> For the complete documentation index, see [llms.txt](https://docs.pharos.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.pharos.ai/platform-overview/pharos-platform-architecture/ai-analysis/org-genie/issue-level.md).

# Issue Level

Use Org Genie in the context of a specific Issue to understand its root cause, investigate related changes, and learn why it started or stopped occurring.

From any Issue detail page, you can open Org Genie in the context of that specific Issue. This provides focused analysis about that particular error pattern, including metadata changes that may have caused or resolved it, without needing to describe the Issue repeatedly.

{% hint style="success" %}
**Context-Aware Investigation**

Issue-level Org Genie automatically understands which Issue you're investigating, allowing you to ask targeted questions like *"What changed around the time this started?"* without re-explaining the Issue each time.
{% endhint %}

***

## Overview

Issue-level Org Genie transforms how you investigate individual errors by maintaining the full context of that Issue throughout your conversation. The AI understands which Issue you're asking about, enabling natural, focused investigation without manually filtering or describing the problem repeatedly.

### Key Advantages

Investigating within an Issue's context offers several advantages over starting from scratch each time.

<table><thead><tr><th width="200">Feature</th><th>Benefit</th></tr></thead><tbody><tr><td><strong>🎯 Automatic Context</strong></td><td>The AI already knows which Issue you're investigating—no need to describe it in every question</td></tr><tr><td><strong>🔍 Focused Analysis</strong></td><td>Questions are automatically scoped to that Issue's context, metadata, and timeline</td></tr><tr><td><strong>📊 Relevant Metadata</strong></td><td>Analysis focuses on changes relevant to that specific error pattern, not org-wide changes</td></tr><tr><td><strong>💬 Persistent History</strong></td><td>Chat history is stored per-Issue—return to your investigation later and continue where you left off</td></tr><tr><td><strong>🔗 Direct Navigation</strong></td><td>Seamlessly move from org-level discovery to issue-level deep-dive with full context preserved</td></tr></tbody></table>

***

## Accessing Issue-Level Org Genie

### Step-by-Step Access

Opening issue-level Org Genie takes three steps.

1. Navigate to any Issue detail page in Pharos
2. Look for the **Org Genie** option (typically in the Issue actions or toolbar)
3. Click to open Org Genie in issue-specific context

{% hint style="info" %}
**Each Issue Has Its Own Chat**

Every Issue maintains its own dedicated Org Genie chat history. This means you can investigate multiple Issues simultaneously without conversations mixing, and you can return to previous investigations without losing context.
{% endhint %}

***

## What Makes Issue-Level Different from Org-Level

When you open Org Genie from an Issue detail page versus the main navigation, the experience is fundamentally different:

### Context Comparison

The same question behaves very differently depending on which context you ask it in.

{% tabs %}
{% tab title="🔍 Issue-Level Context" %}
**What's Different:**

* AI already knows which Issue you're investigating
* Questions automatically scoped to that Issue's timeline and metadata
* Metadata analysis focuses on changes relevant to that error pattern
* Chat history stored per-Issue (return anytime)
* Correlation focused on that Issue's lifecycle (when it started, spiked, or stopped)

**Example:**

```
You: "What changed around the time this started?"

Genie understands:
- "This" = The specific Issue you're viewing
- "Started" = The Issue's first occurrence timestamp
- Scope = Metadata changes near that timestamp relevant to the error
```

No need to specify Issue ID, date ranges, or error type—it's all contextual.
{% endtab %}

{% tab title="🌐 Org-Level Context" %}
**What's Different:**

* AI has no specific Issue in mind
* Questions require you to specify which Issues or time periods
* Metadata analysis covers org-wide changes
* Chat history stored per-user (your personal workspace)
* Correlation spans multiple Issues and broader patterns

**Example:**

```
You: "What changed around the time this started?"

Genie needs clarification:
- "This" = Which Issue? Need Issue ID or description
- "Started" = When? Need date or time period
- Scope = What metadata? All changes or specific components?
```

You must provide context explicitly.
{% endtab %}
{% endtabs %}

{% hint style="success" %}
**Use the Right Context**

* **Issue-Level:** When you know which Issue to investigate and want deep understanding
* **Org-Level:** When you're exploring, prioritizing, or comparing multiple Issues
  {% endhint %}

***

## Example Queries

Issue-level Org Genie excels at questions about the specific Issue's lifecycle, root cause, and correlation with changes.

### Root Cause Analysis

These queries help you pinpoint what caused an Issue or what resolved it.

{% tabs %}
{% tab title="🔍 Identifying Causes" %}
**Questions to Ask:**

* *"What metadata changes happened around the time this Issue first appeared?"*
* *"Could any recent changes have caused this problem?"*
* *"Show me all Flow changes that touch the same components as this error"*
* *"Were there any deployments when this Issue started?"*

**What You Get:**

* Timeline of metadata changes correlated with Issue first occurrence
* Analysis of whether changes could cause this specific error type
* Direct links to relevant configuration modifications
* Plain-language explanations of correlation

**Example Response:**

```
Analysis: This Issue first appeared on January 15 at 2:15 PM.

Relevant metadata changes:
✅ Flow "Case_Assignment_Flow_v3" modified on January 15 at 2:10 PM
   - Added decision element checking Contact.Department field
   - No null handling added
   - LIKELY CAUSE: 60% of Contacts have null Department, causing
     NullPointerException when Flow evaluates the condition

⚠️ Also occurred: Apex class "CaseRoutingHelper" updated same day
   - Changes unrelated to error pattern
   - UNLIKELY CAUSE
```

{% endtab %}

{% tab title="✅ Understanding Resolution" %}
**Questions to Ask:**

* *"This Issue stopped occurring on December 17th. Were there any metadata changes that could have fixed it?"*
* *"What changed that might explain why this Issue is no longer happening?"*
* *"Is this Issue actually fixed or might it come back?"*
* *"Show me the changes that resolved this"*

**What You Get:**

* Identification of fixes (intentional changes that resolved the Issue)
* Confidence assessment (is it truly fixed or just dormant?)
* Related Issues that may have been fixed by the same change
* Recommendations for confirming resolution

**Example Response:**

```
Analysis: This Issue last occurred on December 17 at 11:42 AM.
No occurrences since then (32 days).

Relevant metadata changes:
✅ Flow "Case_Assignment_Flow_v3" updated on December 17 at 11:30 AM
   - Added null check before Department field evaluation
   - If Department IS NULL, routes to "General Support" queue
   - DIRECTLY ADDRESSES: The NullPointerException this Issue tracked

Confidence: HIGH - This appears to be an intentional fix.
No related Issues have emerged, and the change directly addresses
the root cause.

Recommendation: Mark Issue as "Completed - Confirmed"
```

{% endtab %}
{% endtabs %}

### Pattern and Impact Analysis

These queries reveal how an Issue behaves over time and how it relates to other Issues.

{% tabs %}
{% tab title="📊 Behavioral Changes" %}
**Questions to Ask:**

* *"Is the problem now manifesting differently?"*
* *"Has this Issue's frequency or impact changed over time?"*
* *"Are there any trends in when this Issue occurs?"*
* *"Did this Issue spike at any point? When and why?"*

**What You Get:**

* Analysis of occurrence patterns (frequency, timing, affected users)
* Identification of spikes or drops in occurrence rate
* Correlation with time-based factors (deployments, business cycles)
* Assessment of whether the Issue is getting better or worse

**Example Response:**

```
Occurrence Pattern Analysis:

Timeline:
- Jan 10-14: 3-5 occurrences per day (baseline)
- Jan 15: SPIKE to 47 occurrences
- Jan 16-20: 15-20 occurrences per day (elevated)
- Jan 21-present: 2-3 occurrences per day (returning to baseline)

Correlation:
✅ Spike on Jan 15 correlates with Flow change (2:10 PM)
✅ Gradual decrease after Jan 20 correlates with data quality fix
   (Department field backfill process completed Jan 20)

Trend: IMPROVING - Occurrence frequency decreasing toward baseline
Recommendation: Monitor for 7 more days to confirm trend
```

{% endtab %}

{% tab title="🔗 Related Issues" %}
**Questions to Ask:**

* *"Are there any related Issues that might have the same root cause?"*
* *"Did fixing this Issue create any new Issues?"*
* *"Has this Issue split into multiple Issues?"*
* *"Are there similar errors affecting other components?"*

**What You Get:**

* Identification of Issues with similar error signatures
* Issues that emerged after this one was fixed (potential regressions)
* Issues that may share a common root cause
* Recommendations for holistic fixes

**Example Response:**

```
Related Issues:

Similar Pattern:
🔗 Issue #1847 - "LeadAssignmentFlow NullPointerException"
   - Similar null pointer error in Lead routing
   - Also checks Industry field without null handling
   - Started Jan 16 (1 day after this Issue)
   - LIKELY RELATED: Same pattern in different Flow

Potential Regression:
🔗 Issue #1892 - "Cases routing to wrong queue"
   - Started Jan 17 (after Dec 17 fix to this Issue)
   - Cases with null Department now going to General Support
   - May be unintended consequence of the fix
   - INVESTIGATE: Is this the desired behavior?

Recommendation: Audit all assignment Flows for null handling
on optional fields to prevent similar Issues.
```

{% endtab %}
{% endtabs %}

### Change Correlation

These queries tie the Issue to specific deployments and component changes.

{% tabs %}
{% tab title="🚀 Deployment Impact" %}
**Questions to Ask:**

* *"Were there any deployment activities when this Issue spiked?"*
* *"Show me all changes deployed around the time this started"*
* *"Could this Issue be related to the deployment on \[date]?"*

**What You Get:**

* Identification of deployments near Issue lifecycle events
* List of components changed in relevant deployments
* Assessment of whether deployment likely caused the Issue
* Recommendations for rollback if appropriate
  {% endtab %}

{% tab title="⚙️ Specific Component Changes" %}
**Questions to Ask:**

* *"Show me the specific changes made to AccountTrigger on December 16th"*
* *"What changed in the Case Assignment Flow?"*
* *"How did the validation rule change?"*

**What You Get:**

* Detailed breakdown of specific metadata changes
* Before/after comparison (when available)
* Plain-language explanation of what changed
* Assessment of how the change relates to the Issue
  {% endtab %}
  {% endtabs %}

***

## Example Investigation Workflow

Here's a typical issue-level investigation demonstrating the power of contextual conversation:

### Scenario: Understanding Why an Issue Stopped

This walkthrough follows a single investigation from the first question to a confirmed resolution.

{% tabs %}
{% tab title="1️⃣ Initial Question" %}
**You open Issue #1756 and ask:**

*"This Issue stopped occurring on December 17th. Were there any metadata changes that could have fixed this problem?"*

**Org Genie Response:**

```
Analysis of metadata changes around December 17:

✅ Flow "Case_Assignment_Flow_v3" updated on December 17 at 11:30 AM
   (12 minutes before last occurrence)
   
   Changes:
   - Added null check before Contact.Department evaluation
   - If Department IS NULL, route to "General Support" queue
   
   ASSESSMENT: This change directly addresses the NullPointerException
   that this Issue was tracking. The timing (12 minutes before last
   occurrence) suggests this was an intentional fix.
```

**What happened:**\
Org Genie understood "this Issue" refers to Issue #1756, knew its last occurrence date, and searched for relevant metadata changes in that timeframe.
{% endtab %}

{% tab title="2️⃣ Dig Deeper" %}
**You ask a follow-up:**

*"Can you show me the specific changes made to that Flow?"*

**Org Genie Response:**

```
Detailed Changes to Case_Assignment_Flow_v3 (Dec 17, 11:30 AM):

ADDED: Decision Element "Check_Department_Exists"
  Condition: Contact.Department IS NOT NULL
  If TRUE: Continue to existing department-based routing
  If FALSE: NEW path → Route to "General Support" queue

MODIFIED: Existing "Route_By_Department" element
  Now only executes when Department exists (after null check)

Technical Explanation:
Previously, the Flow attempted to evaluate Contact.Department in
a formula without checking if the field was populated. When
Department was null (60% of cases), this caused a 
NullPointerException.

The fix adds a null check first, routing cases with null
Department to a default queue instead of attempting to evaluate
the field.
```

**What happened:**\
Org Genie knew "that Flow" refers to the Flow mentioned in the previous response, and provided detailed technical breakdown without you needing to specify the Flow name again.
{% endtab %}

{% tab title="3️⃣ Verify Understanding" %}
**You ask for confirmation:**

*"Based on those changes, why would that fix the NullPointerException this Issue was tracking?"*

**Org Genie Response:**

```
Technical Explanation:

Root Cause of NullPointerException:
The Flow was evaluating a formula that referenced 
Contact.Department. In Salesforce, attempting to access a null
field in a formula context throws a NullPointerException.

How the Fix Addresses It:
By adding a Decision Element that checks "IS NOT NULL" BEFORE
evaluating the Department field, the Flow now:
1. First determines if Department exists
2. Only proceeds to evaluate Department in formulas if it exists
3. Takes alternative path (General Support queue) if null

This prevents the Flow from ever attempting to evaluate a null
field, eliminating the source of the NullPointerException.

Confidence: HIGH - This is a textbook fix for null reference errors
in Salesforce Flows.
```

**What happened:**\
Org Genie provided technical explanation in context of the specific Issue's error type, explaining how the fix addresses the root cause.
{% endtab %}

{% tab title="4️⃣ Check for Related Issues" %}
**You expand the investigation:**

*"Are there any other Issues that might have been fixed by this same change?"*

**Org Genie Response:**

```
Related Issues potentially resolved by Flow fix:

🔗 Issue #1698 - "Case routing failure - Contacts without Department"
   Status: New
   Last Occurrence: December 17, 11:25 AM (5 min before Flow fix)
   Pattern: Cases not being routed when Contact lacks Department
   
   ASSESSMENT: LIKELY RESOLVED by the same fix. The alternative
   path now handles these cases explicitly.
   
   RECOMMENDATION: Mark as "Completed - Related to #1756"

No other Issues appear directly related to this specific fix.
However, similar null handling patterns may exist in other Flows
(see recommendation below).

Proactive Recommendation:
Audit other assignment Flows (LeadAssignmentFlow, 
OpportunityRoutingFlow) for similar null field references to
prevent analogous Issues.
```

**What happened:**\
Org Genie identified a related Issue that was likely fixed by the same change, plus provided proactive recommendations for preventing similar Issues elsewhere.
{% endtab %}

{% tab title="5️⃣ Follow-Up Later" %}
**You return to the Issue a week later and continue the conversation:**

*"In our previous conversation, we identified a Flow change that fixed this. Has there been any additional activity since then?"*

**Org Genie Response:**

```
Update since our last conversation (December 17):

Issue Status:
✅ No new occurrences in 7 days (previously averaging 5/day)
✅ Related Issue #1698 also resolved (no occurrences)
✅ No regression or related Issues emerged

Metadata Activity:
⚠️ Flow "Case_Assignment_Flow_v3" modified again on December 20
   Changes: Adjusted queue assignments (unrelated to null handling)
   Impact: None on this Issue

Confidence Assessment: CONFIRMED RESOLVED
This Issue is genuinely fixed, not dormant. Sufficient time has
passed without recurrence to confirm the December 17 fix was
effective.

Recommendation: Mark Issue as "Completed - Confirmed" and close.
```

**What happened:**\
Org Genie maintained context from your previous conversation (even days later), remembered which Flow and changes you discussed, and provided an update on activity since then.
{% endtab %}
{% endtabs %}

{% hint style="success" %}
**Contextual Memory**

Notice how each question builds on previous responses without needing to re-explain:

* "that Flow" → Org Genie knows which Flow
* "those changes" → Org Genie remembers what changes
* "our previous conversation" → Org Genie recalls context from days ago

This contextual conversation dramatically accelerates investigation compared to starting fresh with each question.
{% endhint %}

***

## When to Use Issue-Level Org Genie

Understanding when issue-level context is most valuable helps you choose the right approach for each investigation.

### Perfect Use Cases

Issue-level context is most valuable in the following scenarios.

<table><thead><tr><th width="220">Scenario</th><th>Why Issue-Level Works Best</th></tr></thead><tbody><tr><td><strong>🔍 Root Cause Investigation</strong></td><td>Automatic correlation with the Issue's timeline and affected components—no need to specify dates or filters</td></tr><tr><td><strong>✅ Understanding Resolution</strong></td><td>Analyzes changes around when the Issue stopped, confirming if it's truly fixed or just dormant</td></tr><tr><td><strong>📊 Behavioral Analysis</strong></td><td>Tracks occurrence patterns, spikes, and trends for this specific Issue over time</td></tr><tr><td><strong>🔗 Finding Related Issues</strong></td><td>Identifies Issues with similar error signatures or shared root causes</td></tr><tr><td><strong>🚀 Deployment Correlation</strong></td><td>Determines if deployments caused or resolved the Issue by analyzing timing</td></tr><tr><td><strong>⚙️ Component Deep-Dive</strong></td><td>Examines specific metadata changes relevant to the Issue's error type and affected components</td></tr></tbody></table>

### When to Use Org-Level Instead

For broader questions that span many Issues, org-level context is the better fit.

<table><thead><tr><th width="220">Scenario</th><th>Why Org-Level Works Better</th></tr></thead><tbody><tr><td><strong>🌐 Comparing Multiple Issues</strong></td><td>Need to analyze across many Issues, not focus on one</td></tr><tr><td><strong>📈 Org-Wide Trend Analysis</strong></td><td>Understanding overall error patterns, not a specific Issue</td></tr><tr><td><strong>🎯 Prioritization</strong></td><td>Determining which Issues to investigate requires broad view</td></tr><tr><td><strong>📊 Health Checks</strong></td><td>Monitoring org health and identifying anomalies spans all Issues</td></tr></tbody></table>

***

## Best Practices for Issue-Level Investigation

### Effective Questioning Techniques

A few questioning habits get you to root cause faster.

{% tabs %}
{% tab title="⏱️ Reference Timing" %}
**Good Questions:**

* *"Around December 17th"* (specific date)
* *"When this Issue first appeared"* (relative to Issue)
* *"Between the first and last occurrence"* (Issue-relative range)

**Why it works:**\
Org Genie understands relative timing in context of the Issue's lifecycle without requiring you to look up exact timestamps.
{% endtab %}

{% tab title="🎯 Target Specific Metadata" %}
**Good Questions:**

* *"Show me changes to the AccountTrigger"*
* *"What changed in Flow 'Case Assignment'?"*
* *"Were there validation rule modifications?"*

**Why it works:**\
If you suspect a particular component, asking about it directly focuses the analysis and accelerates investigation.
{% endtab %}

{% tab title="🔗 Correlate with Impact" %}
**Good Questions:**

* *"Could this Flow change explain the spike in occurrences last week?"*
* *"Did the deployment on January 15 cause this Issue to start?"*
* *"What changed that might have made this Issue worse?"*

**Why it works:**\
Connecting observable effects (spikes, resolution) with potential causes helps Org Genie provide targeted correlation analysis.
{% endtab %}

{% tab title="✅ Verify Resolutions" %}
**Good Questions:**

* *"Is this Issue actually fixed or might it come back?"*
* *"Has sufficient time passed to confirm resolution?"*
* *"Are there any conditions under which this Issue could recur?"*

**Why it works:**\
Distinguishes between genuine fixes and dormant Issues that may resurface under different conditions.
{% endtab %}

{% tab title="📚 Build on Context" %}
**Good Questions:**

* *"In our previous conversation, we identified a Flow change. Has there been any activity since?"*
* *"You mentioned related Issue #1847—is that still occurring?"*
* *"Can you show me more details about the change you mentioned earlier?"*

**Why it works:**\
Leverages persistent chat history to continue investigations across multiple sessions without losing context.
{% endtab %}
{% endtabs %}

### Investigation Workflow

A typical investigation moves from broad questions to specific ones.

**Recommended Approach:**

1. **Start with High-Level Question**\
   *"What could have caused this Issue?"* or *"Why did this Issue stop?"*
2. **Review AI Response**\
   Note correlated changes, timing, and confidence assessment
3. **Dig into Specifics**\
   *"Show me details on the Flow change you mentioned"*
4. **Verify Understanding**\
   *"How does that change address this specific error?"*
5. **Check for Related Issues**\
   *"Are there similar Issues that might have the same cause?"*
6. **Return Later if Needed**\
   Continue the conversation days later without losing context

{% hint style="success" %}
**Iterative Investigation**

Don't try to ask the perfect question on the first try. Start broad, then refine based on responses. Org Genie's contextual memory makes iterative investigation natural and efficient.
{% endhint %}

***

## Navigating Between Contexts

You can easily move between org-level and issue-level Org Genie as your investigation evolves.

### From Org-Level to Issue-Level

**Scenario:** You're doing a weekly health check and discover a specific Issue that warrants investigation.

**Workflow:**

1. Ask org-level Org Genie: *"Which Issues from the last week warrant immediate attention?"*
2. Org Genie returns: **Issue #1756** - Case Assignment Flow failures (47 occurrences, 12 affected users)
3. Click the Issue link to view Issue #1756 details
4. Open issue-level Org Genie from the Issue detail page
5. Continue with focused investigation: *"What changed that could have caused this?"*

**Why this works:**\
Discovery happens at org-level (broad view), investigation happens at issue-level (focused context).

### From Issue-Level to Org-Level

**Scenario:** Your investigation of a specific Issue reveals it's part of a broader pattern.

**Workflow:**

1. Investigating Issue #1756 with issue-level Org Genie
2. Org Genie identifies: *"Similar pattern in Issue #1847 (LeadAssignmentFlow)"*
3. You realize this might be org-wide problem with assignment Flows
4. Navigate back to org-level Org Genie
5. Ask: *"Show me all Issues related to assignment Flows with null pointer errors"*
6. Get org-wide view of the pattern

**Why this works:**\
When you discover the problem is bigger than one Issue, switching to org-level provides the necessary broad context.

### Context Isolation

Switching between contexts never costs you your place in either investigation.

{% hint style="info" %}
**Separate Chat Histories**

Each context maintains its own chat history:

* **Org-level:** Per-user history (your personal workspace)
* **Issue-level:** Per-Issue history (stored with the Issue)

You won't lose your place in either investigation when switching contexts. Each conversation persists independently.
{% endhint %}

***

{% hint style="success" %}
**Start Your First Issue Investigation**

1. Navigate to any Issue with >10 occurrences
2. Open Org Genie from the Issue detail page
3. Ask: *"What could have caused this Issue?"*
4. Review the response and ask a follow-up: *"Show me details on \[specific change mentioned]"*
5. Continue the conversation until you understand the root cause
6. Return to the Issue days later and continue: *"Has there been any activity since we last talked?"*

Experience how contextual conversation accelerates investigation compared to starting fresh with each question.
{% endhint %}
