Construction AI Glossary

A

Agentic AI

AI systems that monitor data streams and take autonomous action without waiting for human prompts. Unlike chatbots that respond to queries, agentic AI watches, decides, and acts within defined boundaries — completing multi-step workflows such as extracting structured data from site photos, flagging safety issues, and updating progress trackers automatically.

AI Agent

A software system that receives inputs (messages, images, sensor data), processes them using AI models, and takes actions or produces outputs autonomously. In construction, AI agents typically connect to existing communication channels like WhatsApp to extract structured project data from unstructured messages and photos without requiring workers to change their behaviour.

B

BIM (Building Information Modelling)

A digital representation of the physical and functional characteristics of a building or infrastructure asset. BIM models contain geometry, spatial relationships, material properties, and scheduling data. When integrated with AI agents, BIM data can be used to validate progress claims against planned designs and detect deviations automatically.

BTO (Build-To-Order)

Singapore's public housing model where HDB flats are built only after a sufficient number of buyers have committed to purchase. BTO projects are among the largest construction programmes in Singapore, and their scale makes them strong candidates for AI-powered progress tracking and quality management.

C

Computer Vision

A field of AI that enables machines to interpret and extract meaningful information from images and videos. In construction, computer vision models trained on site imagery can detect safety violations (missing PPE, absent guardrails), identify defects (cracks, misalignment), and verify progress by comparing photos against expected work completion.

Construction 4.0

The application of Industry 4.0 principles to the construction sector — integrating IoT sensors, AI, robotics, 3D printing, and digital twins into construction workflows. Construction 4.0 aims to address the industry's historically low productivity growth by automating data capture, improving decision-making with real-time information, and reducing manual administrative tasks.

Construction Rework

Work that must be redone because it was not completed correctly the first time. Research from the Construction Industry Institute puts direct rework costs at 5–20% of total contract value, with indirect costs (schedule delays, disputes, material waste) potentially pushing the figure beyond 30%. Early defect detection through AI agents can catch issues before they compound into costly rework.

D

Daily Progress Report (DPR)

A daily record of construction site activities including work completed, labour counts, materials used, equipment on site, weather conditions, and issues encountered. Traditionally compiled manually by site managers at the end of each day, AI agents can now generate DPRs automatically from WhatsApp messages and photos shared throughout the day.

Defect Management

The systematic process of identifying, documenting, tracking, and resolving construction defects throughout a project's lifecycle. Effective defect management requires structured records with timestamps, photos, severity ratings, and responsible party assignments. AI-powered systems can automate much of this by detecting defects from site photos in real time.

DfMA (Design for Manufacturing and Assembly)

A design approach that optimises building components for factory production and efficient on-site assembly. DfMA reduces construction time, waste, and labour requirements by shifting work from the unpredictable site environment to controlled factory conditions. Singapore's Building and Construction Authority actively promotes DfMA adoption.

Digital Twin

A virtual replica of a physical building or infrastructure asset that updates in real time using sensor data and field inputs. Digital twins enable project teams to monitor performance, simulate scenarios, and detect anomalies without physical inspection. When combined with AI agents, digital twins can be updated automatically from site photos and progress reports.

E

Earned Value Management (EVM)

A project management methodology that integrates scope, schedule, and cost data to assess project performance and forecast completion. EVM relies on accurate progress data — the quality of which directly affects financial reporting. AI-based photo progress tracking provides an independent verification source for earned value claims.

Edge Computing

Processing data near where it is generated (at the "edge" of the network) rather than sending it to a centralised cloud data centre. On construction sites with limited connectivity, edge computing enables AI models to process images and sensor data locally, reducing latency and ensuring critical safety alerts are delivered even when internet connectivity is poor.

I

IoT in Construction

Internet of Things devices deployed on construction sites to monitor environmental conditions (temperature, humidity, dust), track assets and equipment, measure structural loads, and collect real-time data. IoT sensors generate continuous data streams that AI agents can process to detect anomalies, predict maintenance needs, and improve site safety.

L

Large Language Model (LLM)

An AI model trained on vast amounts of text data that can understand, generate, and reason about human language. LLMs power the natural language processing capabilities of construction AI agents — enabling them to parse site messages written in informal shorthand, extract structured data, and generate readable reports from raw project information.

M

Method Statement

A document describing how a specific construction task will be carried out safely and in compliance with regulations. Method statements detail the sequence of work, safety precautions, equipment required, and responsible personnel. In Singapore, method statements are required for high-risk activities and are reviewed by the project's safety team before work begins.

N

Natural Language Processing (NLP)

A branch of AI that enables computers to understand, interpret, and generate human language. Construction-specific NLP must handle domain terminology, abbreviations ("L3 RC 50%" means "Level 3 reinforced concrete 50% complete"), multilingual messages, and the informal communication style typical of site WhatsApp groups.

P

Permit to Work (PTW)

A formal documented system for controlling high-risk construction activities such as hot work, confined space entry, working at height, and excavation. PTW systems ensure that hazards are identified, controls are in place, and authorised personnel have approved the work before it begins. Compliance with PTW requirements is a key regulatory obligation under Singapore's WSH framework.

PPVC (Prefabricated Prefinished Volumetric Construction)

A construction method where three-dimensional building modules — complete with internal finishes, fixtures, and fittings — are manufactured in a factory and transported to site for assembly. PPVC is mandated by Singapore's BCA for selected government land sale sites and can reduce on-site construction time by up to 40%.

Q

QA/QC (Quality Assurance / Quality Control)

Quality Assurance is the systematic process of ensuring construction work meets specified standards through planned procedures and audits. Quality Control is the hands-on inspection and testing of completed work. AI agents enhance QA/QC by providing continuous photo-based monitoring, automated defect detection, and real-time documentation that builds an audit-ready compliance trail.

R

RAG (Retrieval-Augmented Generation)

An AI technique that enhances language model outputs by retrieving relevant information from external knowledge sources before generating a response. In construction AI, RAG enables agents to reference project-specific data — zone names, schedules, trade assignments — when interpreting site messages, ensuring responses are grounded in the actual project context rather than generic knowledge.

S

Safety Management System (SMS)

A structured framework for managing workplace safety risks through policies, procedures, risk assessments, training, and monitoring. In Singapore, contractors are required to implement an SMS that complies with the WSH Act. AI-powered safety monitoring can strengthen an SMS by providing continuous photo-based hazard detection and real-time compliance documentation.

Site Diary

A daily record maintained on construction sites documenting weather conditions, labour on site, equipment usage, visitors, deliveries, and notable events. Site diaries serve as legal records in disputes and are typically required by contract. AI agents can compile site diaries automatically from WhatsApp messages and photos, producing more accurate and timely records than manual entries.

W

Workplace Safety and Health Act (WSH Act)

Singapore's primary legislation governing workplace safety obligations. The WSH Act imposes duties on stakeholders across the construction supply chain — from developers and contractors to workers — and enforces compliance through demerit points, stop-work orders, fines, and criminal liability. Maintaining a clean safety record is a business survival requirement, as accumulated demerit points can restrict foreign worker hiring.

Z

Zero Energy Building (ZEB)

A building that produces as much energy as it consumes over the course of a year, typically through a combination of energy-efficient design and on-site renewable energy systems. Singapore's BCA Green Mark scheme encourages ZEB development, and the country aims to green 80% of buildings by 2030 as part of the Singapore Green Plan.