Built Different. On Purpose.

HomeRunBook isn't another app with a database and some forms. It's built on graph theory, proprietary domain languages, and AI that understands building science. Here's why that matters — and why it's nearly impossible to replicate.

Why Competitors Can't Just Copy This

Every technology choice in HomeRunBook creates compounding advantages. These aren't features — they're architectural decisions that would take years and millions to replicate.

9,500+

Component Definitions

Every component definition includes physical specs, electrical ratings, port types, connection rules, and trade-specific metadata. Years of building science encoded into a curated library.

11

Building Trades

Electrical, plumbing, HVAC, IT, AV, security, fire, solar, gas, irrigation, and telecom. Each with trade-specific semantics, validation rules, and component libraries.

2

Proprietary Languages

CML (Component Modeling Language) defines components. TFL (Topology Formula Language) computes values and validates codes. Both are patent-pending.

DAG

Graph Architecture

Directed acyclic graph data model. Not a flat table. Not a tree. A real graph that captures dependencies, enables traversal, and powers AI reasoning.

3

AI Models Working Together

Gemini for conversation, Gemini Vision for photo recognition, and Gemini for document intelligence. Each model has topology context that generic AI chatbots lack.

<50ms

Real-Time Sync

Custom binary WebSocket protocol with MessagePack encoding. Redis pub/sub for cross-instance coordination. Instant collaboration, not eventual consistency.

Graph-Based Data Model

While competitors store appliances in flat database tables, HomeRunBook models your home as a directed acyclic graph (DAG). Every component is a node. Every connection is an edge. And the relationships between them are what make the platform intelligent.

This isn't just a data structure choice — it's the foundation of everything. Graph traversal enables the AI to trace paths ("what feeds this outlet?"). The formula engine can aggregate values across connected components ("total load on this circuit"). Validation rules can check cross-system dependencies ("this HVAC unit needs a 30A breaker, but it's on a 20A").

Flat databases can't do this. You'd need to rebuild the entire backend from scratch — and re-engineer every feature that depends on relationships. That's why nobody else offers topology-aware intelligence.

// Graph traversal: "What's upstream of the dishwasher?"

Dishwasher
  ← NEMA-5-20RGFCI Outlet
    ← 12/2 NM-BBreaker #7 (20A)
      ← Bus BarMain Panel (200A)
        ← Service EntryUtility Meter

// Result: 5 nodes, 4 edges, complete path traced
// Each edge has type, gauge, capacity metadata

CML: How We Define Building Components

Every one of our 9,500+ components is defined in CML — our proprietary Component Modeling Language. It's not a product database. It's a formal specification that captures the physics, semantics, and constraints of building equipment.

Physical specifications — dimensions, weight, rack units, power consumption, BTU output. Not marketing specs; actual engineering data.

Port definitions — every way a component can connect to another component. A 200A panel has bus bar ports. A NEMA 5-15R receptacle has a plug port. An RJ45 jack has a data port. Port types enforce that only compatible components can connect.

Traits — behavioral tags that enable intelligent grouping and cross-trade relationships. A "powered" trait means the component needs electrical supply. A "networked" trait means it has data connectivity. Traits let the system reason about components abstractly.

Properties — typed fields with enums, ranges, and defaults. Not free-text — structured data that formulas can compute against.

// CML: Component Definition
component "Square D HOM2040" {
  trade: "electrical"
  category: "panel"
  manufacturer: "Square D"

  specs {
    amperage: 200
    spaces: 40
    voltage: "120/240V"
  }

  ports {
    input "service_entry" { max: 1 }
    output "breaker_slot" { max: 40 }
  }

  traits: ["powered", "grounded"]
}

TFL: Excel for Your Home's Topology

TFL is our proprietary formula engine that performs calculations across the graph. Think of it as Excel formulas, but instead of referencing cells, you reference connected components.

Graph-aware functions like UPSTREAM(), DOWNSTREAM(), and CONNECTED() traverse the topology to find related components. SUM the amperage of all devices downstream of a breaker. COUNT the number of devices on a network segment.

Validation rules written in TFL automatically check building codes. "IF circuit_load > breaker_rating * 0.8 THEN WARNING." These rules evaluate in real time as you build your topology.

Computed properties calculate values you'd otherwise have to figure out manually. Total panel load. Voltage drop over a wire run. BTU per square foot for a zone.

// TFL: Validation Rules

// Check 80% breaker load rule (NEC 210.20)
RULE "breaker_load_80pct" {
  load = SUM(DOWNSTREAM(self).amperage)
  capacity = self.amperage * 0.8
  IF load > capacity THEN
    WARNING("Circuit at {load/self.amperage*100}%")
}

// Calculate voltage drop
COMPUTED "voltage_drop" {
  result = (2 * self.wire_length * load) /
           (CONDUCTIVITY(self.gauge) * self.voltage)
  IF result > 0.03 THEN
    WARNING("Drop exceeds 3%")
}

Three AI Models, One Purpose

HomeRunBook doesn't bolt a chatbot onto a database. We integrate three specialized AI capabilities, each with full context of your home's topology.

Conversational AI (Gemini) powers the property assistant. It receives your full topology graph, all component metadata, documents, and maintenance history as context. When you ask "why is my bedroom cold?", it traces the HVAC path in the graph to identify potential failure points — something a generic AI physically cannot do.

Vision AI (Gemini Vision) powers Visual Add. It's not just OCR — it understands equipment labels, recognizes common building equipment, and maps extracted data to our CML component definitions. It knows that "Carrier 24ACC636" is a 3-ton heat pump and fills in the specs automatically.

Document AI (Gemini) processes uploaded PDFs and images, extracting structured information that becomes part of the AI's knowledge base. Upload a warranty document, and the AI can answer warranty questions for that specific component.

Your Home Topology
Conversational
Q&A + Diagnostics
Vision
Photo Recognition
Document
PDF Intelligence

Enterprise-Grade, Consumer-Simple

Built on Google Cloud Platform with the same reliability standards as enterprise SaaS, but with an interface anyone can use.

Google Cloud Platform

Cloud Run for auto-scaling, Cloud Storage for documents and photos, Cloud CDN for global performance, Firestore for serverless NoSQL.

Secure by Design

End-to-end encryption, per-user access controls, permission-based sharing, and SOC 2-aligned security practices. Your home data stays yours.

Real-Time Performance

Binary WebSocket protocol for instant sync. Redis pub/sub for multi-instance coordination. Sub-50ms latency for collaborative editing.

See the Technology in Action

Create your free account and start mapping your home's systems. Experience the difference that graph-based intelligence makes.

Start Free See All Features