Martinez Commercial Builders is a family-owned general contractor based in San Antonio, Texas, with $65M in annual revenue. Founded in 1987, the firm specializes in K-12 education, municipal buildings, and healthcare clinic construction across central and south Texas. When third-generation owner Carlos Martinez took over operations in 2023, he identified preconstruction modernization as his top priority.
"My grandfather started this company with a pencil, a calculator, and a green columnar pad," says Carlos. "My father upgraded to Lotus 1-2-3 and then Excel. When I looked at our estimating process in 2023, we were still fundamentally doing the same thing as my father, just on a bigger monitor. We were competitive because of our relationships and reputation, but I could see that wasn't going to be enough going forward."
Martinez's estimating team consisted of Carlos himself and two estimators with a combined 45 years of experience. They handled approximately 30 estimates per year using Excel templates developed over two decades. The templates were powerful but fragile, with complex inter-sheet references and macros that only one of the three estimators fully understood.
The transition to ElkConstruct began in March 2025 with a parallel estimating approach. The team selected a $12M elementary school project and built the estimate using both their traditional Excel method and ElkConstruct. Carlos expected the AI platform to produce a rough approximation that would need significant manual correction. The results surprised him.
"The AI-generated estimate was within 4% of our manual estimate on the first try," Carlos recalls. "After we calibrated it with our local cost data, the variance dropped to under 2%. But what really sold me was the takeoff accuracy on the architectural scope. The system counted 847 doors across the floor plans. Our manual count was 844. We rechecked and found we had missed three doors in a corridor between the gym and the cafeteria."
Over the following four months, Martinez loaded historical data from 22 completed school projects into ElkConstruct. The AI models quickly adapted to the specific cost patterns of K-12 construction in the Texas market. By July, the team was producing estimates 50% faster than their previous process.
The impact on bid volume was immediate. In the first six months, Martinez submitted 24 estimates compared to 15 in the same period the previous year. Their win rate improved from 22% to 31%, which Carlos attributes to having more time to analyze each bid strategically rather than rushing through the numbers.
"The biggest surprise was how much it changed the way I spend my time," says senior estimator Roberto Garza, who has been with Martinez for 28 years. "I used to spend most of my day counting and measuring. Now I spend that time actually thinking about the project, looking for risks, finding value engineering opportunities, and making sure our pricing strategy is right. It's a better use of my experience."
The financial results have been substantial. Martinez estimates the platform saves them approximately 1,200 estimator hours annually, equivalent to more than half of one full-time position. Combined with the additional projects won through increased bidding capacity, the ROI exceeded 400% in the first year.