Singapore’s shift toward model-based regulatory submissions is changing when architects, engineers and contractors need to resolve design conflicts, validate project information and assign accountability for automated checks. From October 1, CORENET X is set to become mandatory for all new building projects, increasing the practical importance of interoperable building information models and reliable data across the project lifecycle.
TNGlobal previously reported on Nemetschek Group’s partnerships with the Institute of Technical Education and Singapore Polytechnic to build skills around AI-assisted compliance, model validation, digital twins and related workflows. The transition raises a broader question for firms already preparing for CORENET X: what actually changes in day-to-day project delivery once structured models, automated checks and AI-assisted recommendations become part of the workflow?
In this TNGlobal Q&A, Jonathan Ng, Managing Director, Southeast Asia, at Nemetschek Group, discusses where work shifts upstream, why model quality becomes more important as automation increases, how smaller firms can build a practical digital foundation, and what evidence would show that digital delivery is improving outcomes.
As CORENET X moves from policy and preparation into day-to-day project delivery, which workflow changes do you think firms are most likely to underestimate?
The biggest change is that CORENET X requires project teams to think about coordination and information quality much earlier in the process.
In a document-led workflow, different disciplines can develop their work relatively independently and reconcile information closer to submission. With model-based submission, architects, engineers and other project stakeholders need to coordinate around a structured digital model whose geometry and information must remain consistent across disciplines.
In the past, an architect might design a ceiling layout and an engineer might route mechanical, electrical and plumbing systems independently, only discovering a clash during the final 2D drawing overlay or even on-site. Under CORENET X, if a structural beam intersects an HVAC duct, the model validation process flags this immediately. Teams must now resolve that spatial conflict in the digital environment weeks or months earlier than they used to.
That shifts work upstream. Teams need to agree earlier on modeling standards, responsibilities, information requirements and how changes are managed. The firms that adapt best will treat CORENET X as a change in project delivery rather than simply a change in submission format.
Where can AI-assisted compliance genuinely reduce review time or rework, and where should professional judgment remain firmly human-led?
There is significant potential for AI and automation in repetitive, data-intensive parts of review.
For example, digital tools can help identify incomplete model information, highlight inconsistencies and screen for potential compliance issues. If these checks happen during design rather than toward the end of a submission cycle, teams have an opportunity to address issues before they result in further coordination work or rework.
Consider fire safety compliance. AI can calculate travel distances to the nearest exit or verify that specified fire doors have the correct fire-resistance rating in the model’s metadata. These are tasks that traditionally take days of manual checking. However, assessing whether a specific fire-rated material is appropriate given unique site constraints, supply-chain availability or specific building use remains a qualitative decision that requires human professional judgment.
Regulations often need to be understood in the context of design intent, site conditions, engineering considerations and competing requirements. A useful principle is that AI should compress the time spent searching and checking, while leaving judgment and accountability with the professional.
How does model-based submission change the information handoffs among architects, engineers, contractors and regulators? Where are the biggest interoperability frictions today?
Model-based delivery changes the handoff from exchanging drawings and documents to exchanging structured information about the building itself.
Ideally, information created by an architect or engineer should remain usable as it moves through coordination, regulatory submission, construction and eventually operations. That creates opportunities to reduce duplication, but only if different systems can understand that information consistently.
This is where interoperability becomes important. For example, a structural engineer detailing rebar and concrete in software such as Allplan needs the structural model to align with the architect’s geometric model. If teams are forced into proprietary formats, critical metadata can be stripped out during conversion. OpenBIM standards such as IFC help ensure that when an architect hands over the model, the structural engineer does not have to rebuild the geometry from scratch and the regulator receives a federated model in which the different parts can communicate.
This is particularly important to us at Nemetschek. We have long advocated OPEN BIM and open standards because no major built-environment project operates within a single software environment. Architects, engineers, contractors, owners and regulators need to be able to use the tools best suited to their work while still exchanging reliable project information.
What data or model-quality problems are most likely to undermine automated checks, and how should firms govern those inputs before relying on AI-assisted compliance workflows?
Automated checking makes data quality much more visible. A model may look correct visually while still containing incorrect classifications or duplicate information. These issues become important when a system needs to interpret the model.
That means model quality cannot be treated as the BIM manager’s responsibility at the end of a project. Firms need clear information requirements from the beginning and regular validation throughout the design process.
A classic example is a 3D object that visually looks exactly like a fire-rated door, but in the model’s data structure it is classified simply as a generic slab or lacks property sets defining its fire rating. To a human reviewer looking at a screen, it looks fine. To an automated compliance engine, that door does not exist. Deploying sophisticated AI on poorly classified models can simply automate the generation of errors.
Before firms ask whether an AI system is sophisticated enough, they should ask whether the underlying information is reliable enough.
Do smaller design and construction firms face materially different cost, skills or implementation barriers? What minimum digital foundation should they prioritize first?
Smaller firms can face a different transition because they may not have dedicated BIM, information-management or digital-transformation teams. The same person may be responsible for design, coordination, client management and submission, so changes to workflow can have an immediate impact on capacity.
I would prioritize fundamentals such as the ability to create and exchange reliable BIM models, familiarity with IFC and IFC+SG requirements, basic model validation, disciplined version management and at least one or two people internally who understand the end-to-end digital submission workflow.
For a small practice, the minimum digital foundation is often a cloud-based Common Data Environment or an open collaboration platform. They do not necessarily need to invest in massive local servers or hire dedicated BIM managers immediately.
By standardizing accessible platforms that support IFC viewing and basic model validation, a single architect can effectively manage version control and collaborate with external engineers without becoming overwhelmed by IT infrastructure. Once that foundation is reliable, firms can introduce more automation and AI where it removes repetitive work or reduces errors.
As AI recommendations begin to influence design and compliance decisions, how should firms rethink accountability, audit trails and responsibility for errors?
Firms should design accountability into AI-assisted workflows from the beginning.
A useful audit trail should make it possible to understand what information was assessed, which version of a model was used, what the system identified or recommended, and what decision the professional ultimately made. Where a recommendation is accepted or overridden, there should be clarity around who made that decision.
Clear human-in-the-loop decision gates must be established. If an AI compliance tool suggests altering a load-bearing wall to meet spatial regulations, the software should log the recommendation, but the workflow should require the structural engineer to accept, modify or reject that change with a digital signature.
Trust in AI will ultimately depend on how accurate a system appears to be, as well as whether its output can be interrogated, understood and traced.
Which capabilities are most lacking in the current workforce, and how should ITE, polytechnic and industry training evolve beyond learning individual software tools?
The industry increasingly needs people who understand the workflow behind the software.
There is a growing need for people who can work across disciplines. An architect may not need to perform an engineer’s role, for example, but they do need to understand how the information they create affects downstream coordination, regulatory review, construction and operations.
That is one reason our partnerships with ITE and Singapore Polytechnic focus on applied learning and connected digital workflows. Training should expose learners to realistic project scenarios in which they need to exchange models, identify problems, validate information and work with other disciplines.
Instead of simply testing whether a student knows how to draw a wall in a specific software package, modern assessments should simulate real project friction. We give students a federated model containing deliberate architectural and structural clashes or missing IFC classifications, and their task is to run the validation, identify the missing data and coordinate with a peer to resolve it.
We need to teach the workflow, not only the interface. Software will continue to change. The ability to manage reliable digital information and make informed decisions around it is much more durable.
Over the next 12 to 24 months, what evidence would show that CORENET X and AI-enabled delivery are actually improving outcomes? Which measures would be most meaningful?
Approval time will be important, but I would avoid judging the transition through a single metric.
We should also look at whether project teams are identifying more issues before submission and whether the number of avoidable resubmissions or coordination errors falls.
Further downstream, measures such as requests for information, design-related changes and rework during construction can show whether better digital coordination is translating into better project delivery.
Model quality itself is another useful measure. If a greater proportion of models pass validation with fewer fundamental information issues, that would indicate that digital capability is becoming embedded in everyday practice.
The ultimate proof of CORENET X will be found on the construction site. If high-quality, validated models are flowing smoothly through the approval process, we should see a measurable drop in on-site material waste and emergency procurement. Better digital coordination at the design stage directly enables tighter digital supply-chain integration, allowing contractors to order materials with greater precision rather than relying on estimates and buffers.
For us, Singapore is particularly interesting because regulatory digitalization, industry adoption and workforce development are happening in parallel. If that translates into measurable reductions in errors, rework and coordination effort, there will be lessons that are relevant well beyond Singapore.
Jonathan Ng is Managing Director, Southeast Asia, at Nemetschek Group. Based in Singapore, he is responsible for the group’s business across Southeast Asia. He joined Nemetschek in February 2026 and brings more than 20 years of international experience across the architecture, engineering, construction and operations and manufacturing sectors. Before joining Nemetschek, he spent nine years at Hexagon / Leica Geosystems. He holds a postgraduate degree from the University of Oxford and is pursuing a PhD in Informatics and System Science.
Editor’s note: This Q&A has been lightly edited for clarity and TNGlobal house style. The substance of the interviewee’s responses has been preserved.
Share your perspective: TNGlobal welcomes contributed insights and expert commentary from across Asia’s technology and innovation ecosystem. Submit a contribution for editorial consideration, or explore more conversations in our TNGlobal INSIDER and TNGlobal Q&A and Interviews archive.
Nemetschek signs ITE and Singapore Polytechnic MOUs on AI-ready built-environment skills