Jack at the AI Automations by Jack school community invited me to teach a masterclass for his school. The room was full of agency owners, automation builders, and home services operators who wanted to see exactly how I run Capital City Roofing on an AI-first operating system. The full session is on YouTube. This post is the longer version, with the focus on the Capital City Roofing Licensing Platform and BuilderLync that the live session deserved.

The headline number

Two years ago, I started Capital City Roofing and we were doing about three to four million in revenue. We were doing okay on paper. We were drowning in practice. We were running on eleven different software tools that did not talk to each other and we were losing roughly thirty percent of our inbound leads to slow response time. The math was unforgiving. Every month the same operational gaps cost us deals we should have closed.

Today we are running at a ten million dollar run rate. The market is the same. The crew size is the same. The only thing that changed is that we built an AI operating system on top of the company. That is the headline number. Eighty agents, five full-time employees, ten million dollar run rate.

The masterclass walks through how that system works. This post walks through why it matters for any home services operator considering how to think about AI in their own business, with a deliberate focus on the two pieces of the system that are now available to operators outside our flagship operation: the Capital City Roofing Licensing Platform and BuilderLync.

The system in three layers

The architecture I showed in the masterclass has three layers, and I want to make them legible here for anyone who wants to think about it independently.

Layer 1: the operational core. These are the agents that me and my employees use every single day. Lead intake, lead nurturing, reputation management, estimating, sales, closing. They are intuitive. They do specific things. They do them consistently. The operational core is what makes the company run when nobody is watching the dashboard.

Layer 2: the specialists. Marketing, web operations, recruiting, financial analysis, all run by agents with specific knowledge bases and specific outputs. We brought marketing in-house and the marketing layer is now an internal agency that runs on agents we trained on our own brand voice, copy library, and conversion data. The website you are reading this on is managed agentically.

Layer 3: the orchestrator. I can text my orchestrator through Telegram and tell it to do anything across the rest of the system. The orchestrator is the boss. It figures out which agents to invoke, in what order, and reports back when the work is done. That is what makes the whole stack feel like an operating system rather than a pile of tools.

Underneath all three layers is a triple-layer memory architecture: Google Drive, Zengram, and NotebookLM, all kept synced and consistent. Some people ask me whether the redundancy is overkill. It is not. If Google pulls the plug on NotebookLM tomorrow, my data does not go away. If Zengram has an issue, the other two carry the operation. This runs my business. Resilience matters.

Why agents need a knowledge base, not just a model

The most important insight I gave in the masterclass is that the model is not the differentiator. Context is.

When I prompt an agent to do something, I am not relying on whatever the LLM happens to know about roofing or business operations. I am pointing it at a very specific knowledge base I built around Capital City Roofing's actual operations. Our pricing. Our SOPs. Our brand voice. Our certifications. Our service areas. Our customer service standards. Every piece of institutional knowledge lives in a structured repository the agents can read.

That is why my agents do not hallucinate. That is why a customer talking to one of our agents gets a consistent experience that matches how Capital City Roofing actually operates. The model is interchangeable. The knowledge base is the moat.

The same principle applies in every other home services vertical. A plumbing company, an HVAC company, a landscaping company, a pest control company. The operational reality is different. The architecture is the same. Build the knowledge base first. Layer the agents on second. Anything else is automating garbage.

The skill.md insight

I gave a line in the masterclass that I want to repeat here because it is the most important sentence about how to actually scale a system like this.

Everything AI is fantastic, but if you cannot duplicate it, it is useless.

I tried teaching my employees to use the AI stack the way I do. They could not. Not because they are not smart, they are. But because the wizard-behind-the-keyboard model collapses the moment one wizard has to scale across a team. Every operator does the same thing slightly differently. Every customer interaction becomes a one-off. The benefits of consistency disappear.

The fix was to write skill.md files for human beings. They are SOPs. We roll them out. We get the team trained. We get them comfortable. Now the agents and the humans are running on the same playbook, the same expectations, and the same standards. That is the only way the system holds together as the company grows.

This is also the reason the licensing platform works. A licensee in another market is not learning to be Brad. They are learning to run on the same skill.md files, the same SOPs, the same operating system, and the same knowledge base. The platform replaces the wizard with structure.

Why the Capital City Roofing Licensing Platform is the focus

I told Jack and the room directly: everything I taught in the masterclass converges on the Capital City Roofing Licensing Platform and BuilderLync. Those are the two pieces an operator outside our flagship operation can actually access.

The licensing platform is structurally different from a franchise. The buy-in is fifteen thousand dollars instead of fifty plus that most roofing franchises charge. The royalty is five percent capped instead of an open-ended ten percent or higher. The contract is a one-year auto-renewing term instead of a multi-year lockup with painful exit fees. The economics are designed so the platform wins when the operator wins.

But the bigger story is what licensees inherit. They inherit the entire operating system. The SOPs. The brand. The certifications (GAF Master Elite, GAF Commercial Certified, CertainTeed ShingleMaster Premier, GenFlex Commercial Certified, Roofing Alliance Guarantor Member). The training program through Capital City University, our internal LMS. The marketing playbook. The financial systems. The agentic AI workforce. And the technology stack that ties it all together.

A licensee on the platform is not building from scratch. They are running on day one with the operating system that took us two years to build. That is what differentiates a real operating platform from a franchise that gives you a brand and a rulebook.

We have already grown to four locations on this model. We are scaling by people raising their hand and saying they want to start a Capital City Roofing in their market. The vetting process is real. The economics are clean. The conversation starts at licensing@capitalcityroofing.net.

Why BuilderLync is the technology layer

The other half of what I taught is BuilderLync, the AI-driven CRM and operating platform my partners and I built specifically for contractors. BuilderLync is what every Capital City Roofing licensee runs on, and it is also available standalone for any contractor who wants the technology layer without joining the licensing platform.

The reason we built BuilderLync rather than continuing to white-label something else is that we tried everything else. Generic CRMs forced us to bend our operation around the software's idea of how a sales team works. Vertical CRMs that were retrofitted from another trade had the wrong conventions baked in. We tried to make a high-level workflow tool work and could never get the estimating and proposals to function for roofing.

So we built it. BuilderLync includes:

  • A unified contact system. Customers, employees, manufacturers, distributors, vendors, all in one place. No more parallel spreadsheets.
  • A complete jobs and pipeline system with kanban views and division-specific pipelines for residential, commercial, multifamily, and insurance.
  • Integrated payments via Stripe and bookkeeping via QuickBooks, with real-time accounting visibility.
  • Sierra AI as the native voice and text agent, embedded across the customer journey.
  • Job photo capture (the same idea as Company Cam) built into the CRM, with required photo workflows that block forward progress until photos are captured. The customer gets a comprehensive inspection report. The office gets the visual data the crews need.
  • Instant estimator with live satellite measurement and material catalog integration. I can text an address from Telegram, ask for an estimate, and the system produces a 175,000-dollar TPO commercial roof estimate in seconds.
  • DIY satellite-based roof measurement with facet mapping and per-component material calculations.
  • Eagle View integration for full institutional measurement reports.
  • Material order integrations with ABC, QXO, and SRS.
  • Automation builder, recruiting tools, marketing operations, file management, and AI-driven analytics on every dashboard.

The pricing matches the philosophy. The base tier is five hundred dollars a month with unlimited seats. The full package with the agentic AI layer is one thousand dollars a month with unlimited seats. No per-user charges. No token throttling.

The V1 public launch is June 1, 2026. The trial sign-up opens at builderlync.com.

What this means for the room Jack put me in front of

Jack's school community is full of agency owners and automation builders, and the question I got most often after the masterclass was some version of "how do I bring this to my own clients?" The answer is the same regardless of vertical.

Build the knowledge base first. Pick one client, document everything about their operation, and put it in a structured repository. Knowledge base before agents.

Layer agents on second. Start with the highest-friction operational gap. For most home services operations, that is response time on inbound leads. Solve that one with a structured agent. Iterate.

Build the orchestrator third. Once the operational core is running, give yourself a single channel (Telegram works well) where you can talk to the entire system. The orchestrator turns a pile of agents into an operating system.

Document the SOPs. Whatever the agents do, the humans need a parallel set of skill.md files so the system can scale across team members. The moment you stop being the only person who understands how the system works, the leverage compounds.

If you are working with a roofing client specifically, the fastest path is to bring them onto BuilderLync and onto the Capital City Roofing Licensing Platform together. The knowledge base, the SOPs, the technology stack, and the agentic layer are all already built. Your client gets the result. You get to spend your time on the parts that actually require judgment.

Where to find the masterclass and the next conversation

The full masterclass replay is on the AI Automations by Jack channel, embedded above and on the Skool community classroom. Jack runs a great community for builders and operators. If you are in this space and not in his school, you should be.

If you want to talk about the Capital City Roofing Licensing Platform, the conversation starts at licensing@capitalcityroofing.net. I read every one of those personally.

If you want only the technology layer, BuilderLync is the path. V1 trial sign-ups open June 1, 2026.

If you are in Greater Atlanta or Nashville and you need a roof on a residential, multifamily, or commercial property, Capital City Roofing handles it directly.

Thank you to Jack and his team for the invitation. The masterclass was the first time I taught this system end to end. The conversation continues.

Keep Exploring

Related reads on the operating system, the licensing model, and BuilderLync:

About Brad Strawbridge

Brad Strawbridge is the Founder and CEO of Capital City Roofing, a GAF Master Elite, GAF Commercial Certified, and CertainTeed ShingleMaster Premier roofing company serving Greater Atlanta and Nashville with residential, multifamily, and commercial roofing. He is also Co-Founder and Chief Strategy Officer of BuilderLync, an AI-driven CRM and project management platform built for contractors, and Founder of the Feeding the Future Project, a 501(c)(3) nonprofit working to feed one million children in ten years. Brad is an active member of RT3 (Roofing Technology Think Tank), NRCA, and The Roofing Alliance.

bradstrawbridge.com | LinkedIn | capitalcityroofing.net

Tags: AI in roofing, AI-first roofing business, agentic AI workforce, 80 agents, Capital City Roofing Licensing Platform, BuilderLync, AI Automations by Jack, Skool community, masterclass, Brad Strawbridge, knowledge base architecture, NotebookLM, Zengram, Telegram orchestrator, contractor CRM, roofing franchise alternative