Lead Capture
A public-facing website form collects the prospective client's name, email address, company, and inquiry details.
05 / AI & AUTOMATION
Building an AI-powered lead qualification system.
An integrated marketing automation system connecting website inquiries, AI-powered evaluation, and HubSpot CRM, with human oversight built into the follow-up process.

01 / OVERVIEW
Lead qualification is an important part of marketing and sales operations, but reviewing inquiries, evaluating their relevance, updating CRM records, and coordinating follow-ups can involve repetitive manual work.
I designed and developed an AI-powered lead qualification system that connects website inquiries with automated evaluation and HubSpot CRM management.
Built around Northstar Digital Studio, a simulated full-service digital marketing agency, the project demonstrates how AI and workflow automation can support lead management while keeping human judgment central to follow-up decisions.
02 / THE CHALLENGE
Digital marketing agencies receive inquiries from prospective clients with different needs, budgets, and levels of relevance to their services.
Without a structured qualification process, teams may spend considerable time reviewing submissions, manually entering information into their CRM, and deciding which opportunities require attention.
The objective was to capture and validate inquiries, evaluate their relevance, record consistent recommendations in a central CRM, and support human approval before follow-up — reducing repetitive administrative steps without replacing human judgment.

03 / THE SOLUTION
I built an integrated marketing automation system using n8n, Groq AI, HubSpot CRM, and a publicly accessible inquiry form hosted through Cloudflare.
When a prospective client submits an inquiry, the system validates the information and evaluates it against the agency's services, including digital marketing, SEO, paid advertising, social media, analytics, and marketing automation.
Each lead receives a suggested priority, numerical score, qualification rationale, and recommended next action, recorded in HubSpot for human review. A separate workflow supports follow-up task creation for approved leads.
04 / HOW THE SYSTEM WORKS
A public-facing website form collects the prospective client's name, email address, company, and inquiry details.
n8n receives the submission, normalizes the information, and checks required fields before processing.
A Groq-hosted AI model evaluates each inquiry and returns a suggested priority, score from 0–100, rationale, and recommended next action.
The automation creates or updates a HubSpot contact, storing the qualification results in custom properties. New leads enter Pending Human Review.
A reviewer examines the inquiry and AI recommendation in HubSpot before approving or rejecting the lead.
A separate workflow checks for approved leads without an existing follow-up flag, creates an associated HubSpot task, and updates the contact.
THE ACTUAL AUTOMATIONS
The first workflow handles inquiry validation, AI evaluation, and CRM updates. The second checks human-approved leads and creates follow-up tasks.


05 / HUMAN OVERSIGHT BY DESIGN
One of the central design decisions was to treat AI qualification as decision support rather than an automatic sales decision.
The AI evaluates inquiries and recommends priorities, but it does not independently approve leads or initiate outreach. Human reviewers retain control over which opportunities proceed.
The follow-up workflow checks whether a task has already been created, helping prevent duplicate activities and unnecessary CRM updates. The system does not automatically send sales messages.
06 / TESTING & VALIDATION
The system was tested using 20 synthetic lead scenarios representing different inquiry types and qualification priorities. These tests evaluated submission processing, structured AI assessments, and HubSpot record creation.
An additional submission through the public Northstar website successfully completed the intake process, from form submission to AI qualification and HubSpot contact creation.
The approval and follow-up workflow was tested separately. Approved leads generated associated HubSpot tasks, while pending and rejected leads were excluded.
These results demonstrate functionality in a controlled testing environment, not commercial conversion rates, productivity savings, or production-scale performance.

07 / TECHNOLOGY & TOOLS
Workflow orchestration, validation, conditional logic, and CRM automation
AI-powered lead evaluation, scoring, and recommendations
Contact management, qualification records, approval statuses, and follow-up tasks
Public website hosting and connectivity to the locally hosted automation environment
Data normalization, AI response parsing, and workflow logic
08 / WHAT THIS PROJECT DEMONSTRATES
This project brings together marketing operations, AI-assisted decision-making, CRM management, and workflow automation in a connected system. It demonstrates my ability to translate a business process into an automation architecture, integrate multiple technologies, structure AI-generated outputs for operational use, and validate workflow behavior through testing.
More importantly, it reflects an approach to automation that considers not only what can be automated, but where human judgment and control should remain.
Northstar Digital Studio is a simulated digital marketing agency developed for this independent project. The workflows use synthetic lead information and a locally hosted n8n environment. The public website serves as a functional demonstration, with form processing dependent on the local automation environment being online. The system has been tested for the documented workflows but has not been deployed or evaluated as a production-scale commercial service.