Q Division Field Training: Module 200 - Dashboard Inconsistency

[Dalton_Ruer]

Solving the Mystery of Dashboard Numbers That Don't Add Up

🎯 Mission Briefing

Operatives, this is Dork 007 Dork reporting from Q Division headquarters. We have a situation that keeps every intelligence analyst up at night: Numbers that don't add up.

THE SITUATION: The original programmer for our Q Division application has left the company. A talented new designer built a beautiful dashboard, but there's a critical problem: Active Cases (16) + Closed Cases (15) = 31... but Total Cases shows 50. In an intelligence agency where no mission can be left behind, this kind of inconsistency is unacceptable.

YOUR MISSION: Use Qlik Answers to investigate the application and notes from the original programmer. In other words Structured Data -AND- Unstructured Data

DELIVERABLE: A dashboard that accurately reflects all case statuses with proper reconciliation, plus an understanding of how to use AI to diagnose and fix data model inconsistencies.

📦 Field Equipment

What You'll Need:

• Access to Qlik Answers with an existing assistant
• The Q Division application (uploaded in Mission 100 - attached here if needed)
• Approximately 15 minutes mission time

Video Intelligence Briefing: 🎥 Watch the Full Mission Walkthrough

🚀 Mission Steps

Step 1: Connect Your Application to Your Assistant

Navigate to your Q Division Field Academy catalog and locate your Field Training Assistant. This time, instead of adding another knowledge base, we're going to add structured data from your application itself.

Click to add content to your assistant, but instead of selecting a knowledge base, choose your Q Division application. Make sure you're in your Q Division Field Academy space and select the application you uploaded in Mission 100.

Pro tip: Once configured, your assistant can now query both unstructured data (knowledge bases with PDFs) AND structured data (your application's data model). This hybrid capability is what makes Qlik Answers so powerful for investigative work.

Step 2: Ask the Diagnostic Question

Now comes the moment of truth. Ask your assistant:

"Please tell me the total number of cases we have, the total closed cases we have, and the total open cases we have. Can you help me understand why there seems to be an inconsistency with the totals? Because what I see in my application seems wrong. The programmer's no longer with our company."

Watch the multi-agent collaboration unfold in the reasoning panel:

Step 3: Observe the Agent Reasoning Flow

⚠️ FIELD NOTE: Pay close attention to this workflow - it's the heart of how Qlik Answers investigates data issues.

1. Answer Agent Activates: First on the scene, determining what data sources might contain relevant information about case statuses.

2. Semantic Agent Searches: Scans your application's data model to find the most relevant fields. It identifies perfect matches: Total Cases, Closed Cases, Open Cases, and the Status field.

3. Data Analyst Agent Plans: Creates a step-by-step analysis strategy to break down the case counts by status.

4. Chart Agent Visualizes: Builds a visualization showing the breakdown: Resolved, Closed, Escalated, In-Progress, and Open cases.

5. Answer Agent Synthesizes: Combines all findings and delivers the verdict.

Step 4: Analyze the Discovery

The Answer Agent reveals the mystery: there's a hidden "Resolved" status in your data model containing 19 cases!

The complete breakdown:

• Resolved: 19 cases (the missing number!)
• Closed: 15 cases
• Active (Open + In-Progress + Escalated): 16 cases
• Total: 50 cases ✓

The Explanation: "Resolved" represents cases where agents completed their work, but haven't yet received client confirmation to officially close them. It's essentially a holding status between active work and final closure.

The assistant even suggests intelligent follow-up questions like "What is the breakdown of the 19 resolved cases by priority level?" or "How many days on average do cases spend in resolved status before closing?"

Step 5: Request the Dashboard Fix

Now that you understand the problem, ask your assistant to fix it:

"Can you add a KPI to my dashboard to show me the number of resolved cases so that our totals will look correct?"

Watch another beautiful agent collaboration:

1. Answer Agent recognizes the request to add a KPI visualization
2. Semantic Agent finds the appropriate field: Number of Resolved Cases
3. Dashboard Authoring Agent checks for existing relevant visualizations
4. Chart Agent creates the KPI specification
5. Data Analyst Agent designs the analysis package (single metric, no dimensions)
6. Dashboard Authoring Agent adds the KPI showing "19 Resolved Cases" to your dashboard

Step 6: Verify and Finalize

Return to your Q Division application. The AI has created a new sheet called "Case Status Dashboard" with the Resolved Cases KPI displaying the value 19.

Important: Since your original dashboard is a public sheet, the AI created a new private sheet rather than modifying the public one. To complete the mission, you'll need to manually:

1. Make the public dashboard private (or create a working copy)
2. Copy the Resolved Cases KPI from the AI-generated sheet
3. Add it to your main dashboard
4. Style it to match your existing KPIs

Now your dashboard shows: Closed (15) + Resolved (19) + Active (16) = Total (50) ✓

✅ Mission Debrief

What You've Accomplished:

✓ Connected an application to your Qlik Answers assistant

✓ Used Qlik Answers to diagnose a dashboard inconsistency using programmer documentation

✓ Discovered a hidden status field causing reconciliation issues

✓ Generated a new KPI visualization through natural language

✓ Understood multi-agent workflows combining semantic search, data analysis, and dashboard authoring

Validation Check: Can your dashboard now reconcile to 50 total cases with all statuses visible? If yes, mission accomplished!

Challenge Exercise (Optional): Ask your assistant to build a text object for your dashboard that explains the case status values. 

📚 Field Notes: Technical Deep-Dive

For this mission, the Qlik Answers Assistant needed access to both the structured application and it's data as well as the unstructured programmer notes. 

Understanding the Multi-Agent Specialist System

This mission showcased all of the agentic agents that Qlik Answers uses:

Answers Agent: Orchestrates the entire process: Decomposes the questions users ask into sub-tasks, and presents final findings back to users.

Knowledge Agent: Searches through vector database to get relevant unstructured data

Semantic Agent: Expert at understanding your data model's structure, field names, and relationships. It's like a data dictionary that speaks human language.

Data Analyst Agent: Designs analysis packages that specify what data to use, how to analyze it and what output best answers the question(s).

Chart Agent: Specializes in visualization design after receiving the analysis package from the data analyst agent. 

Dashboard Authoring Agent: Consumes outputs from the answer agent, data analyst agent or chart agent, investigates if existing required charts already exist.

Each agent is optimized for its specific task. When they collaborate, you get both depth and accuracy.

Why This Mission Matters for Real Organizations

The scenario in this mission - programmer leaves, new designer inherits an application, numbers don't reconcile - happens constantly in enterprises. Traditional approaches require:

• Digging through data model documentation (if it exists)
• Manually exploring every table and field
• Writing SQL queries to test hypotheses
• Rebuilding institutional knowledge from scratch

With Qlik Answers, you simply ask what's wrong and let the AI investigate the data model itself. The time savings are enormous, but more importantly, you don't lose critical business logic when people change roles.

The "Resolved" Status Pattern

This specific scenario - a holding status between "active" and "closed" - appears in many business processes:

• Support tickets: "Pending Customer Response"
• Sales opportunities: "Verbal Commitment" (before contract signature)
• Project tasks: "Delivered for Review" (before final approval)
• Purchase orders: "Goods Received" (before payment processing)

These intermediate states are often invisible in executive dashboards, leading to the exact reconciliation problems we solved in this mission. The pattern repeats across industries because workflows are rarely binary (open/closed) - they have nuanced intermediate states that matter for operations but get overlooked in reporting.

🎬 That's a Wrap, Operatives

You've successfully completed Field Training Mission 200 and learned how to use Qlik Answers as your data detective. As I mentioned in the video, I honestly never suspected a day when AI could auto-transcribe programmer notes, let alone use those notes alongside the application itself to solve data mysteries. But here we are, operatives. The future of analytics isn't just faster dashboards - it's intelligent systems that understand your data as well as the people who built it.

Questions? Feedback? Have you ever encountered a situation like this in the past where you had to take over maintenance of an application and something didn't add up?