LuxePlatforms
Rather just talk?(816) 477-6802hello@luxeplatforms.comKansas City · Building nationwide
Vertical SaaS · Construction01 / 16

An AI-native operations platform that replaces Procore + Innergy for custom millwork shops

A custom millwork shop's entire back-office, absorbed into one screen.

A universal SwiftUI app plus a hundreds-of-tools AI backend that runs an entire commercial millwork operation from one screen: estimating, submittals, scheduling, finance. Built to show how much of a millwork back-office a single AI window can absorb.

In production · backend live on Fly.io

A look at the interface.

Representative UI — abstracted, never a client's real data.
🔒app.workspace.dev/overview
Studio
Overview
Analytics
Projects
Inbox3
Reports
Team
Settings
Storage68%
Workspace/Overview
Search…
Overview
Last 30 days · updated 2m ago
Last 30 daysNew report
Revenue
$128.4k
12.5%
Active users
8,642
4.1%
Conversion
3.84%
0.6%
Avg. session
4m 12s
8.2%
Performance
RevenueTarget
JanFebMarAprMayJun
Recent activity
Atlas migration
$24,800
Active
Onboarding flow
$11,250
Review
Billing revamp
$38,400
Active
Mobile parity
$9,600
Done
Dashboard / ops console · representative UI
All-in-one
Replaces Procore + Innergy
Live
In production today
AI
Turns a PDF into an estimate
Role
Solo, design, iOS, backend, AI
Year
2026
Status
In production · backend live on Fly.io
Category
Vertical SaaS · Construction
Built with
  • SwiftUI
  • ARKit
  • Apple Intelligence
  • watchOS
  • Python
  • MCP
  • Fly.io
(01) The Challenge

A custom millwork shop juggles estimating, submittal tracking, shop scheduling, and AR/retainage chasing, all across manual office labor and two separate worlds: construction (Procore) and shop ops (Innergy). The goal was to absorb roughly half that office work into a single AI window, with real shop-floor-to-quote data flow, not toy demos.

(02) The Approach

How it
works

The moat is a closed loop. Shop-floor time-clock actuals, tagged by scope, flow back into the estimating engine and recalibrate labor defaults the moment a punch lands. No competing construction time clock does that automatically. On-device Apple Intelligence drafts and summarizes; a server-side vision model turns PDFs into estimates; embedding search ranks similar past jobs. A hardened decoder layer absorbs recurring server quirks so new backend tools light up in the app with zero per-view code.

What I built

(03) Inside the build
01

Closed-loop quoting engine

Shop-floor time-clock actuals, tagged by scope, auto-calibrate the labor defaults in future quotes the instant a punch lands. No competing construction time clock closes that feedback loop automatically.

02

26-tile Office hub

Native tiles span Procore parity (RFIs, daily logs, AIA billing) and Innergy parity (estimate builder, shop-schedule Gantt, inventory), then finance: QuickBooks OAuth, AP inbox, payroll export.

03

On-device + server AI

Apple Intelligence handles draft replies and voice-memo summaries on device. A server-side vision model turns PDFs into estimates and analyzes RFP bids, and embedding search ranks similar past projects.

(04) Highlights

The
receipts

  • 113 Swift files across three shipping targets: app, widgets, and watch
  • Live-Activity punch tracking in the Dynamic Island
  • ARKit install overlay anchors shop drawings to detected walls
  • Verified auto-calibration: real time-clock actuals flow straight into a source-labeled quote
← Previous
Deterministic Code-Generation Agent
Next →
Social Fitness Meetup App