How We Built Classic Pool's AR Takeoff App: Three Days of CAD Into Ten Minutes

Classic Pool's AR Takeoff Platform

The Objective

Classic Pool Tile & Stone distributes pool coping and tile to builders across the trade. Quoting a coping job meant manual templating and CAD preparation to produce accurate measurements and a layout, and that work ran one to three days per job. We built native iOS and Android applications that scan the pool surroundings, stitch multiple captures into one continuous view, generate the measurements, and plot the selected coping onto the actual pool using AI and geometric algorithms. Takeoff and layout now take under ten minutes on site.

Service Tags

Augmented Reality Computer Vision Native Mobile Development Microservices Architecture

Industry Tags

Wholesale Distribution Pool and Outdoor Living Building Products

Tech Stack

iOS Native Android Native .NET Azure Functions

The Impact

Under 10 minutes
Pool takeoff and coping layout on site
Manual templating and CAD preparation previously took one to three days per job.
2 native applications
Live in production on both stores
Built native on iOS and Android because AR capture quality depends on direct camera and motion sensor access.
One session
From scan to signed order
Capture, stitch, visualize, adjust, and order happen without leaving the application.
The Client

Classic Pool Distributes Coping and Tile to Pool Builders

Classic Pool Tile & Stone is a wholesale distributor of pool coping and tile, founded in 2005 and headquartered in Spotswood, New Jersey. They supply granite, travertine, and slate coping alongside a full line of swimming pool and spa tile, depth markers, deck tiles, and mosaics.

Their customers are pool builders and contractors. That distinction shapes everything about the problem, because a builder buys differently from a homeowner. A builder cannot sell coping until they have quoted the job, and they cannot quote the job until they know the exact perimeter measurements and can show the homeowner what the material will look like installed.

Which is why Classic Pool came to us. The quoting step had become the bottleneck in their own sales cycle, and every day a builder spent producing measurements and a layout was a day the order sat unplaced.

The Problem

Quoting a Job Required One to Three Days of Templating and CAD

Coping runs the entire perimeter of a pool. It is expensive, permanent, and impossible to evaluate from a sample chip.

For the homeowner, that meant committing to a decade-long installation based on a two-inch piece of stone and a catalogue photograph of somebody else’s pool. Hesitation was the natural response, and on a purchase this size hesitation stalls the sale.

For the builder, the cost was slower and larger. Producing an accurate quote required manual templating on site to capture the perimeter geometry, then CAD preparation to turn those measurements into a coping layout. That work ran one to three days per job. A homeowner who was still deciding could absorb three days of a builder’s effort and then walk away.

Photographing the pool solves neither problem. One photograph captures one angle of a shape that curves, so it yields no usable measurements and no believable visualization. The builder still needs the template, and the homeowner still has to imagine the material following geometry they can see but cannot picture finished.

Our Approach

We Mapped the Full Pool Shape Before Placing Any Coping

Placing coping accurately requires knowing the actual shape of the pool, and no single camera frame captures a curved perimeter. That single constraint set the entire build order.

AR-Guided Capture

We built guided scanning that walks the user around the pool area, capturing the frames the stitching needs so nothing is left to chance.

Image Stitching

Multiple captures resolve into one continuous view of the pool perimeter before any material is placed on it. This is where most augmented reality work in this category fails. Render onto a partial or distorted scan and the customer immediately recognizes the result as fake, which costs more trust than showing them nothing at all.

Measurement Generation

The stitched view produces the perimeter measurements the builder previously captured through manual templating. This is the component that replaces the CAD step outright.

AI and Geometric Coping Placement

Placement combines AI with geometric algorithms, and each does the half it is suited to. The AI handles detection and surface identification across varied pool shapes and lighting conditions. The geometric layer handles how a physical material follows a curved edge, which is a solved mathematical problem and does not need a model to approximate it.

Catalogue and Visualization

The builder selects from the Classic Pool catalogue and the coping renders onto the scanned pool in real time, with adjustment before anything is finalized.

In-App Ordering

Design finalization and ordering happen in the same session, so the quote does not have to travel back to an office before it becomes an order.

What We Built

Two Native AR Apps, Image Stitching, and In-App Ordering

Every component served the builder standing at the edge of a pool with a homeowner waiting for a number.

AR-Guided Pool Area Scanning
Multi-Image Stitching
Continuous Perimeter View
Surface and Edge Detection
Automated Measurement Generation
Coping Catalogue Selection
AI and Geometric Coping Plotting
Real-Time Preview and Adjustment
Design Finalization
Direct In-App Ordering
Native iOS Application
Native Android Application
.NET Backend
Python Vision Microservices
Azure Functions Processing Layer
How It Was Built

Real Pools Behave Differently From Test Renders

We built the capture and stitching pipeline against actual pools. Curved perimeters and reflective water behave differently from flat test geometry, and the difference shows up immediately in the stitched view. A pipeline validated on renders would have shipped and then failed on the first job with an irregular freeform pool and afternoon sun on the water.

That same demand for precision drove the platform decision. AR capture quality depends on direct access to the device camera and motion sensors, and a cross-platform abstraction layer between the app and that hardware costs exactly the accuracy this build could not afford. Measurement accuracy is the product here. A visualization that looks good and measures wrong produces a quote that loses money on the job.

Behind the apps, the processing architecture separated concerns deliberately. Vision and geometric work ran as Python microservices on Azure Functions, kept apart from the .NET backend handling catalogue and commerce. Image processing is bursty and computationally heavy, and ordering is neither, so keeping them separate lets image processing scale on demand without provisioning the commerce layer for peaks it never sees.

Splitting the work between AI and geometry was a deliberate architectural position. Machine learning is good at detection and surface identification across the variation real pools present. Computing how a fixed-width material follows a curved edge is deterministic geometry with a known solution, and reaching for a model there would add inference cost, latency, and a failure mode for no accuracy gain.

Both applications shipped to the App Store and Play Store and are live in production.

Outcome

Takeoff and Coping Layout in Under Ten Minutes on Site

A builder scans the pool on site and has the measurements and a coping layout in under ten minutes, work that previously took one to three days of manual templating and CAD preparation. The homeowner sees the selected coping along their own perimeter, adjusts until it looks right, and the order is placed before anyone leaves.

For Classic Pool, that compresses the gap between a builder walking a job and an order landing. The quoting bottleneck that sat between interest and purchase is now a ten-minute step performed at the pool, where it used to be a multi-day task performed back at an office.

The catalogue can also grow without touching the vision pipeline. New coping options render against the same scanned shape, so expanding the product range stays a content decision.

Testimonials

"The application created by AppVerticals scans the pool area, stitches multiple images into a single view, and generates both the measurements and the coping layout using AI and geometric algorithms. Work that took one to three days of templating and CAD now takes under ten minutes on site, and the customer sees the finish on their own pool before ordering. The app is live on both the App Store and Play Store, and we are grateful for the team's work in getting it there."

Andreas Luethi

Business Owner Classic Pool Tile & Stone

Is your sales cycle waiting on a measurement and quoting step that still runs on manual templating and CAD?

We have replaced multi-day takeoff processes with on-site AR capture, accurate enough to quote from and clear enough to sell from. We know what this takes.

Start the conversation