Fiber Network Planning Engine

FiberCAD

Automated fiber network design that turns addresses into construction-ready plans in minutes, not weeks. Standards-compliant. Integrity-verified. LLM-assisted.

10x
Faster than manual CAD
$0
Esri licensing fees
100%
Standards-validated output

The Problem

Billions in BEAD funding. Not enough engineers to design the networks.

Today's Workflow

  • Manual CAD design — experienced designer draws every conduit, splitter, and handhole by hand in AutoCAD
  • 16-step process — from EOP linework to final QC, each step requires domain expertise
  • 3-GIS costs $50K-$500K/year — plus Esri licensing at $5K-$15K per seat, plus 6-18 month deployment
  • Talent bottleneck — takes months to train a designer, years to build judgment for optimization
  • QC is manual — 17-step review checklist performed by a human on PDF output

FiberCAD

  • Automated planning — feed in an area boundary, get a complete network design with splitters, conduit, and equipment
  • Standards built-in — design rules are the software, not a PDF that humans interpret
  • No Esri dependency — open-source GIS stack, cloud-native, zero vendor lock-in
  • LLM-assisted review — AI evaluates trade-offs and suggests optimizations that normally require senior engineers
  • Self-validating — every design is checked against standards before output, with a cryptographic integrity chain

How It Works

Six layers of graduated enrichment, from raw addresses to construction-ready output.

L0
Input
Address list or area boundary
L1
GIS Data
Buildings, roads, parcels from OSM
L2
Topology
Boundary groups, splitter tree
L3
Physical
Conduit routes, equipment
L4
Compliance
Standards validation
L5
Artifacts
DXF, BOM, reports

Competitive Positioning

Not another GIS tool. A design automation engine with built-in intelligence.

Capability 3-GIS VETRO FiberMap IQGeo / Comsof FiberCAD
Automated splitter planning Partial Partial Yes Yes
Provider-specific rule engine Config No Yes Yes
Self-validating output No No No Yes
Integrity chain (signed layers) No No No Yes
LLM-assisted design review No No No Yes
No Esri dependency Requires Esri Yes Optional Yes
AutoCAD DXF output Partial No No Yes
Deployment time 6-18 months Weeks Months Minutes
Annual cost (est.) $50K-$500K+ $10K-$50K $25K-$200K TBD (target: <$10K)

Key Differentiators

Layered Data Architecture

Every design layer is independently versioned, cryptographically signed, and auditable. Change one algorithm, regenerate downstream layers without losing upstream data. Full provenance from raw addresses to construction drawings.

Baremetal Tool Layer

Purpose-built tools for spatial queries, validation, and design analysis. The LLM never processes raw geometry — it calls deterministic tools and reasons about structured summaries. Reliable, fast, and token-efficient.

Domain-Expert Built

Built by fiber designers who know what a midsheath splice is, why bore pits cannot be in driveways, and why T-connections are not permitted. The standards document IS the software, not a reference PDF.

Rust-Ready Architecture

Pure domain logic with no framework coupling. Designed for migration to Rust for production performance. The Python MVP proves the algorithms; the Rust port makes them fast.

Market Opportunity

$42.5B
BEAD funding allocated for broadband
1,500+
New fiber ISPs entering the market
80%
Cannot afford 3-GIS + Esri
Minutes
Time to first design (vs. months)

Delivery Timeline

Week 1 — MVP

  • Day 1-3: Core engine — ingest, planning, validation, DXF output
  • Day 4: Integrity chain with Ed25519 signing
  • Day 5: MCP tools + LLM design review integration
  • Day 6: Full Lumos V6 rule set + BOM workbook
  • Day 7: Resilience (checkpoint, circuit breaker) + testing

Week 2 — Production

  • Day 8-9: Design optimizer + multiple input formats
  • Day 10-11: Additional standards profiles + enhanced DXF
  • Day 12-13: Performance at scale (10K+ addresses)
  • Day 14: Final polish + customer presentation

Ready to automate fiber design?

FiberCAD replaces months of manual CAD work with minutes of automated planning.