AI Challenge results
Creative Workshop for the Construction Industry
Future-proof, creative solutions make the construction industry safer and more efficient, and ease the burden on those involved.
From the circular economy to skilled trades to modular construction: these diverse approaches demonstrate how creative solutions can help ensure the future of the construction industry.
Creative Workshop for the Construction Industry
Our event partners
The workshop wasorganized in collaboration with Fraunhofer IOSB, the City of Karlsruhe’s Energy Forum, TALKCONCRETE, Objektkultur, and Cyberforum.
The AI Challenge Theme: “AI in the Construction Industry”
The use of artificial intelligence (AI) methods in the construction industry holds great potential. AI can help make processes more efficient, safer, and more sustainable—from planning, through production and the construction site, all the way to building operations. Existing data can be utilized more effectively, workflows can be digitized, and new business models in the construction industry can be developed.
Through the AI Challenge, we aim to foster a dialogue between users in the construction industry and providers of AI solutions to generate concrete project ideas that can be pursued beyond the event and that will sustainably strengthen the industry’s future viability.
Our topics
In interactive workshops, potential solutions were developed along four key thematic strands (TS)—all of which offer direct benefits to the construction industry:
Topic Thread #1 - Circular Economy in the Construction Industry
The goal of this TS was to use resources and materials more efficiently throughout the life cycle of buildings. The following questions were addressed:
- AI-powered detection and sorting of building materials and secondary construction raw materials, e.g., concrete, reinforcement (steel, fibers, carbon), wood, metals, insulation materials, etc.
- Material Passports and Digital Twins for Buildings (e.g., Building Resource Passport)
- Assessment of reuse potential or recyclability, with a particular focus on reuse over recycling
- AI-Based Optimization of Dismantling and Deconstruction Processes
Session Chair: Dr. Jan Burke (Fraunhofer IOSB)
Ö=C=Ö - Economy and Ecology Optimally Combined with AI Support
DEVELOPMENT
Project objective
Planning Tool/Wizard for Architects for Construction/Design
Resource- and CO2-optimized proposals for building design that meet technical and economic objectives
DEVELOPMENT
AI system and data
AI Agent
- collects and organizes data
- estimates missing attributes in product passports
- minimizes the environmental footprint and the use of critical resources
- develops conceptual proposals with numerical metrics (economic viability, including life cycle assessment)
Value proposition
Resource Conservation/Optimization
Cost reduction, including for building operations
Prudent and Reusable Use of Strategic Resources
Convenient, fast, and accurate planning without any costly surprises
Innovative/disruptive solutions powered by AI
OPERATION
Business model
cost reduction
Open-source project with private and public participation
Internal maintenance and further development of the database
OPERATION
Resources and partners
Resources:
Data on life cycle phases (from manufacturing through demolition and reuse), recycling costs, and resale value
Partners:
- Real Estate Developers, Architects
- Legislator
- Research/University/Colleges
There are still many unknowns in the concept of a circular economy for buildings (materials and their separability, embedded “gray” energy, and the economic viability of reuse or refurbishment). Yet from the perspectives of resource efficiency, carbon footprint, and economic viability, it is absolutely essential to identify the improvements that can be easily achieved in this area. However, existing buildings are often poorly documented, and new buildings are rarely planned with a time horizon that matches their useful life.
This is where the Ö=C=Ö project comes in, which aims to optimally integrate economics and ecology with the help of AI. The first task is to gather information on materials and construction methods in order to capture the energy, material, and life cycle assessments across the entire life cycle of buildings. To this end, we plan to deploy an AI agent that will build a database from all available data sources, standardize the data, and learn to distinguish important data from less meaningful information. The goal is to develop a planning and architectural tool that can provide reliable predictions of operating and recycling costs during the planning phase and effectively support both sustainability and economic efficiency. To achieve this, sustainability aspects—such as avoided environmental damage—must be incorporated into the overall assessment. The availability or foreseeable scarcity of resources must also be factored in.
Since both the general public and several major industrial sectors will benefit from more sustainable planning, the plan is to launch the initiative as an open-source project so that the greatest possible number of data sources and users can be reached. Results and methodologies will be presented at professional conferences and published in academic journals; last but not least, the tool can also provide valuable support in the training of the next generation of professionals.
Topic Thread #2 - Craftsmanship and Execution
The goal of this TS was to investigate the extent to which the skilled trades can be supported by AI-based tools and process optimization. The following questions were addressed:
- Assistance systems for tradespeople (e.g., digital building surveys, image or speech recognition, and automated translation on construction sites)
- Physical assistance provided by (humanoid) robotics and personalized exoskeletons
- Knowledge Management and Preserving Expertise in the Skilled Trades
- AI-powered documentation, work planning, staffing, error detection, occupational safety
- AR/VR Approaches
- Digital Documentation and Real-Time Quality Assurance
Session Chairs: Matthis Leicht and Gerrit Holzbach (Fraunhofer IOSB)
DeliveryScan AI
DEVELOPMENT
Project objective
- AI-powered overview of material deliveries to multiple construction sites.
- Definition of delivery zones on construction sites for suppliers.
- Automatic tracking of which materials have already been delivered.
- Reconciliation of recorded materials with expected deliveries, purchase orders, or bills of materials.
- Providing up-to-date delivery information to construction managers.
DEVELOPMENT
AI system and data
Camera-based detection of materials in the staging area.
Use of mobile cameras, smartphones, or fixed cameras.
Optional additions such as RFID chips, color coding, or other identification features.
Image-based detection of materials, components, pallets, or packaging units.
Reconciliation with delivery data, purchase orders, bills of materials, and, if applicable, CAD/design data.
Interfaces to existing construction, ERP, logistics, or documentation systems.
Value proposition
A clear overview of which materials have already arrived at which construction sites.
Better Management of Multiple Construction Sites Through Centralized Delivery Status Information
Fewer follow-up questions between the construction management team, suppliers, and site personnel.
Early detection of missing, incorrect, or unclear deliveries.
Improved planning of construction processes, follow-up trades, and material usage.
Automatic documentation of incoming shipments in the staging area.
OPERATION
Business model
SaaS solution for construction companies with multiple job sites.
Project-, construction site-, or user-based licensing.
Add-on modules for RFID connectivity, ERP/BIM integration, or a supplier portal.
Introduction via pilot construction sites with designated staging areas
Scaling across multiple construction sites, suppliers, and material groups.
OPERATION
Resources and partners
Resources:
Construction companies with multiple concurrent construction sites.
Construction Managers, Project Management, Purchasing, and Logistics.
Suppliers, freight forwarders, and subcontractors.
Mobile devices or fixed cameras in the deployment area.
RFID infrastructure or color-coding, as needed.
Delivery data, orders, bills of materials, material lists, and, if applicable, CAD/design data.
Providers of ERP, BIM, logistics, or construction site documentation systems.
Construction companies often manage multiple job sites at the same time and need a reliable overview of which materials have already arrived and where. This transparency is often lacking when construction managers are not on site and suppliers deliver materials on their own.
DeliveryScan AI allows you to define a delivery area on the construction site where suppliers can unload materials. There, the materials are recorded using a camera—such as a smartphone, a mobile device, or a fixed-installation solution—and, optionally, via RFID or color coding, and then cross-referenced with delivery data, purchase orders, bills of materials, or planning data.
The results are provided to the construction manager and can be integrated into existing construction project, ERP, logistics, or documentation systems. This provides an up-to-date overview of which materials have been delivered, where they are located, and whether there are any discrepancies.
SiteSense AI
DEVELOPMENT
Project objective
- AI-Powered Condition Monitoring on Construction Sites
- Voice-guided documentation by on-site staff
- Comparison of the actual status with target, planning, and delivery data
- Early Detection of Deviations
DEVELOPMENT
AI system and data
Voice Dialogue / Chatbot
Image Analysis of Construction Site Photos
Structured checklists and data entry forms
Construction plans, delivery dates, target conditions
Optional: CAD/BIM data
Value proposition
Greater Transparency Regarding the Condition of the Building
Relief for Construction Management and Quality Assurance
Fewer On-Site Appointments for Construction Managers
Faster Detection of Deviations
Better Documentation and Traceability
OPERATION
Business model
SaaS Solution for Construction Companies
Project-Based Licensing
User-Based Licensing
Pilot Projects for Implementation
- Add-on modules for inventory management, image analysis, or BIM integration
OPERATION
Resources and partners
Construction companies and construction site personnel
Construction Management and Quality Assurance
Planning, Delivery, and Target Data
Mobile Devices
BIM/ERP software provider
AI and Research Partners
On construction sites, regularly documenting the current status of construction is a critical but time-consuming task. This is particularly challenging when construction managers, inspectors, or technicians cannot be on site themselves and must rely on information provided by employees at the construction site.
SiteSense supports this process through AI-powered condition assessment conducted directly on-site. Employees are guided through the documentation via voice prompts, can capture photos or additional information, and can provide a structured description of the current actual condition. The collected data is then compared with planning, delivery, or target data to document construction progress, deviations, and available materials in a traceable manner.
This results in a digital assistance system that reduces the workload for site management, quality assurance, and documentation, and provides greater transparency regarding the actual conditions on the construction site.
Topic Thread #3 - The Construction Industry and the Construction Process
The goal of this TS was to investigate the extent to which AI-based tools can increase efficiency in construction project management and improve communication. The following questions were addressed:
- AI-Based Analysis and Optimization of Construction Progress Reports and Schedules
- Automatic Detection of Deviations (Time, Cost, Quality)
- Communication and Responsibility Management Among Trades
- Building Information Model (BIM)-Based Data Analysis and Decision Support in Construction Management
- Automated BIM Model Adaptation for Existing Buildings
Session Chair: Reinhard Herzog (Fraunhofer IOSB)
AI Mapper - AI-powered mapping of data models
DEVELOPMENT
Project objective
Consolidating Heterogeneous Domain Models (Planning, Cost Estimation, Inventory)
Avoid Duplicate or Extra Work
Avoid remolding, shorten lead times
A consistent, traceable database through a canonical target model
Ensuring Data Sovereignty and Compliance When Handling Project Data
DEVELOPMENT
AI system and data
Core: Canonical target model + mapping engine (geometry & semantics)
Data sources: BIM/specialized models, point clouds, cost estimation/billing data
AI: Automated standard mappings, AI suggestions in cases of ambiguity, human-in-the-loop
Operation: On-premises/private cloud, encryption, strict role- and permission-based management
Value proposition
Faster Quotes and Invoices
Fewer media breaks and errors, higher data quality
Reusing existing models instead of creating new ones
Transparent, auditable transformations and decisions
OPERATION
Business model
- Target customers: Construction companies, architectural firms, large building owners/contractors
- Services: AI mapping platform + target model setup + consulting/training
- Revenue: One-time project setup + ongoing use (subscription/project basis)
OPERATION
Resources and partners
Internal:
BIM/Technical Experts, Data/AI Specialists, Software Development, IT Operations/Data Protection
External:
Construction software vendors, cloud/hosting partners, pilot customers from the field
(Optional) Universities/Research on AI Methodology and Standardization
The “AI-Mapper” project addresses data silos and duplication of effort in the construction industry. Today, different software solutions, file formats, and modeling guidelines mean that technical models often have to be created multiple times—for example, when a new bidding model is created for iTWO-based cost estimation and billing, even though a design model—or possibly even an as-built model derived from point clouds—already exists. This slows down the bidding process and creates duplication of effort.
This is where the AI Mapper comes in: it defines a canonical target model as a common semantic framework and links domain-specific models to this model via a mapping layer. Geometry and content—such as company-specific components, properties, and classifications—are mapped to this target model automatically or semi-automatically, without forcing or replacing existing workflows and perspectives on architecture, structural systems, or construction. Design and as-built models can thus be used directly for cost estimation, bidding, and billing.
Known, curated mappings run in a deterministic and reproducible manner. In unknown or ambiguous cases, an AI generates mapping suggestions, which are reviewed and approved by subject matter experts as part of a human-in-the-loop process and then fed back into the mapping library as versioned rules. Governance, monitoring, and clearly defined quality criteria ensure transparency, traceability, and data quality. Through data-sovereign operations—such as on-premises or in a controlled private cloud, with encryption and strict access controls—control over sensitive project and company data is maintained, laying the foundation for fast, reliable decisions throughout the entire construction and lifecycle.
Topic Thread #4 - Serial and Modular Construction
The goal of this TS was to investigate the extent to which the industrialization of construction (e.g., serial and modular construction) can be supported by AI-assisted planning and production. The following questions were addressed:
- Optimizing Manufacturing and Assembly Processes with AI
- Sustainability Assessment and Comparison of Alternatives Through Simulation
- Automated Quality Inspection in the Production Process
- AI-Supported Education and Training
- AI-Powered Permitting Processes for Municipalities
Topic Chair: Dr. Christian Kühnert (Fraunhofer IOSB)
AI Elements – Conversion of Architectural Drawings into Prefabricated Building Specifications
DEVELOPMENT
Project objective
- Support
- Automation in the Conversion of Architectural Drawings into Manufacturing Specifications
- Improving Quality and Efficiency Through Automated Validation Checks
DEVELOPMENT
AI system and data
- Computer vision-based plan analysis, combined with a generative AI model for automatic plan translation
- AI System Provides Optimized Manufacturing Data
- Diverse historical design data (CAD, BIM, PDFs, scans)
Value proposition
- Time savings achieved by reducing manual work
- Efficient use of personnel and scalability without a proportional increase in staff
- Minimizing Planning Errors
OPERATION
Business model
- Development of a subscription model and cloud solution
- Option for on-premises deployment and licensing
OPERATION
Resources and partners
- Technicians as domain experts, data scientists, and software developers
- Hosting of AI models and other necessary IT infrastructure
- Annotated cutting plans and CAD/BIM data that contain this ground truth
In the manufacture of prefabricated components for building construction, the process is structured such that a designer first creates the architectural building design and then passes it on to a technician, who uses it to determine how to produce prefabricated components that fit precisely—a complex task in which the technician must analyze the architect’s plan and translate it into system-specific, manufacturable component specifications. Since building plans and the resulting specifications are similar, the question arises as to whether and how AI methods can support the engineer’s work. Specifically, the following AI-based applications are conceivable according to the AI-Elements project concept:
- Plausibility checks to verify that the building design and the layout plan for the building components match
- Automatic conversion of the building design into a layout plan
- Calculation of an optimized cutting plan, e.g., in terms of material efficiency or CO₂ emissions
Become part of our network!
The community management of the KI-Allianz Baden-Württemberg specifically connects business, science and politics in the regions in order to promote the exchange of knowledge and the application of AI technologies. The Community Management of the AI Alliance is also represented in the Karlsruhe region.
If you are interested in a result or an entire topic and would like to find out more or get involved, please contact our Community Management for the Karlsruhe region at
:
Akiza Hagami
akiza.hagami.neb@ki-allianz.de
Impressions and comments from "Smart Sustainable Solutions"
We are all dependent on each other.
The great thing about AI (artificial intelligence) is that I can bring many aspects together.
How can AI help us ask the right questions?
What is special about this workshop format is that the providers do not develop solutions that can subsequently be offered to users, but that users themselves are directly involved in the design.
Our plan worked. The participants were inspired by the kick-off event and there was a lack of time, not a lack of ideas.