Built the AI sales engine that lets contractors design, render, and sell a $10,000 lighting job — all from their phone, standing in a muddy backyard, in under 20 minutes.

Landscape lighting contractors close five-figure jobs standing in dark backyards at dusk. The homeowner is right next to them. The sale is one "yes" away.

But when Matt (founder of XOutdoorLighting in California) came to me, his contractors were losing that moment. Why?

Because they had to juggle four separate apps to close a single job. CompanyCam for photos. A separate rendering tool. A different app for BOM. And then a video editor for the "wow" moment. By the time everything was assembled and sent to the homeowner, the emotional momentum was gone. Deals died in editing timelines.

Matt's vision was radical: What if the contractor never had to leave the driveway? Design, render, and pitch — all from one phone, right there on the property.

That's what we built in Milestone 4.

By the time M4 kicked off, I had already delivered three milestones — the offline field logging engine, the metadata foundation, and the AI-ready data bridge. Nechtr was a solid field tool. But it wasn't the "sales weapon" Matt kept describing.

M4 was the transformation.

Before M4, adding a fixture meant tapping through a catalog, scrolling brands, selecting types, then typing wattage. Fine at a desk. Terrible at 8 PM in someone's backyard with gloves on.

We built Capture & Speak — inspired by a workflow Matt loved in CompanyCam. Contractor snaps a photo, immediately speaks natural language like "Metro on the oak tree, three of them," and the app does the rest. On-device speech recognition catches the words. A custom nickname engine resolves "Metro" to the exact Metropolitan SKU. GPT-4o-mini structures it into fixture data. Done.

Zero API cost for speech. Works fully offline. Three taps became one voice command.

The whole design philosophy came from Matt: "It shouldn't feel like they're drawing — it should feel like they're ringing up a lighting design." We rebuilt the entire fixture picker as a POS-style terminal. Visual tiles. Brand filter pills. No nested menus, no scrolling catalogs. Optimized for daylight visibility and dirty gloves.

There's a competitor called Dusk Pro. Their AI generates gorgeous nighttime renderings — but the founder himself admitted they hallucinate. Fixtures appear that weren't specified. Beam spreads are wrong. Homeowners get excited, then contractors deliver something that doesn't match.

Matt didn't want gorgeous. He wanted accurate.

We architected the rendering pipeline around Nechtr's existing BOM metadata. Every AI prompt now includes the exact fixture brand, wattage, and placement. The AI isn't guessing — it's rendering the actual design the contractor built. We wired in two models (Flux Kontext Pro and Nano Banana via Replicate) so contractors can choose, and so the future token economy has a foundation to sit on.

The rendered image comes back with high-contrast optimization — critical because contractors are showing homeowners the result on a phone screen in daylight.

The magic moment: the contractor takes a two-minute setup video once. HeyGen builds them a personal Digital Twin avatar. From then on, every project auto-generates a cinematic 1080p walkthrough video where the contractor's own avatar narrates the design against the rendered nighttime scene.

The homeowner receives a personal video message from their contractor showing exactly what their yard will look like. No editing timeline. No CapCut. No jumping between apps.

A guy in Austin was charging $250 for these videos manually. With Nechtr, contractors generate them for around $10 in API costs — in seconds, on-site.

Most of Matt's target contractors already live in the CompanyCam + Jobber ecosystem. Rather than force them to abandon what works, we integrated in.

Jobber creates a job. Jobber pushes to CompanyCam (existing integration). CompanyCam webhooks into Nechtr. The whole thing chains automatically. Photos taken in CompanyCam appear in the correct Nechtr project instantly — no double work, no re-uploading.

Three apps, one integration.

Matt's business isn't a solo operation — it's a team with admins, sales reps, and field installers. So we architected M4 with a full multi-tenant model from the ground up.

Three roles: Admin (billing, users, catalog), Designer (AI generation, BOMs, sales), Technician (field capture only). Cost-bearing endpoints — AI renders, video generation — are gated behind permission checks. If a Technician tries to burn API credits, a push notification fires to the Admin for approval.

We also built the Company Catalog Filter — Admins toggle on the brands their company actually stocks, and every contractor's fixture search only pulls from that curated list. No sifting through 10,000 irrelevant fixtures.

Everything runs on AWS EC2 with a Node.js/Express server, Nginx reverse proxy, and SSL via Certbot. All AI API keys (Replicate, HeyGen, OpenAI) live server-side — the mobile app never touches them. Firebase handles auth, database, and storage. TestFlight for delivery.

Replaced 3 separate apps in the contractor's workflow

Cut proposal-to-video time from hours to under 5 minutes

Videos generate for $10 vs. $250 contractors were paying third-party editors

Point-and-Speak eliminated the drag-and-drop paradigm entirely

Cost-safe multi-tenant architecture ready for scale

Flutter · Node.js · AWS EC2 · Firebase · Replicate (Flux Kontext Pro + Nano Banana) · HeyGen · OpenAI GPT-4o-mini · CompanyCam OAuth + Webhooks · Codemagic · TestFlight

6 weeks · Milestone 4 of an ongoing engagement (now transitioning to a full-time retainer)

M4 turned Nechtr into a sales engine. M5 turns it into a marketplace — subscription tiers, a portfolio system where contractors showcase completed jobs, a web app, and the "Pollin" token economy where BOM purchases generate credits that fuel more AI renderings.

The vision Matt described from day one wasn't a lighting app. It was the operating system for the outdoor lighting industry. M4 is where we crossed the line from tool to platform.

"The competition is trying to make contractors better graphic designers. We're going to make them better contractors by removing the design work entirely." — Matt, Founder, XOutdoorLighting