AI Automation & CRM
AI Voice Agent Sales & Client Onboarding System
Built and evolved a complete GoHighLevel + n8n AI Voice Agent system, starting with a major sub-account cleanup and optimization project and expanding into automated sales, purchasing, client onboarding, sub-account provisioning, Voice AI setup, and QA.
The system connects the customer's purchase journey with automated backend provisioning, allowing each purchased Voice AI service to be tracked, redeemed, and converted into a configured client environment.
01 / CONTEXT
The situation
The Office Manager was developing a service around AI Voice Agents, with GoHighLevel being used as the CRM, sales platform, automation layer, and client environment.
I initially joined the project to help resolve a large number of issues inside an existing GoHighLevel sub-account. What started as a focused troubleshooting and optimization project quickly expanded into the development of the wider Voice AI sales and onboarding system.
The goal became more than simply making the existing setup work. The system needed to support a repeatable process where a customer could purchase a Voice AI service and then move through onboarding, provisioning, configuration, and approval without requiring the entire process to be handled manually.
02 / PROBLEM
Where the work got stuck
The initial GoHighLevel setup had a large number of issues across the account, including problems with workflows, Voice AI behavior, booking, calendars, transfers, email and Zoom configuration, qualification logic, and the overall customer journey.
As the Voice AI service evolved, another challenge appeared:
How do you turn a Voice AI purchase into a repeatable onboarding process?
A customer could purchase the service, but the backend still needed to:
- Recognize the customer and their purchase
- Track the customer's entitlement
- Prevent the same purchase from being redeemed twice
- Support customers purchasing multiple Voice AI products
- Create a new GoHighLevel sub-account
- Configure the Voice AI environment
- Apply the required workflows and settings
- Validate the setup
- Move the client through QA and approval
The system needed to handle these steps reliably while remaining reusable for future customers.
THE SHIFT
From friction to flow.
Before / the friction
The project began with an existing GoHighLevel sub-account containing multiple configuration and automation problems. The Voice AI and CRM setup required significant manual troubleshooting across different parts of the system. There was also no structured entitlement architecture connecting a customer's purchase to the provisioning of their Voice AI environment. The process was largely dependent on manually configuring and validating each customer environment.
After / the system
**Customer purchases
Purchase webhook
Entitlement registered
Onboarding submitted
Purchase validated
Sub-account provisioned
Voice AI configured
QA
Client ready**
The system now follows a structured lifecycle: The architecture also supports multiple purchases for the same customer while maintaining individual purchase records and redemption states. This creates a foundation for scaling the Voice AI service without rebuilding the onboarding process for every customer.
03 / SOLUTION
What I built
I evolved the system into a GoHighLevel + n8n onboarding architecture that separates the customer's identity, purchase entitlement, and onboarding process.
The architecture uses GoHighLevel for the CRM, sales, opportunities, payments, Voice AI, calendars, and client-facing automation, while n8n handles the backend orchestration and provisioning logic.
A purchase can trigger an n8n webhook when the opportunity reaches the Purchased stage.
n8n then:
- Receives and normalizes the purchase data
- Identifies or creates the customer
- Registers the purchase as an entitlement
- Tracks whether that specific purchase has been redeemed
- Receives the customer's onboarding information
- Validates the available entitlement
- Creates the required GoHighLevel sub-account
- Configures the Voice AI environment
- Sets up the required workflows and supporting systems
- Sends the configuration through QA and approval
The entitlement model was designed so that one customer can have multiple purchases, including multiple Voice AI environments, without using email as the unique identifier for purchases.
Each purchase is treated as its own entitlement, allowing the system to distinguish between:
Same customer + different purchase = new Voice AI setup
while also preventing:
Same purchase + repeated onboarding = duplicate provisioning
04 / WORKFLOW
How it works
Initial GoHighLevel Optimization
I started by working through the existing sub-account and resolving the major issues affecting the CRM and Voice AI setup.
This included troubleshooting workflows, calendars, booking, Voice AI behavior, transfers, email configuration, Zoom, qualification logic, and the customer journey.
Voice AI Sales System
The next stage focused on turning the Voice AI into a functional sales and booking assistant.
The system was designed to:
- Handle inbound conversations
- Capture customer information
- Qualify prospects
- Distinguish genuine prospects from vendors and spam
- Handle appointment booking
- Use live calendar availability
- Transfer qualified prospects when appropriate
- Support the sales process without falsely claiming actions had been completed
Purchase → Entitlement
The next layer introduced n8n.
When a customer reaches the Purchased stage in GoHighLevel, the purchase data is sent to n8n through a webhook.
n8n registers the purchase as an individual entitlement.
Instead of putting a simple redeemed flag on the customer, redemption belongs to the purchase itself.
This allows one customer to purchase multiple products while keeping each purchase independently trackable.
Automated Client Onboarding
The onboarding system then validates the customer's available entitlement before provisioning begins.
The architecture uses states such as:
Available → Provisioning → Redeemed
with failure or cancellation states available when necessary.
This prevents duplicate submissions from accidentally creating multiple environments from the same purchase.
GoHighLevel Sub-account Provisioning
Once the purchase is validated, the system can create the customer's GoHighLevel environment and associate the new location with the onboarding record.
The customer, purchase, company, location, and onboarding records remain connected so the system knows exactly which purchase produced which environment.
Voice AI Setup
The provisioning process then moves into Voice AI configuration.
This includes configuring the Voice AI environment, prompts, actions, qualification behavior, booking logic, transfers, and supporting workflows required for the client's setup.
QA & Approval
The final stage is QA.
The newly provisioned environment is checked before it is considered ready for the customer.
This creates a final control point between automated provisioning and client delivery.
05 / ARCHITECTURE
The system behind it
CUSTOMER PURCHASE │ ▼ ┌──────────────────┐ │ GoHighLevel │ │ Opportunity │ │ → PURCHASED │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ n8n Webhook │ │ Purchase Event │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ Customer / User │ │ Identification │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ Purchase │ │ Entitlement │ │ Available │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ Onboarding │ │ Validation │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ GHL Sub-account │ │ Provisioning │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ Voice AI │ │ Configuration │ └────────┬─────────┘ │ ▼ ┌──────────────────┐ │ QA & Approval │ └──────────────────┘
06 / OUTCOME
What changed
The project evolved from fixing an existing GoHighLevel implementation into building a more structured system for selling and onboarding AI Voice Agents.
The result is a repeatable architecture that connects:
Sales → Purchase → Entitlement → Onboarding → Sub-account → Voice AI → QA
Instead of treating every new client as a separate manual setup, the system provides the foundation for provisioning Voice AI environments through a consistent automated process.
It also gives the business a clearer separation between customer records, purchases, and individual onboarding instances, making the system easier to extend as the number of Voice AI customers grows.
TOOLS IN THE SYSTEM
CLIENT PROOF
What the client said.
Next step
Have a similar problem?
Tell me what is happening and we can figure out what makes sense for your business.
No commitment. No pressure. Just a conversation about your process.