Setup Wizard

Prev Next

QuantumOps Setup Wizard

The QuantumOps setup wizard guides you through configuring your AI-powered MSP operations platform in 11 steps. Your progress is automatically saved, so you can leave and return at any time.

Estimated completion time: 15-20 minutes

[insert screenshot of Setup Wizard welcome screen here]


Before You Begin

Ensure you have the following ready before starting. See Review Prerequisites for detailed preparation instructions.

Required

Item Example Where to Get It
HaloPSA URL yourmsp.halopsa.com Your HaloPSA login page
HaloPSA Client ID 12345678-abcd-... Configuration → Integrations → API Applications
HaloPSA Client Secret secret_xxxxx... Same location as Client ID
Billing Information Company name, address Your business records

Optional

Item Purpose
Anthropic API Key Bring-your-own-key for Claude models
OpenAI API Key Bring-your-own-key for GPT models
PagerDuty/SMS/Email config On-call notifications

Wizard Overview

The wizard consists of 11 steps organized into four phases:

Phase 1: Foundation

  1. Welcome — Overview and prerequisites checklist
  2. HaloPSA Connection — API credentials and webhook setup
  3. Billing & Organization — Company info, spending limits, multi-user access

Phase 2: AI Configuration

  1. AI Configuration — Model selection, provider settings, rate limiting

Phase 3: Q-Director Setup

  1. Q-Director Core — Incremental analysis, categorization, stale detection
  2. Q-Director Automation — Triage assist, auto-assignment, escalation
  3. Q-Director Intelligence — Sentiment, duplicates, recurring patterns

Phase 4: Features & Launch

  1. Web Search — Configure AI research capabilities
  2. Customer Success — CSAT surveys and health monitoring
  3. Dashboard Setup — HaloPSA iframe integration URLs
  4. Import & Finalize — Data import options and launch

Step 1: Welcome

The welcome screen provides an overview of what you're about to configure and an interactive prerequisites checklist.

[insert screenshot of Welcome tab here]

What You'll See

  • Setup time estimate: 15-20 minutes
  • Auto-save confirmation: Your progress is saved automatically
  • Prerequisites checklist: Click each item to mark it complete
  • Quick tour option: Interactive walkthrough of the wizard

Prerequisites Checklist

The wizard displays six prerequisite items you can check off:

  • HaloPSA Agent URL — Your HaloPSA instance URL
  • HaloPSA Client ID — From your API application
  • HaloPSA Client Secret — From your API application
  • Webhook URL — Provided during setup for HaloPSA configuration
  • Billing Information — Company details for invoicing
  • AI API Keys — Optional, or use managed services

Click Let's Get Started to proceed.


Step 2: HaloPSA Connection

This step establishes the secure connection between QuantumOps and your HaloPSA instance.

[insert screenshot of HaloPSA Connection tab here]

API Credentials

Field Format Notes
HaloPSA Agent URL yourmsp.halopsa.com Do NOT include https://
Client ID GUID format From HaloPSA API application
Client Secret Password field Masked for security

Testing Your Connection

  1. Enter all three credential fields
  2. Click Test Connection
  3. Wait 10-15 seconds for validation
  4. Look for the green success indicator
    Connection Success Response:
✅ Connection Successful!
Successfully authenticated with HaloPSA.

Regional Settings

Field Options Impact
Region US, CA, EU, AP Affects API endpoint routing
Version Production, RC, Beta, Bleeding Edge HaloPSA version compatibility

💡 Recommendation: Select Production unless you're specifically testing HaloPSA beta features.

Webhook Configuration

QuantumOps provides a custom automation runbook you'll configure in HaloPSA after completing the wizard.

Copy the Webhook URI — You'll need this for QuantumOps configuration:

https://[your-instance].halopsa.com/api/automation/[unique id]

Ticket Type Exclusions

Optionally exclude specific ticket types from AI processing. Common exclusions:

  • Alerts/Informational — Automated maintenance tasks and alerts
  • Internal — Internal team tickets
  • Quotes — Quote and Opportunity tickets

ℹ️ Note: This selector only becomes available after a successful connection test.

Security Information

Your credentials are protected by:

  • AES-256 encryption at rest
  • Azure Key Vault for secret storage
  • TLS 1.3 for all connections

Step 3: Billing & Organization

Configure your billing information, spending controls, and multi-user access settings.

[insert screenshot of Billing & Organization tab here]

Billing Information

Field Required Notes
Billing Name ✅ Yes Company or individual name for invoices
Country ✅ Yes Select from dropdown
Address Lines Optional Street address, suite, etc.
City, State, ZIP Optional Complete address for invoicing

Spending Controls

Set a monthly AI consumption limit to control costs (if desired):

Preset Use Case
$100 Small MSP, < 20 tickets/day
$250 Medium MSP, 21-50 tickets/day
$500 Larger MSP, 51-200 tickets/day
$1000 Enterprise, 200+ tickets/day
Unlimited No spending cap

💡 Tip: Most MSPs processing 50-200 tickets/day consume $250-500/month in AI costs depending on feature set enabled.

Display Preferences

Time Zone — All timestamps in QuantumOps will display in your selected time zone. This affects:

  • Ticket activity times
  • Alert timestamps
  • Report generation
  • Dashboard displays

Multi-User Access

Toggle Enable Multi-User Access to allow team collaboration:

When Enabled:

  • Organization Name field appears (required) — This is the public-facing name shown on login screens and within the application. It can differ from your billing name (e.g., "Acme IT Solutions" vs. "Acme IT Solutions LLC").
  • Invite Team Members button becomes available
  • Team members can access the entire QuantumOps management console
  • Role-based access controls become available in Q1 2026

Inviting Team Members

With multi-user enabled, click Invite to send email invitations:

  1. Enter colleague's email address
  2. Click Send Invitation
  3. Invitee receives email with setup instructions
  4. They'll appear in the Team Members list once registered

Step 4: AI Configuration

Configure your AI providers, model preferences, and rate limiting settings.

[insert screenshot of AI Configuration tab here]

Primary AI Model

The primary model handles the bulk of platform usage—ticket analysis, summarization, and reasoning tasks.

Model Provider Cost Best For
Claude 4.5 Sonnet Anthropic Medium Best balance of cost/quality/speed
Claude 4.5 Opus Anthropic Premium Highest accuracy
GPT-5/GPT-5.2 OpenAI Premium Advanced reasoning, slow
Azure Foundry All Supported Models Varies Variable performance

Recommended: Claude 4.5 Sonnet offers the best cost-performance ratio for MSP ticket analysis, plus the highest performance and a 1M token context window that some functions in the application depend on (especially for large MSPs). Not using Claude 4.5 Sonnet or a 1M token context window model as your primary AI model will prevent large batch operations from being as efficient. This primarily applies to large semantic memory or profile generation tasks for large MSPs that sometimes exceed the standard 200k token context window due to the number of semantic memories and interactions that need to be consolidated or analyzed. Since queries in excess of 200k tokens cost double per token after 200k, QuantumOps only uses 1M token context for those explicitly large tasks, and only after determining a token context window will be exceeded or has been exceeded while attempting a cheaper approach.


Secondary AI Model (Fast)

The secondary model handles lightweight, incremental tasks that don't require complex reasoning—such as updating analysis or trajectories when a user or agent emails on a ticket, and performing sentiment/tonality analysis.

Model Provider Cost Best For
Claude 4.5 Haiku Anthropic Low Fast, efficient, high quality incremental updates
Claude 4.5 Sonnet Anthropic Medium Higher quality when needed
GPT-5/GPT-5.2 Fast OpenAI Low Budget alternative
Azure Foundry All Supported Models Varies Variable performance

Recommended: Claude 4.5 Haiku provides the speed, efficiency, and accuracy needed for high-frequency incremental refreshes at minimal cost.


Orchestration AI Model

The orchestration model handles complex chat queries and guides responses from Qubit. This model sees minimal usage but requires advanced reasoning and tool-use capabilities when invoked.

Model Provider Cost Best For
Claude 4.5 Opus Anthropic Premium Complex orchestration & reasoning
Claude 4.5 Sonnet Anthropic Medium Cost-conscious alternative
GPT-5/GPT-5.2 OpenAI Premium Advanced multi-step reasoning, slow
Azure Foundry All Supported Models Varies Variable performance

Recommended: Claude 4.5 Opus provides the highest reasoning capability for orchestrating complex tasks and guiding nuanced responses at a competitive cost and speed.

Anthropic (Claude) Configuration

Toggle Enable Anthropic to configure Claude models.

API Key Options:

  • Leave empty → Use QuantumOps managed service (recommended)
  • Enter your key → Bring Your Own Key for direct billing

Rate Limiting Configuration:

Setting Purpose Recommendation
Service Tier Matches your Anthropic account tier Check at console.anthropic.com
Timeout (min) Max wait for AI response 3-5 minutes
Smart Queuing Queues requests when rate limited ✅ Enable (recommended)

Service Tier Defaults:

This is an optional setting

QuantumOps reads the dynamic token and usage headers from the Anthropic API to automatically throttle calls and prevent rate limits from being hit.

Tier Token Capacity Request Capacity
Tier 1 40,000/min 50/min
Tier 2 80,000/min 1,000/min
Tier 3 160,000/min 2,000/min
Tier 4 400,000/min 4,000/min
Monthly Billing 100,000,000/min 400,000/min

OpenAI Configuration

Toggle Enable OpenAI to configure GPT models.

API Key:

  • Required for BYOK
  • Leave empty to use managed service

Managed vs. BYOK Comparison

Feature Managed Service BYOK
Setup complexity None API key management
Billing Through QuantumOps Direct with provider
Rate limiting Automatic Manual configuration
Failover Automatic Your responsibility
Cost visibility Aggregated Full granularity

Step 5: Q-Director Core

Configure foundational AI features including real-time analysis, categorization, and stale detection.

[insert screenshot of Q-Director Core tab here]

Incremental Lifecycle Analysis

When enabled, QuantumOps analyzes tickets automatically as they're created and updated via webhooks—no manual refresh required.

You should turn this on

Incremental refreshes save you money while keeping your dashboard data fresh. Incremental refresh configured with Claude Haiku is extremely low cost and saves you money in AI consumption. Without incremental refresh, tickets that are updated can run a full analysis job using your Primary AI Model as frequently as every 2 hours, costing you significantly more. Incremental refreshes prevent full re-analysis of tickets at update and allow us to analyze just what has changed and update our assessment, trajectories, and evaluate for sentiment/tonality alerts.

Setting Default Purpose
Enable Incremental Refreshes ✅ On Process updates to tickets incrementally
AI Model claude-haiku-4-5-20251001 Fast model for incremental work
Stale Recheck (hours) 24 Re-analyze stale tickets periodically (does not use AI)
Scan Interval (minutes) 30 Background scan frequency for finding newly stale tickets (Does not use AI)
Timeline Retention (days) 90 Keep ticket timeline data for this long (recommend 9999 for infinite retention)

Ticket Categorization

Q-Director can automatically assign categories based on ticket content:

Setting Options Purpose
Enable AI Categorization Toggle Turn feature on/off
Timing OnCreation, OnClosure, DuringTriageAssist When to categorize
Allow Category Creation Toggle Let AI create new categories
Category Creation Mode Category, ResolutionCategory, Both What types to create
Re-categorize on Closure Toggle Update category at resolution
Set Resolution Category Toggle Auto-set how ticket was resolved

💡 Recommendation: Start with OnCreation timing and Allow Category Creation enabled. This lets Q-Director learn your patterns while providing immediate value.

Stale Ticket Detection

Configure when tickets are flagged as "stale" and require attention:

Inactivity Thresholds:

Setting Default Purpose
Inactivity Days 3 Business days without activity
Max Ticket Age 14 Days before age-based staleness
Inactivity Trigger Any Activity What resets the counter
Business Days Only ✅ On Skip weekends in calculations

Assignment & SLA Settings:

Setting Default Purpose
Track Agent Follow-Up ✅ On Detect pending agent action
Follow-Up Days 2 Days before follow-up is overdue
Assignment Delay (hours) 4 Hours unassigned before alert
SLA Hold Time (hours) 8 Hours on SLA hold before alert

Health Score Penalties:

Stale conditions apply penalties to ticket health scores:

Condition Default Penalty Impact
Per Day Stale 5 points Gradual decline
Age Threshold Exceeded 15 points Major hit
No Assignment 10 points Ownership issue
SLA Hold Exceeded 10 points Process issue

Step 6: Q-Director Automation

Configure automated ticket handling, routing, and escalation workflows.

[insert screenshot of Q-Director Automation tab here]

Triage Assist

When enabled, Q-Director provides AI-powered triage suggestions:

Feature Purpose
Routing Suggestions Recommend team and agent assignment
Priority Recommendations Suggest priority based on content
Impact/Urgency Setting Auto-determine during triage

Toggle Set Impact & Urgency During Triage to have Q-Director automatically set these values based on ticket content analysis.

Auto-Triage & Assignment

Automatically triage and assign tickets that remain idle:

Setting Default Purpose
Enable Auto-Triage Toggle Activate automatic processing
Idle Time (minutes) 15 Wait time before auto-triage
Schedule Time Zone Your TZ Business hours reference
Triage Outcome ID HaloPSA outcome for triage actions

💡 Recommendation: Start with 30-minute idle time and monitor before reducing. Set to 15 minutes for urgent ticket types.

Research & Sherlock

Configure how AI research results are handled:

Setting Purpose
Post Results to Ticket Add research findings as ticket actions
Custom Action Filtering Exclude specific outcomes from AI analysis
Excluded Outcomes Comma-separated list (e.g., "Rule Applied, SLA Hold")

On-Call Notifications

Alert staff for tickets arriving outside business hours:

Notification Type Configuration
PagerDuty Webhook URL
SMS Phone number
Email Email address

Exclude Clients from Notifications — Comma-separated list of clients that shouldn't trigger on-call alerts.

Escalation Paths

Configure automatic escalation workflows for specific conditions. Escalation rules can trigger based on:

  • Ticket age
  • Priority level
  • Client importance
  • SLA proximity

Step 7: Q-Director Intelligence

Configure advanced AI analysis features including sentiment detection, duplicate identification, and pattern recognition.

[insert screenshot of Q-Director Intelligence tab here]

Sentiment & Tonality Analysis

Detect customer emotional state and communication tone:

Setting Range Purpose
Analysis Sensitivity Low/Medium/High Alert frequency
Sentiment Threshold 1-10 Alert when ≤ threshold
Tonality Threshold 1-10 Alert when ≤ threshold
Feedback Loop Toggle Show feedback buttons in alerts

Custom Instructions — Add business-specific guidance for sentiment analysis:

Ignore standard support frustration. Focus on:
- Threats to cancel service
- Mentions of competitor evaluation
- Executive involvement signals

Duplicate Detection

Automatically identify similar tickets to reduce redundant work:

Setting Options/Default Purpose
Detection Scope SameClient, SameUser, Global Where to look for matches
Lookback Days 90 How far back to search
Ticket Status Open, OpenAndRecentlyClosed Which tickets to compare
Similarity Threshold 75% Match confidence required
Subject Line Matching Toggle Also match on summary text
Auto-Flag Duplicates Toggle Automatically mark duplicates
Show Merge Button Toggle Enable ticket merging UI

Recurring Incident Detection

Identify patterns of repeated issues across users, clients, and globally:

Threshold Default Triggers When
User Level 3 Same user, 3+ similar issues
Client Level 5 Same client, 5+ similar issues
Global Level 10 All clients, 10+ similar issues
Window (days) 30 Lookback period for patterns

ML Feedback Loop

Train Q-Director from corrections:

Setting Purpose
Enable Feedback Loop Learn from manual overrides
Require Manual Review Human approval before retraining
Min Corrections Before Retrain Threshold to trigger learning

Step 8: Web Search Configuration

Configure AI-powered web research for ticket analysis.

[insert screenshot of Web Search tab here]

Provider Options

Provider Setup Best For
Tavily + Bing None required Most MSPs (recommended)
Anthropic Web Search API key with web search enabled More precise and targeted results

Tavily + Bing (Recommended)

Automatically configured with:

  • Pre-configured domain prioritization for tech documentation
  • Automatic result filtering for relevance
  • Integration with ticket context for better results
  • No additional configuration required

Anthropic Web Search

Anthropic Web Search is Extremely Expensive

Anthropic Web Search is a highly precise and targeted AI search engine that returns extremely impressive results, but it is extremely expensive to use. If using Anthropic Web Search, expect your total AI spend to double from the expected/average amount.

When selected, configure:

Setting Default Purpose
Max Searches/Request 5 Limit web searches per AI call
Allowed Domains Empty = all Whitelist specific domains
Blocked Domains Empty Blacklist specific domains

Example Domain Configuration:

Allowed: microsoft.com, docs.microsoft.com, stackoverflow.com
Blocked: facebook.com, twitter.com, pinterest.com

When Web Search is Used

The AI automatically triggers web search when:

  • Looking up error codes
  • Finding vendor documentation
  • Researching known issues
  • Encountering unfamiliar technologies

Step 9: Customer Success Configuration

Configure proactive customer success monitoring and CSAT surveys.

[insert screenshot of Customer Success tab here]

Customer Success Dashboard

Toggle Enable Customer Success Dashboard to activate:

  • Client health score tracking
  • Sentiment trend monitoring
  • At-risk client alerts
  • CSAT tracking
  • Response time monitoring
  • Success reports

CSAT Survey Configuration

Setting Options Purpose
Rating Scale 1-5, 1-10 Survey response options
Link Expiration (days) 1-30 How long survey links work
Post as Private Actions Toggle Hide feedback from customers

Alert Settings

Setting Default Purpose
Silence Duration (hours) 24 How long to silence acknowledged alerts

Step 10: Dashboard Setup

Configure QuantumOps dashboards as iframe tabs within HaloPSA.

[insert screenshot of Dashboard Setup tab here]

Available Dashboards

Dashboard Entity Purpose
Ticket Dashboard Ticket AI insights, health score, trajectories
Agent Assist Ticket Interactive resolution guide
Triage Dashboard Ticket AI-assisted ticket routing
Client Dashboard Client Client profiles and analytics

How to Add Dashboards to HaloPSA

  1. Log into HaloPSA as administrator
  2. Navigate to Configuration → Custom Objects → Custom Tabs
  3. Select the entity type (Ticket or Client)
  4. Click New → Set Type to iFrame
  5. Paste the URL from the wizard and save

Dashboard URLs

Each dashboard provides a URL with HaloPSA placeholders:

https://[your-instance].qops.app/TicketDashboard/[your-tenant-id]/$FAULTID?agentid=$agentid&agentname=$agentname
https://[your-instance].qops.app/TicketTriageDashboard/[your-tenant-id]/$FAULTID?agentid=$agentid&agentname=$agentname
https://[your-instance].qops.app/ClientDashboard/[your-tenant-id]/$CLIENT_ID
https://[your-instance].qops.app/InteractiveGuide/[your-tenant-id]/$FAULTID?agentid=$agentid&agentname=$agentname

The ${id} placeholder is automatically replaced by HaloPSA with the ticket or client ID.

Recommended Tab Order

For Incident/Service Request Tickets:

  1. Ticket Dashboard
  2. Agent Assist
  3. Progress (HaloPSA native)

For Intake/Triage Ticket Types:

  1. Triage Dashboard
  2. Agent Assist
  3. Progress (HaloPSA native)

ℹ️ Note: Dashboard configuration can be completed after the wizard. URLs are also available in Q-Director Settings.


Step 11: Import & Finalize

Import historical data and launch your QuantumOps instance.

[insert screenshot of Import & Finalize tab here]

Data Import Options

Import Duration Purpose
Import Agents 1-2 minutes Sync HaloPSA agents for routing and workload
Import Historical Tickets 10-60+ minutes Train AI on your patterns

Import Agents (Required)

Syncs your HaloPSA agents into QuantumOps for:

  • Intelligent routing suggestions
  • Workload balancing
  • Performance tracking
  • Skill matrix population

Import Historical Tickets (Optional)

Imports past tickets to:

  • Train AI on your resolution patterns
  • Enable similar ticket matching
  • Populate semantic memory
  • Build client/user profiles

💡 Recommendation: Start with 90 days of import only. The import screen allows you to estimate what this training will cost and set a maximum spend for training.

Setup Summary

Before completing, review your configuration:

Setting Status
HaloPSA Connection ✅ Connected
Billing Name ✅ Configured
Primary AI Model ✅ Selected
Q-Director ✅ Enabled
Web Search ✅ Configured
Data Import Ready

Completing Setup

Click Complete Setup & Launch to:

  1. Save all configuration
  2. Start selected data imports
  3. Redirect to main dashboard

Troubleshooting

Connection Test Fails

"Invalid Client ID or Secret"

  • Verify exact values from HaloPSA
  • Check for extra spaces
  • Ensure API application is enabled

"URL not found"

  • Remove https:// prefix
  • Verify HaloPSA instance is accessible
  • Check for typos

"Permission denied"

  • Ensure API application has required permissions
  • Verify admin:all permission is granted
  • Check if API application is disabled

Webhook Not Receiving Events

  1. Verify webhook URL is exactly as provided
  2. Check HaloPSA webhook is enabled
  3. Confirm events are selected (Created, Updated, Action Added)
  4. Test with a manual ticket update

AI Processing Not Starting

  1. Verify AI provider is configured (managed or BYOK)
  2. Check API key validity if using BYOK
  3. Review rate limiting settings
  4. Check consumption limit hasn't been reached

Next Steps


Related Documentation