I joined Mauricio Cardenal on The AI & Marketing Show for Home Service Pros to break down exactly how I use AI to close more deals at Capital City Roofing — without spending another dollar on lead generation.

The core insight: most contractors assume growth requires more leads. I looked at my business and saw a different problem. Leads were already coming in. Revenue was rotting on the vine. Speed to lead was broken, follow-up was inconsistent, and tribal knowledge locked in people's heads meant the business couldn't scale past whoever was running it that day.

Speed to Lead Is Still the Biggest Revenue Leak

Before any AI system, before any CRM, before any marketing strategy — the most immediate revenue leak for most home service companies is response time.

The first person that can get on the phone with that customer has a good chance of sitting at their kitchen table and giving them a quote. A lot of times homeowners, once they talk to someone, they feel comfortable and stop looking.

The second leak is what happens after the first call. Most contractors give a quote, walk out the door, and have no structured follow-up plan. I call it rotting on the vine.

A structured follow-up sequence — consistent, psychology-backed, running automatically — is the fastest way to recover revenue that's already slipping through your hands.

The 10-Agent AI System

I've built a 10-agent AI orchestration layer that handles everything except going to the customer's house and selling the roof. Here's the actual workflow:

  1. An incoming lead triggers the system
  2. Agents book the appointment, put it on the rep's calendar, send reminders
  3. The system orders the roof measurement report
  4. It attaches the report to the proposal template and generates the full proposal
  5. A task goes to the human: "Ready for review"
  6. The rep checks it, adjusts pricing if needed, and sends it
  7. Agents handle the entire follow-up sequence from there — emails, texts, and prompts to the sales rep to call at specific intervals

It does everything that doesn't require a human. And we do it responsibly and tastefully.

AI Transparency: My Non-Negotiable

During business hours, my intake team answers the phones — real people. But when the team is busy or after hours, AI agents step in with full transparency.

The AI always identifies itself: "This is an AI roofing expert that's been formally trained by Capital City Roofing. But if you need to speak to a human, we have a lot of them ready to answer your questions."

An AI agent answering after hours is not a downgrade — it's a massive upgrade over a voicemail that nobody calls back.

The $200,000 Proof Point

My licensing partner Blake Grissom of Revive Roofing in Charleston had a shot at a 10-to-11 building townhome HOA community. The kind of job that requires a capital improvement plan, annotated photographic report, presentation deck, and slide-by-slide script. Normally takes a week.

I told him to fly the drone, collect the data, drop everything into a structured shared folder. I plugged it into the system.

The output was instant: a thorough, accurate, beautifully formatted proposal package.

The HOA board said they had never seen something so thorough, so beautifully written, so fast, ever. They awarded the $200,000 job without entertaining another bid. The competing contractors hadn't even gotten their proposals back — and probably wouldn't have for seven to ten business days.

Speed, accuracy, and professionalism — all delivered by the system.

The Tech Stack: Claude, Google Drive, and NotebookLM

I use Claude as my primary AI model, connected to Google Drive and NotebookLM. Every SOP, pricing detail, service description, and customer script lives in NotebookLM as a structured knowledge base. When my agents prompt against that system, they're pulling from Capital City Roofing's actual data — not guessing.

There's no chance of hallucination. It's based on our company data that we have uploaded into the notebooks. The output is very consistent.

The system isn't complicated to set up, but it requires discipline: everything the AI might need has to be documented and loaded. None of this works without first building your SOPs.

Marketing Budget Benchmarks

I'm direct about marketing spend: most contractors are underfunding it, which is why they feel like marketing doesn't work.

My benchmarks:

  • 5% of revenue to maintain your current position
  • 10% of revenue to grow at meaningful velocity

But don't spend 10 until you know what you're doing. Start at 5% and work your way up. You need to know your cost per lead, cost per booked appointment, booking rate by lead source, and the ROI of each channel. Without those numbers, adding budget just means spending more to get the same results.

BuilderLync and the Licensing Platform

BuilderLync is the CRM I co-founded with three other roofing company owners and a developer. Because we all own roofing companies, the platform is built around what roofing businesses actually need. Currently in beta, targeting a V1 launch in June. $500/month unlimited seats, $1,000/month enterprise.

For contractors who want more than just the CRM, the Capital City Roofing licensing platform offers three tiers: $2,500/month for startups (includes the tech stack, website, and systems), up to $12,000/month for full-service operations where Capital City runs both the front-end (phone team, appointment booking, CRM management) and the back-end (accounting, bookkeeping, compliance, warranty filing).

If you were to outsource all of that to third parties, it would be significantly more expensive. And if you hired W2 employees to do it all, even more than that.

Key Takeaways

  1. Fix speed to lead first. The contractor who calls back fastest wins.
  2. AI agents are better than voicemail — always. After-hours AI coverage is a legitimate upgrade right now.
  3. Be transparent when AI is talking to customers. Saying "this is an AI roofing expert trained by our company" builds trust; pretending it's human destroys it.
  4. Pick one AI model and go deep. Claude, ChatGPT, Gemini — stop platform-hopping and learn one well.
  5. Connect your AI to your own data. Claude + Google Drive + NotebookLM eliminates hallucinations.
  6. Document your SOPs before you automate. AI can't replicate tribal knowledge. Get it out of people's heads first.
  7. Spend 5% to maintain, 10% to grow — but only increase once you know your numbers.
  8. AI frees humans for human work. The goal isn't to replace your team — it's to get them in front of customers.

A portion of every roof Capital City Roofing replaces funds the Feeding the Future Project — our nonprofit working to feed one million children in ten years. The systems exist so the mission can scale.

Listen / Watch

If you're a homeowner — capitalcityroofing.net

If you're a contractor — capitalcityroofing.net/licensing

If you want the CRM — builderlync.com