Q-Director Configuration Guide

QuantumOps Configuration Guide

A comprehensive guide to configuring your QuantumOps tenant and Q-Director AI automation settings.

Table of Contents

1. Overview
2. Tenant Configuration (Update Tenant)
   • Tenant Information
   • Organization Management
   • AI Provider Configuration
   • Vector Store Configuration
   • HaloPSA Integration
   • Ticket Type Exclusions
   • Webhook Configuration
3. Q-Director Settings
   • HaloPSA Dashboard URLs
   • Incremental Lifecycle Analysis
   • Stale Ticket Detection
   • Sentiment & Tonality Analysis
   • Ticket Categorization
   • Triage Assist
   • Research & Sherlock Module
   • AI Analysis Configuration
   • Triage Dashboard
4. Best Practices
5. Troubleshooting

Overview

QuantumOps provides two primary configuration interfaces:

Why Two Configuration Pages?

The Platform Settings page handles foundational, infrastructure-level settings that typically need to be configured once during initial setup. These settings define how QuantumOps connects to external services and how the UI behaves.

The Q-Director Settings page controls behavioral and operational settings that service desk managers may tune regularly to optimize AI-powered automation. These settings define what QuantumOps does with tickets once connected.

Platform Settings (Update Tenant)

Access via: Settings Gear in the top right corner of any page or navigate to /UpdateTenant

Tenant Information

Basic organizational and billing information associated with your QuantumOps tenant.

Fields

Display Time Zone

Why configure this?

All timestamps in QuantumOps are stored as UTC for accuracy and consistency. The Display Time Zone setting controls how these timestamps appear throughout the application:

• Ticket dashboards and grids
• Chat analytics and logs
• Timeclock reports and entries
• Documentation sync status
• Stale ticket calculations

Recommendation: Set this to match your service desk's primary operating timezone. You can define agent specific timezones in their technician configuration profile later.

Organization Management

Multi-user capabilities for team-based access to QuantumOps.

Enable Multi-User

When enabled, you can invite additional team members to access your QuantumOps tenant. This feature integrates with Auth0 for secure identity management.

Prerequisites:

1. Enable the Multi-User toggle
2. Configure an Organization Name (required)
3. Save changes
4. Use the User Management section to invite team members

Organization Name

This name identifies your organization with our identity provider (Auth0). It appears:

• When users are invited to join your organization
• During the login process
• In Auth0 management portals

Requirements:

• Must be unique across all QuantumOps tenants
• Will be converted to lowercase with spaces removed for the internal identifier
• The display name preserves your original formatting

Example:

• You enter: Acme Technology Solutions
• Internal ID: acmetechnologysolutions
• Display Name: Acme Technology Solutions

AI Provider Configuration

QuantumOps supports multiple AI providers for flexibility and redundancy.

OpenAI Integration

Enable this if you want to use OpenAI models.

Anthropic (Claude) Integration

Enable this for access to Claude models. Recommended for most QuantumOps features.

##### Rate Limiting Configuration

Why configure rate limits?

Anthropic enforces API rate limits based on your service tier. Configuring these limits correctly:

• Prevents request failures during high-volume periods
• Optimizes ticket processing speed
• Enables smart queuing to handle rate limit events gracefully

How to find your tier:

1. Log into console.anthropic.com
2. Navigate to Settings → Limits
3. Note your current tier and limits

##### Service Tiers

Selecting a tier automatically configures:

• Token Capacity/Min
• Request Capacity/Min
• Recommended Max Wait Time

##### Advanced Settings

Primary AI Model

Select the default model for ticket analysis and AI operations.

Available options depend on enabled providers. Models are grouped by provider with cost tier indicators:

• ⭐ Recommended: Best balance of capability and cost
• Low Cost: Suitable for high-volume, simpler analyses
• High Performance: Best quality for complex reasoning

Vector Store Configuration

Vector stores power semantic search, similar ticket detection, and knowledge base matching.

Weaviate (Recommended)

Why Weaviate?

• Fully managed—no additional infrastructure required
• Automatic vectorization
• Data isolation through tenant naming
• Optimized for QuantumOps workloads

Auto-Vectorization: When QuantumOps save tickets or memories, Weaviate automatically creates vector embeddings.

Azure AI Search (Legacy)

Use your own Azure AI Search instance for organizations with existing Azure infrastructure requirements.

Note: You must manually create indexes using the "Create Search Index" button when using Azure AI Search.

HaloPSA Integration

Connect QuantumOps to your HaloPSA instance for ticket retrieval and processing.

Required Fields

Setting Up HaloPSA API Access

1. In HaloPSA, navigate to Configuration → Integrations → HaloPSA API
2. Create a new API Application:
   • Application Name: QuantumOps Integration
   • Authentication Method: Client Credentials
   • Scopes: Enable all read scopes and write scopes for tickets/actions
3. Copy the generated Client ID and Client Secret
4. Paste into QuantumOps Update Tenant page
5. Click Test Connection to verify

Test Connection

Always test your credentials before saving. The test verifies:

• URL accessibility
• Credential validity
• Token generation success
• Basic API communication

Ticket Type Exclusions

Configure which HaloPSA ticket types should be excluded from AI analysis.

Why exclude ticket types?

Some ticket types may not benefit from AI analysis:

• Internal administrative tickets
• Automated monitoring alerts
• Time tracking entries
• Project tasks

How to configure:

1. Click Configure
2. Select ticket types to exclude from the list
3. Selected types will appear in the "Excluded Types" field
4. Save your tenant configuration

Webhook Configuration

Real-time ticket processing requires webhook integration via a custom automation runbook in HaloPSA.

Webhook URI

The URI that HaloPSA uses to receive QuantumOps actions.

Format: https://yourinstance.halopsa.com/api/automation/<unique_id>

Recreate Webhooks Button:

Use this if:

• Webhooks stopped working
• You changed your QuantumOps instance URL
• You need to reset webhook configuration in HaloPSA

This automatically configures the required webhooks in HaloPSA:

• Ticket Created
• Ticket Updated
• Action Added

Q-Director Settings

Access via: Settings → Q-Director Settings or navigate to /qdirector-settings

Q-Director is QuantumOps' AI-powered ticket automation engine. It provides intelligent ticket lifecycle management through configurable rules and AI analysis.

Guided Tour

Click the Tour button in the header to start an interactive walkthrough of all Q-Director settings. The tour highlights each section with explanations of functionality and best practices.

HaloPSA Dashboard URLs

Embed QuantumOps dashboards directly within HaloPSA using custom iframe tabs.

Available Dashboards

Setup Instructions

1. In HaloPSA, navigate to Configuration → Custom Objects → Custom Tabs
2. Select the target entity (e.g., Ticket)
3. Click New
4. Set Type to iFrame
5. Paste the URL from Q-Director Settings
6. Configure tab order (lower numbers appear first)
7. Save

Copy URLs: Use the green copy button next to each URL for easy clipboard copying.

Coming Soon Dashboards: Some dashboards are marked "Coming Soon" and cannot be copied yet. These will be enabled in future releases.

Incremental Lifecycle Analysis

Automatically analyze tickets in real-time when they're updated via webhooks.

Enable Incremental Refreshes

When enabled, every webhook event triggers an AI analysis of the affected ticket. This provides:

• Real-time ticket insights
• Immediate staleness detection
• Up-to-date sentiment scores
• Current categorization

Configuration Options

Why configure the AI model?

• Haiku models are fast and cost-effective for quick incremental updates
• Sonnet models provide better analysis but cost more
• Opus models give highest quality but highest cost

For incremental analysis that runs on every ticket update, Haiku is usually the best balance.

Stale Ticket Detection

Identify tickets that require attention based on configurable staleness rules.

Ticket Type Filter

Select which ticket types are monitored for staleness.

If no types are selected, all ticket types will be analyzed. This is useful for focusing staleness tracking on support tickets while ignoring project tasks or internal tickets.

Inactivity Thresholds

Inactivity Trigger Options:

Track Agent Follow-Up Requirements

When enabled, tickets become stale if:

1. The agent was the last to respond
2. The customer hasn't replied
3. The agent hasn't followed up within the configured days

Use case: Ensures agents don't forget to follow up on tickets waiting for customer response.

Age & Assignment Rules

Max Ticket Age catches tickets that have ongoing activity but never get resolved. Even with regular updates, a 30-day-old open ticket may need escalation.

Assignment Delay ensures new tickets get picked up quickly by agents.

SLA Breach Tracking

Separately track SLA breaches as a more severe status than regular staleness.

Why track separately? SLA breaches have contractual implications and should appear distinctly in dashboards. Regular staleness is an operational metric; SLA breach is a compliance issue.

AI Stale Analysis

Use AI to understand why tickets are stale and provide actionable recommendations.

Analysis Trigger Options:

Per-Ticket-Type Overrides

Configure custom staleness rules for specific ticket types. For example:

• Emergency tickets: 4-hour inactivity threshold
• Project requests: 7-day threshold
• Enhancement requests: 14-day threshold

Click Configure Per-Ticket-Type Overrides to access the override dialog.

Sentiment & Tonality Analysis

AI-powered analysis of customer sentiment and agent communication quality.

Sensitivity Level

Controls how aggressively the AI flags potential issues.

Start with Medium and adjust based on alert volume and accuracy.

Alert Thresholds

Alerts fire when scores are at or below these values (scale: 1-10).

Lower values = more permissive, fewer alerts
Higher values = more strict, more alerts

Custom Analysis Instructions

Add organization-specific guidance for the AI to consider during analysis:

Example Instructions:

- Our VIP clients (Acme Corp, Beta Industries) should have lower alert thresholds
- Password reset tickets should never trigger sentiment alerts
- Pay special attention to mentions of "contract renewal" or "switching providers"
- Tickets from the Sales team are often urgent but not frustrated

These instructions are included in every analysis while preserving the standard output format.

Exclusions

Exclude specific ticket types or clients from sentiment analysis:

• Excluded Ticket Types: Skip analysis for monitoring alerts, internal tickets, etc.
• Excluded Clients: Skip analysis for internal company accounts, test accounts, etc.

Feedback Buttons

When enabled, sentiment alerts sent to Slack or Teams include thumbs up/down buttons. User feedback helps improve future analysis accuracy.

Ticket Categorization

Automatically categorize tickets using AI analysis.

Enable Categorization

When enabled, Q-Director analyzes ticket content and assigns appropriate categories.

When to Categorize

Allow Q-Director to Create Categories

When enabled, if AI encounters a ticket that doesn't fit existing categories, it can create new ones in HaloPSA.

Category Creation Mode:

• Category Only: Create regular categories
• Resolution Category Only: Create resolution categories
• Both: Create both types as needed

Recommendation: Start with this disabled to use only existing categories. Enable once your category structure is established.

Additional Options

Why recategorize on closure? Initial categorization is based on limited information. Final categorization based on the full resolution context is more accurate for reporting.

Triage Assist

AI-powered ticket routing and assignment recommendations.

Enable Triage Assist

When enabled, AI analyzes incoming tickets and provides:

• Team/agent assignment recommendations
• Priority suggestions
• Impact and urgency assessments
• Related ticket identification

Ticket Types for Triage

Select which ticket types receive triage assistance. Leave empty for all types.

Auto Triage and Assign

⚠️ Warning: This enables fully automatic ticket assignment without dispatcher intervention.

When enabled:

• AI routes tickets to teams/agents automatically
• Assignments are based on AI analysis, skills, and availability
• Dispatchers can still override assignments

##### Auto Triage Configuration

##### On-Call Notifications

Notify staff when tickets arrive outside business hours.

Research & Sherlock Module

Sherlock performs deep research on tickets to find solutions and related information.

What Sherlock Does

1. Similar Incident Lookup: Finds previously resolved tickets with similar issues
2. Semantic Knowledge Base Matching: Searches documentation and KB articles
3. Web Searches: Finds external resources and vendor documentation
4. AI Analysis: Synthesizes findings into actionable recommendations

Post Sherlock Results to Ticket

When to enable: Useful for agents who want research findings directly in HaloPSA without switching applications.

When to disable: Keep HaloPSA tickets cleaner; access findings through QuantumOps dashboard instead.

AI Analysis Configuration

Fine-tune which ticket data is included in AI analysis.

Custom Action Filtering

Enable to exclude specific action types from AI processing.

Why filter actions?

Some HaloPSA actions add noise without meaningful content:

• SLA Hold/Release events
• Rule Applied notifications
• Automated status changes
• Internal system actions

Excluding these:

• Improves analysis quality by focusing on meaningful content
• Reduces token usage and costs
• Speeds up processing with less data to analyze

Action Outcomes to Exclude

Select action types that should NOT be sent to AI:

• Rule Applied
• SLA Hold
• SLA Release
• Status Changed (automated)
• Internal Note (optional)

⚠️ Warning: Never exclude:

• Email User / Email Update / First User Email
• Close / Escalated
• Any user-facing communications

Triage Dashboard

A specialized dashboard for dispatchers embedded within HaloPSA tickets.

Enable Triage Dashboard

When enabled, the triage dashboard iframe shows AI-powered insights directly within HaloPSA tickets.

Ticket Types for Dashboard

Select which ticket types display the triage dashboard. Leave empty for all types.

Duplicate Detection

Automatically detect duplicate tickets based on content similarity.

Recurring Incident Detection

Identify when similar issues are reported repeatedly.

Global Threshold helps identify widespread issues that may indicate major incidents affecting multiple clients.

Machine Learning Feedback

Improve AI accuracy using dispatcher corrections.

Recommendation: Enable tracking with 50 corrections minimum before retraining for reliable improvement.

Dashboard Display & Performance

Best Practices

Initial Setup Order

1. Configure Update Tenant first:

   • Set HaloPSA credentials and test connection
   • Configure AI providers (start with Anthropic)
   • Set up vector store (use Weaviate default)
   • Configure timezone and organization

2. Then configure Q-Director Settings:

   • Start with Incremental Lifecycle Analysis
   • Add Stale Ticket Detection rules
   • Enable Triage Dashboard
   • Gradually enable automation features

Recommended Starting Configuration

For Small Teams (< 5 technicians)

Incremental Lifecycle Analysis: Enabled
├── AI Model: claude-haiku-4-5-20251001
├── Stale Recheck: 24 hours
└── Scan Interval: 60 minutes

Stale Ticket Detection: Enabled
├── Inactivity Threshold: 5 business days
├── Inactivity Trigger: No Agent Action
└── AI Stale Analysis: Disabled (save tokens)

Triage Assist: Enabled
└── Auto Triage: Disabled (manual recommendations only)

Sentiment Analysis: Enabled
├── Sensitivity: Medium
└── Alert Thresholds: 4/4

For Medium Teams (5-15 technicians)

Incremental Lifecycle Analysis: Enabled
├── AI Model: claude-haiku-4-5-20251001
├── Stale Recheck: 12 hours
└── Scan Interval: 30 minutes

Stale Ticket Detection: Enabled
├── Inactivity Threshold: 3 business days
├── Track Agent Follow-Up: Enabled (2 days)
├── SLA Breach Tracking: Enabled
└── AI Stale Analysis: Incremental Refresh

Triage Assist: Enabled
├── Auto Triage: Optional (business hours only)
└── Ticket Categorization: On Closure

Triage Dashboard: Enabled
├── Duplicate Detection: Enabled (90%)
├── Recurring Detection: Enabled
└── ML Feedback: Enabled

For Large Teams (15+ technicians)

Enable all features with:

• Auto triage with schedules
• Full escalation paths
• Aggressive stale detection (2 business days)
• Pre-generated AI predictions
• Lower refresh intervals

Cost Optimization

1. Use Haiku for incremental analysis - Fast and cheap for frequent updates
2. Use Sonnet for triage decisions - Better quality for important routing
3. Exclude noisy actions from AI - Reduce token usage
4. Set appropriate analysis intervals - Avoid over-processing
5. Use stale analysis triggers wisely - "Scheduled Only" uses fewer tokens

Troubleshooting

HaloPSA Connection Issues

Symptom: "Connection failed" or timeout errors

Solutions:

1. Verify the HaloPSA URL doesn't include https:// prefix
2. Confirm Client ID and Secret are correct (copy fresh from HaloPSA)
3. Check that API scopes include read/write ticket permissions
4. Verify your HaloPSA region matches the dropdown selection

Webhooks Not Triggering

Symptom: Tickets aren't being analyzed in real-time

Solutions:

1. Click "Recreate Webhooks" in Update Tenant
2. Verify webhook URL is accessible from HaloPSA
3. Check HaloPSA webhook logs for delivery failures
4. Ensure Incremental Lifecycle Analysis is enabled

High Token Usage

Symptom: Anthropic bills higher than expected

Solutions:

1. Switch incremental analysis to Haiku model
2. Increase "Min Hours Between Analysis" for stale tickets
3. Enable custom action filtering to exclude noisy events
4. Set "Scheduled Only" for AI stale analysis
5. Reduce sentiment analysis sensitivity

Inaccurate Categorization

Symptom: Tickets getting wrong categories

Solutions:

1. Disable "Allow Q-Director to Create Categories"
2. Review and clean up existing HaloPSA categories
3. Enable "Recategorize on Closure" for learning
4. Add custom instructions for edge cases

Dashboard Not Loading in HaloPSA

Symptom: Iframe shows blank or error

Solutions:

1. Verify the dashboard URL was copied correctly
2. Check that Custom Tab type is set to "iFrame"
3. Confirm your QuantumOps instance allows iframe embedding and that you are using the HaloPSA URL you configured to access your Halo portal. (e.g. if you set up QuantumOps with "yourmsp.halopsa.com" but have a branded domain of "psa.yourmsp.com", the latter will not load dashboards due to CORS restrictions)
4. Clear browser cache and try again

Additional Resources

• Setup Wizard - First-time configuration walkthrough
• Initial Training / Data Import - Importing historical tickets
• What is QuantumOps? - Platform overview
• Review Prerequisites - Review Prerequisites