Case Study
n8nstudio.tech · June 2026
Parker
AI Lead Generation Agent for Texas Elite Roofing
A fully automated 13-workflow n8n system that discovers leads across 6 platforms, enriches them with AI, scores by priority, and calls hot prospects via voice AI — every 30 minutes, without any manual work.
13
Workflows
6
Lead Sources
24/7
Automated
0
Manual Work
The Problem
Manual lead gen wasn’t fast enough
Texas Elite Roofing operates in one of the most competitive roofing markets in the US. DFW sees frequent hailstorms — but the opportunity only goes to whoever reaches homeowners first.
Too Slow
Manual lead generation relied on referrals and word of mouth — no systematic way to find homeowners with fresh storm damage.
No Storm Awareness
No system to monitor weather events and immediately identify affected DFW neighborhoods with potential roof damage.
No Outbound Calling
Hot leads sat in a spreadsheet waiting for a human to call — losing deals to faster-moving competitors every hour.
The Core Problem
In DFW roofing, the company that contacts a storm-damaged homeowner first wins the job. Every hour of delay after a hail event is lost revenue. Texas Elite Roofing needed a system that never sleeps, never misses a lead, and calls prospects automatically — before the competition does.
The Solution
Parker — 13 workflows, fully automated
Parker handles the complete lead-to-call pipeline without any human involvement. From discovering potential leads across 6 data sources, to enriching with AI, scoring by priority, and calling the hottest prospects using a natural voice AI agent.
| Layer | Workflows | What It Does |
|---|---|---|
| Discovery | WF02–WF06 | Scans Google, Maps, Facebook, Nextdoor, LinkedIn, Instagram, TikTok + live DFW weather |
| Intelligence | WF07, WF08 | xAI Grok AI enrichment + proprietary lead scoring engine (0–100 score) |
| CRM & Alerts | WF09, WF10 | Google Sheets CRM upload + instant Gmail escalation alerts for hot leads |
| Voice AI | WF11, WF12 | Retell AI outbound calls every 30 min + inbound call handling with AI transcription |
| Infrastructure | WF00, WF01 | Model abstraction layer (swap AI models via Supabase) + master orchestrator |
n8n Cloud
Core automation — 13 workflows & all data flow
Retell AI
Voice agent — outbound + inbound calls
xAI Grok
Lead enrichment & context generation
Groq LLM
Inbound transcript processing & extraction
SerpApi
Google Search & Maps lead discovery
Apify
Social media scraping across 5 platforms
Tomorrow.io
Real-time DFW hail & storm detection
Google Sheets
CRM database — 7-tab pipeline tracking
Supabase
AI model config for abstraction layer
Gmail
Instant Tier 1 escalation alerts
How It Works
From lead discovery to first conversation
Parker runs the entire pipeline automatically. Here’s exactly what happens every weekday.
1
WF01 — Orchestrator
8AM: The system wakes up
Every weekday at 8AM, the Orchestrator fires all lead scan workflows simultaneously, waits for completion, then runs enrichment → scoring → CRM upload → alerts → calling in sequence.
WF01 — Daily Schedule Trigger → Initialize Session → Parallel Scans → Enrichment → Scoring → CRM → Alerts
2
WF02–WF06 — Lead Discovery
6 sources scanned in parallel
Parker simultaneously searches Google, Maps, Facebook Groups, Nextdoor, LinkedIn, Instagram, and TikTok for DFW homeowners with storm damage or roofing needs — plus monitors live weather for hail events.
WF02 — SerpApi searches Google + Maps → Parse → Deduplicate → Save to Raw_Leads
WF06 — Tomorrow.io weather for Dallas, Frisco, DFW → Analyze → Storm Detected? → Log to Storm_Triggers
3
WF07 — AI Enrichment
xAI Grok analyses every lead
Every raw lead is processed by xAI Grok (grok-3-mini), which generates ownership probability, storm damage likelihood, estimated roof age, and personalised outreach notes Parker uses during calls.
WF07 — Fetch Raw_Leads → xAI Grok AI enrichment → Save to Enriched_Leads
4
WF08 — Lead Scoring
Every lead scored 0–100
Parker’s proprietary scoring engine assigns every enriched lead a score and a tier. Only the hottest leads get called automatically — the rest go to the CRM for manual follow-up.
Tier 1 — Hot
60–100
Instant Gmail alert + Parker calls automatically
Tier 2 — Warm
25–59
Added to CRM for manual sales team follow-up
Tier 3 — Cold
0–24
Archived for future reference — no immediate action
WF08 — Fetch Enriched_Leads → Parker Scoring Engine → Save to Scored_Leads
5
WF09 + WF10 — CRM & Alerts
Leads uploaded, team notified instantly
All scored leads are formatted and uploaded to the Parker CRM Google Sheet. A daily summary report is auto-generated. For every Tier 1 lead, a Gmail alert fires immediately.
WF10 — Fetch Tier 1 Leads → Build Alert → Send Gmail → Log Escalation Activity
6
WF11 — Outbound Calling
Parker calls every 30 minutes
Every 30 minutes, 9AM–8PM CST Mon–Fri, Parker checks for uncalled Tier 1 leads and dials them via Retell AI. Parker qualifies the homeowner, captures key info, and logs every outcome automatically.
WF11 — Schedule (30 min) → Calling Hours Check → Read CRM → Filter Tier 1 Uncalled → Build Retell Payload → Call → Log → Update CRM
7
WF12 — Inbound Handling
Inbound calls handled 24/7
When a prospect calls the Parker number, Retell AI answers instantly. The transcript is processed by Groq AI, the lead is scored and saved to CRM, and a Tier 1 alert fires if they’re hot.
WF12 — Retell Webhook → Parse → IF Call Ended → Groq AI Transcript Extract → Score → Save to CRM + Call_Logs → Alert
Supporting Workflows
Social Media Scans
WF03 — Facebook & Nextdoor: Apify scraping → Parse signals → Merge → Save to Raw_Leads
WF04 — LinkedIn: People + Company search via Apify → Merge → Save to Raw_Leads
WF05 — Instagram & TikTok: Hashtag scans → Parse signals → Merge → Save to Google Sheets
CRM Upload & Model Layer
WF09 — CRM Upload: Fetch Scored_Leads → Format → Upload to CRM_Parker_Leads → Generate Daily Report
WF00 — Model Abstraction Layer: Fetch active AI model config from Supabase → Extract → Return to caller
Infrastructure
CRM, Retell Agent & Data Layer
Parker maintains a 7-tab Google Sheets CRM, a configured Retell AI voice agent, and a Supabase backend — all connected and automated.
Parker CRM — Sales Team View
CRM_Parker_Leads tab — Lead ID, Full Name, Phone, Property Address, Lead Tier, Lead Score, Call Status, all auto-populated by Parker
Raw Leads — All Scraped Data
Raw_Leads tab — every scraped lead from all 6 sources with source, title, link, snippet, date found, and raw JSON data
Retell AI Voice Agent
Parker’s Retell AI agent — GPT 5.1 model, Leland male voice, $0.11/min, 970–1450ms latency, full qualification script configured
Supabase — AI Model Config
Supabase agent_config table — switch between Claude, Grok, and Gemini by updating one database row, no workflow edits needed
Results
What Parker delivers
13
Workflows Deployed
6
Lead Sources Active
30m
Calling Frequency
100%
Test Pass Rate
| Metric | Before Parker | After Parker |
|---|---|---|
| Lead discovery | Manual — hours per week | Automated every weekday 8AM |
| Sources monitored | 1–2 (referrals, Google) | 6 simultaneous platforms |
| Storm response | Hours to days | Detected automatically, leads prioritised instantly |
| Outbound calling | Manual — limited by team hours | Every 30 min, 9AM–8PM CST, fully automated |
| Lead scoring | Gut feeling | AI-scored 0–100, 3 priority tiers |
| CRM updates | Manual data entry | Automatic — every lead and call logged instantly |
| After-hours coverage | None | 24/7 inbound via Parker voice agent |
| Team time on lead gen | Several hours/week | Near zero |
Key Outcome
Texas Elite Roofing now has a fully autonomous lead generation machine running every business day. Parker discovers leads, qualifies them with AI, scores by priority, and calls the hottest prospects — all before the sales team starts their morning.
Built by n8nstudio.tech
Want a system like Parker?
I build production-grade AI automation systems for home services, HVAC, roofing, and real estate businesses.
Get in Touch