Details

The role of AI agents in construction project management
ID Brelih, Anja (Author), ID Robnik Šikonja, Marko (Author), ID Klinc, Robert (Author)

.pdfPDF - Presentation file, Download (403,66 KB)
MD5: 1BEFACFD33FCF5D42D7527BA5C449E3D
URLURL - Source URL, Visit https://publications.rwth-aachen.de/record/1014054/files/1014054.pdf This link opens in a new window

Abstract
The integration of Large Language Models (LLMs) into Artificial Intelligence (AI) agents presents new opportunities for enhancing construction project management through improved collaboration, reasoning, and workflow automation. This paper investigates how LLM-powered AI agents can support complex construction tasks such as scheduling, resource allocation, risk assessment, and information coordination. We conduct a structured review of recent literature across construction informatics, AI-driven project management, and agent-based systems to identify trends, capabilities, and limitations. Our findings show that while AI agents offer strong potential to assist project teams with dynamic decision-making and routine task automation, their effectiveness is limited by data quality, domain adaptation, and the stochastic nature of LLMs. These insights are significant for researchers and practitioners aiming to introduce AI agents into real-world construction workflows, where reliability, explainability, and human oversight are essential. The paper highlights the need for further exploration of Human-in-the-Loop (HITL) designs and calls for standardisation of agent capabilities to ensure safe, transparent, and practical deployment in the construction industry.

Language:English
Keywords:AI agents, LLM, project management, construction workflows, construction informatics
Work type:Article
Typology:1.08 - Published Scientific Conference Contribution
Organization:FGG - Faculty of Civil and Geodetic Engineering
Publication status:Published
Publication version:Version of Record
Year:2025
Number of pages:Str. 147-154
PID:20.500.12556/RUL-178027 This link opens in a new window
UDC:004.8:624
DOI:10.18154/RWTH-CONV-254903 This link opens in a new window
COBISS.SI-ID:262894083 This link opens in a new window
Publication date in RUL:16.01.2026
Views:538
Downloads:223
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Record is a part of a monograph

Title:Tagungsband des 36. Forum Bauinformatik 2025 : 24. – 26. September 2025, Aachen
Editors:Baris Özcan
Place of publishing:Aachen
Publisher:Rheinisch-Westfälische Technische Hochschule - RWTH
Year:2025
COBISS.SI-ID:262884355 This link opens in a new window

Licences

License:CC BY 4.0, Creative Commons Attribution 4.0 International
Link:http://creativecommons.org/licenses/by/4.0/
Description:This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.

Secondary language

Language:Slovenian
Abstract:
Vključitev velikih jezikovnih modelov (VJM) v agente umetne inteligence (UI) ponuja nove priložnosti za izboljšanje upravljanja gradbenih projektov z izboljšanim sodelovanjem, razumevanjem in avtomatizacijo delovnih tokov. Ta prispevek preučuje, kako lahko agenti UI, ki jih poganja VJM, podpirajo kompleksne gradbene naloge, kot so načrtovanje, dodeljevanje virov, ocena tveganj in usklajevanje informacij. Izvedli smo strukturiran pregled najnovejše literature s področja gradbene informatike, upravljanja projektov na podlagi UI in sistemov na podlagi agentov, da bi identificirali trende, zmogljivosti in omejitve. Naše ugotovitve kažejo, da agenti UI sicer ponujajo velik potencial za pomoč projektnim delovnim skupinam pri dinamičnem sprejemanju odločitev in avtomatizaciji rutinskih nalog, vendar je njihova učinkovitost omejena zaradi kakovosti podatkov, prilagajanja domen in stohastične narave VJM. Ti vpogledi so pomembni za raziskovalce in st rokovnjake, ki želijo agente UI uvesti v realne gradbene delovne tokove, kjer so zanesljivost, razložljivost in človeški nadzor bistvenega pomena. Članek poudarja potrebo po nadaljnjem raziskovanju zasnov načela človek v zanki (angl. Human-in-the-Loop, HITL) in poziva k standardizaciji zmogljivosti agentov, da se zagotovi varna, pregledna in praktična uporaba v gradbeni industriji.

Keywords:agenti UI, veliki jezikovni modeli, upravljanje projektov, gradbeni delovni tokovi, gradbena informatika

Projects

Funder:ARIS - Slovenian Research and Innovation Agency
Project number:P2-0210
Name:E-Gradbeništvo

Funder:ARIS - Slovenian Research and Innovation Agency
Project number:P6-0411
Name:Jezikovni viri in tehnologije za slovenski jezik

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back