the vibes, embrace exponentials, and forget that the code even exists.” — Andrej Karpathy, February 2, 2025 Te concept is straightforward: in- stead of writing code in a programming language, you describe what you want in plain English—the way you’d explain it to a colleague—and an AI model gen- erates working software. You test it, give feedback conversationally, and the AI refines the result. By the end of 2025, Collins English


Dictionary had named “vibe coding” its Word of the Year. By early 2026, surveys indicated that over 90% of U.S. developers had adopted some form of AI-assisted coding, and the global market for these tools was projected to reach $8.5 billion. But here’s what matters for con-


struction: vibe coding isn’t just for soft- ware developers. It’s for anyone who understands a problem well enough to describe it. And if there’s one thing construction professionals know how to do, it’s describe operational problems with precision.


How It Works (And Why It’s Easier Than It Sounds)


ect-specific automation that no software company can anticipate—because only you know how your company actually works. Tat’s where vibe coding comes in.


What Is Vibe Coding? In February 2025, Andrej Karpathy—a computer scientist, co-founder of OpenAI, and former head of AI at Tesla—posted a now-famous message on X outlining a new way to build software. He called it “vibe coding.” “Tere’s a new kind of coding I call ‘vibe coding,’ where you fully give in to


Reading about vibe coding can make it sound almost too simple. Describe what you want, and AI builds it? In practice, it really is that accessible, but the results depend entirely on how well you describe the problem. Te AI is fast and capable. Your job is to be specific. Te process follows a simple loop.


You start by describing what you need in everyday language—not in technical terms, but the way you’d brief a new project engineer. Something like, “I need a tool that pulls our open change orders from a spreadsheet, groups them by subcontractor, calculates the total pending cost exposure, and highlights anything over $50,000.” Te AI generates


a working version. You try it. You tell it what to fix, saying, “Add a column for days pending,” or “Make the chart a bar chart instead.” It revises. You repeat it until it works. Te platforms that make this possi-


ble range from beginner-friendly app builders such as Lovable, Bolt, and Replit (where you need zero coding experience) to more powerful tools such as Cursor, Claude Code, and GitHub Copilot for those willing to go deeper. The key insight is this: the person building the tool doesn’t need to under- stand JavaScript, Python, or databases. Tey need to understand the problem. Your knowledge of construction work- flows, trade coordination, and project controls is the most valuable input.


Finding the Gaps: Where Vibe Coding Fits Vibe coding is not about replacing your existing platforms. It’s about automating the work that lives in the spaces between them—the manual tasks, the custom reports, the data re-entry that nobody’s solved because it’s too specific to your firm for any vendor to anticipate. Te opportunities exist across every


department in a construction company. Here’s a preview of where the gaps are widest:


Estimating and Preconstruction


Bid day is still a largely manual scramble. Subcontractor proposals arrive in differ- ent formats. Scope comparisons are done in spreadsheets built from scratch each time. Historical bid data sits in old files that nobody has time to analyze. Vibe coding opens the door to tools that auto- mate proposal intake, normalize bid data for comparison, and mine your historical estimates for patterns that improve future pricing. Tis is one of the highest-value areas for automation.


CALIFORNIA CONSTRUCTOR JULY/AUGUST 2026


13


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36