Skip to main content
Ari Zilnik
Construction Join Head of Design

An AI recommendation engine that saved a construction customer $500K in 30 days

Spotting the cost savings estimators would miss.

An AI recommendation engine that saved a construction customer $500K in 30 days

Join is project management for the people who build actual buildings. As Head of Design, I lead a small team and ship product across the platform.

Ideas is an AI recommendation engine that surfaces cost-saving material alternatives inside construction estimates. A less experienced estimator might spec a finish at full cost when a substitute would do the same job for half. That knowledge gap compounds across every line item in every project. Ideas closes it automatically.

The original plan was a materials marketplace. Customer interviews killed that direction. Estimators didn’t know what alternatives existed, so they’d never browse a catalog. I took on both design and an acting PM role, pivoted the concept to a recommendation engine, and we shipped in two months.

Suggestion carousels

AI-generated suggestions appear as contextual carousels ranked by cost-saving potential. Each carousel groups alternatives by where they appear in the estimate and by construction phase, so estimators see relevant options without searching.

Join Ideas browsing page with two suggestion carousels: items found in the estimate and items commonly considered at this construction phase

Human review layer

These suggestions affect real buildings and real budgets. The AI will be wrong sometimes, so a materials researcher validates every suggestion before it reaches an estimator. The model is a first-draft engine, not a source of truth.

I used AI-coded prototypes to test card density with real estimate data before engineering built anything. Each card shows UniFormat and MasterFormat classification codes. Estimators told us in interviews these were non-negotiable for trust.

Three Join Ideas suggestion cards showing alternate materials with dollar amounts and construction classification codes

Forced comparison

Every recommendation has to be weighed against alternatives before an estimator can accept it. The model suggests. The estimator decides. Nothing gets auto-applied.

Detail view of a material suggestion with tradeoffs and a cost comparison chart

Results

One enterprise customer saved $500K on their first project estimate within 30 days of launch. Decision-making time dropped 50% across the product.

"Join secures growth investment from construction giants DPR Construction and STO Building Group."
EIN Presswire, 2025

Credits

Design + acting PM: Ari Zilnik

Materials research: Kyle Willis

Engineering: Kevin Rakestraw

Engineering: Nick Zukoski

Shipped at Join, 2024.

Impact

$500K

Construction savings for one enterprise customer in 30 days

50%

Reduction in decision-making time

2 months

From research kickoff to launch