Winner of the openBIM Awards 2025, Technology Category

Qonic Intelligence (QI) AI-Powered openBIM Classification

Q: Won't IFC exports be too inconsistent for AI?

QI's multi-modal architecture handles real-world data quality issues by processing geometry, metadata, spatial context, and hierarchical relationships simultaneously.

Q: What happens as IFC standards evolve?

QI features dynamic class expansion—automatically adapting to new IFC classes without forgetting prior knowledge. The system grows with your needs.

Location: Belgium | Status: Live in Production

Company: Qonic NV

Download the case study

Key results:

98% classification accuracy
100+ IFC classes supported
50-80% reduction in model prep time
8M+ objects analyzed in training
<25MB model size, client-side processing

openBIM Solutions: IFC 2x3, IFC 4, IFC 4.3, BCF, bSDD

Timeline: 1 year development

The Challenge

The IFC Classification Problem

Most project teams struggle with a fundamental openBIM challenge: how do you consistently classify building elements across 100+ IFC classes when most team members aren't BIM experts?

The consequences are significant:

Time Drain

Manual classification consumes 50-80% of model preparation time. Contractors routinely reclassify entire models between design stages.

Quality Issues

Misclassified elements create incomplete clash detection sets, inaccurate quantity take-offs, and coordination failures that propagate into construction.

Knowledge Barrier

The IFC schema's richness becomes its limitation—only technical specialists can leverage its full potential, forcing teams toward oversimplified models.

Cost Impact

Classification errors lead to rework, material waste, disputes, and lost trust in digital workflows.

"We often see that the IFC schema is not fully understood by designers and contractors. This is key to raising IFC maturity across our industry."

Louis Casteleyn

Brussels Airport Company

Why traditional methods fail:

Manual expert classification doesn't scale.
Simple heuristic rules (identifying only beams, columns, walls) miss 90% of the IFC schema's value.

Why openBIM?

The Strategic Foundation

Qonic built QI on openBIM standards for three compelling reasons:

Universal Applicability

IFC is vendor-neutral and globally recognized. QI works with exports from Revit, ArchiCAD, Rhino, SketchUp, Blender, and even point cloud/mesh data—no tool lock-in.

AI Enablement

Standardization is fundamental to machine learning success. IFC's consistent structure enabled training on 8+ million real-world elements. Without this standard, reliable pattern detection would be impossible.

Long-Term Value

Properly classified IFC models remain usable beyond handover for facility management, renovations, and lifecycle analysis—maximizing digital investment.

Won't IFC exports be too inconsistent for AI?

What happens as IFC standards evolve?

Solution in Action

How QI Works

Purpose-Built AI Architecture User Experience

Purpose-Built AI Architecture

Unlike generic computer vision models, QI was designed specifically for BIM data:

1. Spherical Scanning Technology — Captures complete 3D geometric signatures from all directions, ensuring accuracy even with partial or simplified geometry

2. Multi-Modal Analysis — Processes five data types simultaneously:

Geometric features (shape, volume, proportions)
Textual metadata (names, types, properties);Spatial relationships (adjacency, positioning)
Hierarchical context (parent-child relationships)
Contextual intelligence (surrounding elements)

3. Self-Growing Neural Network — Automatically expands to recognize new IFC classes and building system types as they emerge

4. Continuous Learning — User corrections feed back into training, improving accuracy with every project

User Experience

Simple Workflow:

Upload IFC file or native format (Revit, Rhino, SketchUp)
QI analyzes at 100+ objects/second
Review top 5 classification suggestions with confidence scores
High-confidence predictions auto-apply; uncertain cases prompt validation
Corrections improve future classifications automatically

Real-Time Intelligence:

Visual feedback shows classification reasoning
Confidence scores indicate reliability
One-click corrections for user control
Client-side processing ensures data privacy

Metrics

Time Savings

Faster model preparation for coordination and quantity take-offs

up to

0%

Process Efficiency

Vendor-agnostic: Works with 5+ authoring tools plus mesh geometry
Scalable: Validated on federated models across all disciplines
Energy-efficient: Client-side processing (<25MB) vs. cloud AI

Quality Improvements

98% accuracy even for ambiguous geometric shapes
Measurable reduction in BCF coordination issues
Complete clash detection sets through automated validation

Cost Reduction

Eliminated repeat classification across disciplines
Reduced construction rework from early error detection
More reliable cost estimates from consistent quantity data

Download the case study

Sustainability Impact

Lower carbon footprint from reduced rework and material waste
Energy efficiency through compact, client-side models
Extended data lifecycle with properly structured IFC models

Key takeaways

What Made This Work

Standardization enables AI — IFC's consistent structure was essential for training reliable models
Purpose-built beats generic — Domain-specific architecture outperforms adapted solutions
Real-world training matters — 8M+ elements from actual projects ensured robustness
Continuous learning prevents obsolescence — Dynamic architecture stays current as technology evolves
User feedback accelerates improvement — Corrections create a virtuous cycle

Critical Success Factors

Build on open standards from day one
Invest in domain expertise, not just data science
Design for edge cases and imperfect data
Make AI decisions transparent to build trust
Create feedback loops for continuous improvement

Lessons for Your Organization

For Technology Developers:

Start with openBIM standards—proprietary approaches limit scale and lock users into specific ecosystems

For Project Teams:

Adopt classification validation early, before errors propagate downstream into coordination and construction

For Industry Leaders:

Recognize that AI doesn't replace expertise—it democratizes technical knowledge across entire project teams

Stakeholder Voices

Owner Perspective Contractor Perspective Design Team Perspective

Owner Perspective

"At Brussels Airport, we adopt an IFC-first approach to ensure data consistency and predictability, enabling our asset management systems to fully leverage the information. Yet, we often see that the IFC schema is not fully understood by designers and contractors. Qonic's classification tool assists designers with accurate IFC mappings and significantly improves the IFC quality of existing models. This advancement is key to raising IFC maturity across our industry."

Louis Casteleyn

Building Information Management, Brussels Airport Company

Contractor Perspective

"At Cordeel, QI allows us to augment existing models during the tender phase with higher utility value in line with our stated BIM requirements. The algorithm makes strong suggestions for IFC entities—saving significant time."

Ben Arents

BIM & IT Innovation Manager, Cordeel Group

Design Team Perspective

"The Qonic classification tool enables users to correctly classify IFC models in a simple way; even without having in-depth knowledge of the IFC schema themselves. This combined with Qonic's other capabilities makes it an excellent tool for creating structured, truly useful models."

Ir. arch. Luis Briones Alonso

Head of BIM, Arcade

About This Case Study

buildingSMART International publishes case studies to demonstrate the real-world value of openBIM standards and services including IFC, IDS, BCF, and bSDD. We showcase diverse implementations to inspire innovation and guide adoption.

Featured in this case study: Qonic NV (Belgium)

buildingSMART International does not endorse specific commercial products. Company information is provided for educational context. Organizations interested in similar approaches should evaluate multiple solutions and consult with openBIM professionals.

Browse All Case Studies

Become a Member

About buildingSMART International