One Guy, a Laptop, and Ten Patents: How Forge Was Built

I built Forge alone. No co-founder, no venture backing, no team of engineers. Just me, a laptop, and a conviction that the software powering America's contractors was fundamentally broken — not because it lacked features, but because it lacked architecture.

This is the story of how a family business philosophy, a Wyoming Series LLC, and ten provisional patents became an AI operating system for the trades. It is not a story about disruption. It is a story about building something that lasts.

The Origin

My family motto is "Fide et Familia" — Faith and Family. That is not decorative Latin on a crest. It is an operating principle. Everything I build, I build with the assumption that my children and their children will inherit it. That changes how you think about software.

I am Mark Lord, and I work out of Albuquerque, New Mexico. The company is Dominus Foundry LLC, structured as a Wyoming Series LLC — a legal architecture that mirrors the technical one, but I will get to that. Before Forge, I spent years watching contractors struggle with software that was never designed for how they actually work. The tools on the market were built by people who had never stood on a roof, never negotiated a insurance supplement, never dispatched a crew at 5 AM while juggling three adjusters on the phone.

The contractor software market is enormous and poorly served. Not because the incumbents are stupid — they are not — but because they all made the same architectural mistake. They started with a generic platform and tried to bolt on industry-specific features after the fact. CRMs that were really just Salesforce clones with a roofing skin. Project management tools that were Asana with different field labels. Communication systems that were duct-taped together from five different vendors.

I wanted to start from the opposite direction. What if you built the intelligence layer first and let the application grow out of it?

Dominus Foundry was incorporated as a Wyoming Series LLC specifically because the legal structure maps to the technical vision. Each series can hold its own assets, its own IP, its own liability boundary. Each vertical we enter gets its own isolated series. The law and the code follow the same pattern.

The Technical Bet

Forge is not a SaaS product with AI sprinkled on top. It is an AI-native operating system built on a federated multi-kernel architecture.

Let me explain what that means in practice. Most vertical SaaS companies build one monolithic application and sell it to everyone. When they expand into a new vertical, they fork the codebase or stretch the existing schema until it breaks. The result is technical debt that compounds with every new market.

Forge works differently. The core is a shared intelligence layer — what I call MetisOS — that handles identity, communication, spatial reasoning, compliance logic, and negotiation intelligence. On top of that core, each vertical gets its own isolated kernel. A roofing kernel. An HVAC kernel. A solar kernel. Each one is purpose-built for its domain, with its own data models, its own workflows, its own AI agents. But they all share the foundational intelligence.

The stack is Next.js and Supabase on the web side, with native SwiftUI for the iPad applications where performance and hardware access matter — particularly for the LiDAR work. I chose Supabase because it gives me Postgres with real-time subscriptions, row-level security, and edge functions without the overhead of managing infrastructure. I chose Next.js because server components and the app router let me build complex interfaces with minimal client-side JavaScript. And I chose SwiftUI because when you need to talk to an iPad's LiDAR sensor at 60 frames per second, you cannot be going through a web view.

The communication layer — internally called Torch — integrates Telnyx for telephony, Deepgram for real-time transcription, and Claude for conversational intelligence. Every phone call that comes through Forge is transcribed in real time, analyzed for intent and sentiment, and automatically linked to the relevant project. The system does not just record calls. It understands them.

Treasury handles payroll and financial operations through Check HQ's embedded payroll infrastructure. Contractors can run payroll, manage worker classifications, and handle tax compliance without leaving Forge.

Every piece of this was built to be AI-native from day one. The AI is not a feature. It is the foundation.

The LiDAR Breakthrough

This is the part that gets other engineers excited, and it is the part I am most proud of technically.

Hyperion is Forge's spatial intelligence module. It turns an iPad Pro into a commercial roof measurement and structural analysis tool using the built-in LiDAR sensor. Every iPad Pro since 2020 has had a LiDAR scanner — the same core technology that powers autonomous vehicles — and almost nobody in construction is using it properly.

Here is the problem Hyperion solves. When a commercial roofer needs to bid a job, they need accurate measurements of the roof. Traditionally, this means either climbing up there with a tape measure — slow, dangerous, and imprecise — or ordering satellite imagery and tracing it manually — faster but expensive and still imprecise for complex geometries. Some companies use drones, which is better but requires FAA certification, flight planning, and post-processing time.

With Hyperion, a contractor walks the perimeter of a building holding an iPad. The LiDAR sensor captures a three-dimensional point cloud of the structure in real time. Forge processes that point cloud to extract roof geometry, calculate square footage, identify penetrations and obstructions, and — this is the key part — make automated structural adequacy determinations.

That last piece matters enormously in commercial roofing. Before you can specify a roofing system, you need to know whether the existing structure can support it. That calculation depends on local building codes, wind load requirements, snow load tables, and the structural properties of the existing deck. Hyperion automates that entire analysis. What used to take an engineer days of manual calculation now happens in seconds on a tablet.

The 3D models Hyperion generates are not just visualizations. They are intelligent documents that carry measurement data, structural analysis results, and material specifications. They can be shared with adjusters, engineers, and building departments. They become the single source of truth for a project.

The Patent Portfolio

Ten provisional patents. Filed under Dominus Foundry LLC. Here is what they cover and why it matters.

The portfolio breaks down into four domains:

• •Spatial Intelligence — The LiDAR-based measurement and structural analysis pipeline. Methods for automated roof geometry extraction from point cloud data. Systems for real-time structural adequacy determination using device-captured spatial data combined with building code databases.

• •Compliance Automation — Systems that automatically map project specifications to local, state, and federal building codes. Methods for generating code-compliant material specifications from spatial measurements. Automated permit document generation.

• •Communication Intelligence — The real-time call analysis pipeline. Methods for extracting actionable project data from contractor-adjuster conversations. Systems for automated sentiment and intent classification in insurance negotiation contexts.

• •Negotiation Systems — AI-driven supplement identification from damage documentation. Methods for automated scope-of-work generation from field measurements and insurance claim data. Systems for optimizing claim outcomes through historical pattern analysis.

I filed provisionals because they establish priority dates while giving me twelve months to file full utility patents with refined claims. The strategy is deliberate. Each provisional covers a broad technical approach, and the full filings will include specific implementation claims that are harder to design around.

Why does this matter? Because in vertical SaaS, the moat is not the code. Anyone can build a Next.js app. The moat is the intelligence — the proprietary methods for turning raw data into decisions. These patents protect the methods, not the implementation. You can build your own roofing app. You cannot use my method for automated structural adequacy determination from LiDAR point clouds without licensing it.

The First Customer

All Weather Roofing is a commercial roofing company that has been operating in Albuquerque for over 24 years. Their office is about ten minutes from my house. They are the first company running Forge in production.

I chose them deliberately. Not because they were convenient — although that helped — but because they represent the exact profile of contractor that Forge is built for. Established company. Serious volume. Complex commercial projects. Sophisticated enough to understand what the technology does, practical enough to tell me when it does not work.

Building for real users from day one changed everything about how Forge was designed. Every feature in the system exists because someone at All Weather needed it, not because it looked good in a pitch deck. The call transcription works the way it does because I sat next to their office manager and watched how she handled forty calls a day. The measurement tools work the way they do because I went on roof inspections and saw what actually happens in the field.

There is a temptation in software to build for imaginary users. To design features based on what you think people need. Working with All Weather eliminated that temptation entirely. When your first customer is ten minutes away and will call you directly when something breaks, you build differently. You build things that work.

The Vision

I am not building Forge to flip it. I am not building it to raise a Series A and hand it to professional management. I am building it to last.

The model I think about most is Berkshire Hathaway — not the stock price, but the structure. A federation of businesses, each excellent in its own domain, sharing capital and intelligence but operating independently. That is what Dominus Foundry is designed to become. Each vertical we enter — roofing, HVAC, solar, plumbing, electrical — gets its own series in the LLC, its own kernel in the software, its own team when the time comes. But they all share the MetisOS intelligence layer, the patent portfolio, and the operational playbook.

The Wyoming Series LLC structure makes this legally clean. Each series is its own liability boundary. If we enter a new vertical and it does not work, that failure is contained. The other series are protected. This is the same principle as the technical architecture — isolation with shared intelligence.

I think about this in hundred-year terms. Not because I expect to be around in a hundred years, but because building with that time horizon forces better decisions. You do not take shortcuts when you are building something your grandchildren will inherit. You do not accumulate technical debt when the architecture has to last decades. You do not chase trends when the foundation has to be permanent.

The trades are not going away. Commercial roofing, HVAC, plumbing, electrical — these are the systems that make buildings work. The people who install and maintain those systems deserve software that is as well-built as the work they do. That is what Forge is. Not a disruption. Not a pivot. A foundation.

If you are a commercial contractor and this resonates, I would like to hear from you. Call (866) 710-3313 or email foundry@dominusfoundry.com. I am still the one who picks up the phone.

Fide et Familia. Faith and Family. That is how we build.