompt customers for add-on hair care serums after package bookings.   │
└────────────────────────────────────────────────────────────────────────────────────────┘

Report Contents & Badging​

  • Hot Topics Handled: Breakdown of the customer inquiry types resolved by the agent.
  • Key Issues Identified: Any mistakes, incorrect pricing quotes, or missed follow-ups made during conversations.
  • Important Chats Requiring Attention:
    • Conversations flagged with Urgency Badges (critical, high, medium, low).
    • Clicking any flagged customer name opens the conversation thread in the Team Inbox in a new tab.
  • Upselling Log: Specific instances where the agent successfully upsold a customer, or missed an obvious buying signal.
  • Actionable Coaching Plan: Step-by-step guidance for supervisors to mentor the staff member.

Step-by-Step Generation Flow​

  1. Navigate to AI & Reports > Agent Summary.
  2. Choose the team member from the Select Agent dropdown.
  3. Choose the evaluation timeframe (e.g., Last 7 Days, Last 30 Days, or custom dates).
  4. Click Generate AI Summary Report.
  5. Once compiled, the report is permanently saved to the Report History Archive.

4. Report History Archive (agent-report-history)​

To avoid consuming AI processing quotas repeatedly for past evaluations, all previously compiled Agent Summary reports are permanently stored in the Report History Archive:

  • Searchable Database: Filter past reports by agent name, supervisor author, or date range.
  • Status Badges: Visual indicators show Done (green), Processing (blue), or Error (red).
  • Zero-Token Replay: Review complete historical scorecards instantly without incurring AI API charges.
  • Longitudinal Growth Tracking: Compare an agent's monthly evaluations to verify whether coaching recommendations improved response quality and customer ratings.
  • Download & Share: Print or copy report summaries for formal quarterly employee appraisals.

5. Chat Summaries Log (chat-summaries)​

Whenever a conversation is summarized (either automatically upon resolving a thread, or manually by an agent clicking Summarize Chat in the Team Inbox), the output is logged in the centralized Chat Summaries Log.

The 3 KPI Telemetry Cards​

At the top of the Chat Summaries Log, three summary cards track aggregate usage:

┌───────────────────────────────────┬───────────────────────────────────┬───────────────────────────────────┐
│ TOTAL SUMMARIES                   │ UNIQUE CUSTOMERS                  │ CONTRIBUTING AGENTS               │
│ 1,420                             │ 1,180                             │ 16                                │
│ [Cumulative summaries created]    │ [Distinct customer threads]       │ [Staff triggering summaries]      │
└───────────────────────────────────┴───────────────────────────────────┴───────────────────────────────────┘

Draft Filter Bar Controls​

Selection modifies local draft state, ensuring smooth user experience without triggering premature API requests:

  • Agent Selector: Filter summaries by a specific team member or All Agents.
  • Date Range Picker: Start and end date inputs with calendar selector.
  • Search Input: Searches across customer name, phone number, and generated summary text.
  • Execution Buttons: API requests are dispatched only when clicking Filter or Reset.
┌────────────────────────────────────────────────────────────────────────────────────────┐
│ CHAT SUMMARIES AUDIT LOG                                                               │
├────────────────────────────────────────────────────────────────────────────────────────┤
│ [👤 Agent: All ▼] [📅 Start Date - End Date] [🔍 Search Mobile/Name...] [Filter] [Reset] │
│                                                                                        │
│ Customer Contact   | Agent Assigned | Context Depth | Summary Snippet       | Actions  │
│ ------------------ | -------------- | ------------- | --------------------- | -------- │
│ Priya Sharma       | Rajesh Kumar   | last_50       | Inquired on Balayage; | 👁️ View  │
│ +91 98765 *****    |                |               | booked for Sat 3 PM.  | 🚀 Open  │
│ Amit Patel         | Ananya Sharma  | last_100      | Billing dispute on OTP| 👁️ View  │
│ +91 91234 *****    |                |               | payment; resolved.    | 🚀 Open  │
└────────────────────────────────────────────────────────────────────────────────────────┘

Context Depth Options​

Wabixo supports adjustable context windows when generating per-chat summaries:

  • last_50 (50 msgs): Fast summary for routine, short inquiries.
  • last_100 (100 msgs): Standard summary capturing inquiry and resolution.
  • last_150 (150 msgs): Extended depth for complex support disputes.
  • last_200 (200 msgs): Comprehensive multi-shift audit.

View Summary Dialog & Live Inbox Navigation​

  • Rich Markdown Preview Dialog: Clicking View (👁️) opens the full formatted summary with bold highlights, bullet points, intent classification, and customer sentiment badges.
  • Copy to Clipboard: 1-click copy of the entire summary formatted for CRM paste.
  • Open in Chat (🚀): Clicking the button navigates directly to /dashboard/teamInbox?chatId=<id>, loading the customer's live message thread.

6. Customer Support AI Agent Playbook & FAQs​

Use this knowledge base to resolve customer inquiries regarding AI analysis, agent summaries, and privacy.

Q1: "Why is the AI & Reports tab locked with a message asking to upgrade?"​

Resolution: AI conversational intelligence, sentiment clustering, and Agent Summary scorecards require our Connect Plan (₹1,799/mo) or Scale Plan. On the Lite and Launch plans, AI features are locked. Account owners can upgrade under Settings > Billing & Subscription.

Q2: "Can our customers see the AI chat summaries or agent performance scorecards?"​

Resolution: No. AI summaries and agent evaluations are 100% internal and visible only to authorized workspace staff with supervisor or admin permissions. No AI evaluations are ever sent to customers over WhatsApp.

Q3: "Does the AI support conversations conducted in Hindi or Hinglish?"​

Resolution: Yes. Our AI Engine includes native multi-language and transliteration intelligence. If an agent and customer conversed in Hinglish (e.g., "Package ka price kitna hai aur Sunday ko slot milega kya?"), the AI model understands the conversational context and outputs structured English summaries and scorecards accurately.

Q4: "Why does generating an AI Chat Analysis take 1 to 2 minutes?"​

Resolution: Unlike simple database queries, AI Chat Analysis processes thousands of conversation messages across dozens of customers. The AI engine reads the transcripts, removes noise, performs semantic clustering, extracts quotes, and computes sentiment scores before saving the report. A progress indicator tracks generation status.