AI Agents for Automated Design and Compliance Verification in Buildings and Infrastructure: A Case Study Approach with No-Code/Low-Code Implementation

Abstract

The buildings and infrastructure sector faces increasing demands for accuracy, efficiency, and regulatory adherence. Manual design calculations and compliance checks are slow and error-prone. This paper presents a case study-driven approach to automating these tasks using AI agents within a no-code/low-code workflow automation framework.

We examine three applications: ASHRAE 62.1 ventilation rate calculations, ADA accessibility compliance for door clearances, and structural foundation sizing.

The methodology utilizes a visual workflow automation platform to extract information from building codes, perform calculations, and validate results with integrated Large Language Model (LLM)-based quality assurance checkpoints. For each case, we document the agent design, integration methodology, quality assurance protocols to manage hallucination risks, and quantifiable outcomes. The implementation of LLM-based QA/QC checkpoints ensures robustness by validating that information extracted from codes is correct and calculations are performed accurately, enhancing trust in AI-driven workflows.

Results from three real-world case studies demonstrate an 80% reduction in task completion time and substantial decrease in calculation errors. Beyond efficiency gains, AI agents enhance resilience by streamlining design processes, minimizing regulatory non-compliance risks, reducing dependency on manual expertise, and adapting dynamically to evolving standards and design requirements.

This approach democratizes digital transformation in buildings and infrastructure, making design calculation and compliance verification tools accessible to both large and small firms without specialized AI expertise. The integration of LLM validation addresses a critical challenge in adopting AI for safety-critical applications. Future directions include expanding AI agent capabilities for more complex design scenarios and integrating real-time data sources for dynamic compliance monitoring.