At the 2025 Bentley Illuminate conference, engineers from VHB unveiled a groundbreaking application of generative AI that's poised to transform how infrastructure professionals interact with complex design software. The demonstration showcased Bentley Copilot, an internal AI agent that learns software applications so engineers can focus on design rather than software mastery. This innovative approach represents a significant shift in how AI integrates with professional workflows, moving beyond simple automation to become an adaptive learning partner.

The Vision Behind Bentley Copilot

Bentley Copilot represents a fundamental rethinking of AI assistance in professional software environments. Unlike traditional help systems or chatbots that provide static responses, this AI tutor actively learns the software's functionality, interface, and workflows. The system uses advanced machine learning algorithms to map software capabilities and user interactions, creating a dynamic knowledge base that evolves as the software updates and user patterns change.

According to industry analysis, infrastructure design software typically requires 6-12 months for engineers to achieve proficiency, with complex applications like Bentley's MicroStation, OpenRoads, and OpenBuildings requiring even longer mastery periods. Bentley Copilot aims to dramatically reduce this learning curve by providing context-aware guidance that adapts to individual user skill levels and project requirements.

How the AI Tutor Technology Works

The core innovation of Bentley Copilot lies in its ability to learn software functionality through multiple data streams. The system analyzes:

  • Software documentation and help files to build a foundational understanding of features
  • User interaction patterns to identify common workflows and pain points
  • Project data and templates to understand context-specific applications
  • Real-time user queries to continuously improve response accuracy

This multi-layered learning approach enables the AI to provide guidance that's not just technically accurate but also practically relevant to the specific design challenge at hand. The system can recognize when users are attempting complex operations and offer step-by-step assistance, suggest alternative approaches, or warn about potential compatibility issues.

Integration with Microsoft Teams and Existing Workflows

One of the most practical aspects of Bentley Copilot is its seamless integration with Microsoft Teams, making the AI assistance available within the collaboration platforms that engineering teams already use daily. This integration strategy reflects Bentley Systems' understanding that AI tools must fit naturally into existing workflows rather than requiring users to adopt new platforms.

The Teams integration allows for:

  • Real-time collaboration where multiple team members can consult the AI tutor simultaneously
  • Shared learning sessions where teams can work through complex design challenges together
  • Knowledge retention through chat history that becomes a searchable resource
  • Cross-platform accessibility from desktop, mobile, and web interfaces

This approach aligns with Microsoft's broader Copilot ecosystem strategy, creating consistency for users who may already be familiar with other Microsoft Copilot implementations.

Impact on Infrastructure Design Efficiency

Early demonstrations and pilot implementations suggest Bentley Copilot could deliver substantial efficiency improvements across multiple dimensions of infrastructure design:

Reduced Learning Time

Traditional software training requires extensive classroom time, self-study, and on-the-job experience. Bentley Copilot provides immediate, context-sensitive guidance that can reduce initial software familiarization time by up to 70%, according to preliminary data from VHB's internal testing.

Accelerated Problem Solving

When engineers encounter unfamiliar scenarios or complex design challenges, they typically spend significant time researching solutions, consulting documentation, or seeking help from colleagues. The AI tutor can provide immediate guidance, potentially cutting problem-solving time from hours to minutes.

Improved Design Quality

By providing real-time validation and best practice suggestions, Bentley Copilot helps maintain design consistency and compliance with standards. The system can flag potential issues before they become costly errors and suggest optimizations that might not be immediately obvious to human designers.

Knowledge Preservation

As experienced engineers retire or move between projects, institutional knowledge often gets lost. Bentley Copilot captures and formalizes this knowledge, making expert-level guidance available to less experienced team members.

Technical Architecture and Implementation

Bentley Copilot builds on several advanced AI technologies working in concert:

Natural Language Processing

Advanced NLP capabilities allow the system to understand complex technical queries expressed in natural language. Users can ask questions like "How do I create a custom cross-section for this roadway while maintaining design standards?" and receive specific, actionable guidance.

Computer Vision Integration

For design software, visual context is crucial. The system incorporates computer vision to understand what users are seeing on screen, enabling more precise guidance based on the current visual state of the design.

Reinforcement Learning

The AI continuously improves its responses based on user feedback and interaction patterns. When users indicate that a suggestion was helpful or unhelpful, the system adjusts its future responses accordingly.

Knowledge Graph Technology

A sophisticated knowledge graph connects software features, design principles, project requirements, and user contexts, enabling the system to provide holistic guidance rather than isolated answers.

Industry Context and Competitive Landscape

Bentley's approach to AI assistance comes at a time when multiple CAD and BIM software providers are exploring similar technologies. However, Bentley's focus on infrastructure-specific applications and deep software integration sets it apart from more generic AI assistance tools.

Competitors like Autodesk have introduced their own AI assistants, but Bentley's strategy of creating a tutor that learns the software represents a different philosophical approach. While some systems focus on automating repetitive tasks, Bentley Copilot emphasizes education and skill development, potentially creating more sustainable long-term benefits for users.

Real-World Applications and Case Studies

VHB's demonstration at Bentley Illuminate provided concrete examples of how the technology works in practice:

Roadway Design Optimization

In one scenario, an engineer working on a complex intersection design used Bentley Copilot to quickly identify the most efficient method for creating transition lanes while maintaining proper superelevation. The AI tutor provided step-by-step guidance through OpenRoads Designer, suggesting specific tools and settings that the engineer hadn't previously used.

Bridge Modeling Assistance

Another demonstration showed how the system could help structural engineers create complex bridge components by understanding the relationship between design parameters and structural requirements. The AI could explain why certain approaches were preferable based on load calculations and material properties.

Utility Coordination

For projects involving multiple underground utilities, Bentley Copilot demonstrated ability to guide users through clash detection and resolution processes, suggesting coordination strategies based on industry best practices and regulatory requirements.

Future Development Roadmap

Based on the Illuminate conference presentation and industry trends, Bentley Copilot is expected to evolve in several key directions:

Expanded Software Coverage

While initial implementations focus on core Bentley applications, the technology is designed to scale across the entire Bentley portfolio, potentially including specialized tools for geotechnical engineering, construction management, and asset operations.

Enhanced Predictive Capabilities

Future versions may incorporate more advanced predictive analytics, anticipating user needs before they're explicitly stated and suggesting optimizations based on analysis of similar completed projects.

Integration with Digital Twins

As digital twin technology becomes more central to infrastructure management, Bentley Copilot could evolve to provide guidance not just during design but throughout the asset lifecycle, from construction through operations and maintenance.

Customization and Specialization

The system may eventually support organization-specific customization, allowing companies to train the AI on their proprietary standards, templates, and workflows.

Challenges and Considerations

Despite the promising demonstrations, several challenges remain for widespread adoption:

Data Privacy and Security

Engineering firms handle sensitive project data, and any AI system must maintain rigorous security standards. Bentley will need to demonstrate robust data protection measures, particularly for cloud-based AI services.

Accuracy and Reliability

In engineering contexts, incorrect guidance could have serious consequences. The system must achieve extremely high accuracy rates before engineers will trust it for critical design decisions.

Integration Complexity

While the Teams integration simplifies access, deeper integration with Bentley's desktop applications will be necessary for the most seamless user experience.

Cost and Accessibility

Pricing models and accessibility for smaller firms will be important factors in determining how broadly the technology gets adopted across the infrastructure industry.

The Broader Implications for AI in Professional Software

Bentley Copilot represents a significant milestone in the evolution of AI assistance beyond consumer applications into specialized professional domains. The approach demonstrates several important principles that may influence how other professional software developers approach AI integration:

Contextual Understanding Matters

Generic AI assistance often falls short in professional contexts where domain-specific knowledge is essential. Bentley's focus on infrastructure-specific understanding shows the importance of specialized training.

Education Over Automation

While automation has its place, there's significant value in systems that enhance human capability rather than replacing it entirely. The tutor approach acknowledges that professional expertise involves judgment and creativity that pure automation can't replicate.

Seamless Integration is Critical

The success of professional AI tools depends heavily on how well they integrate into existing workflows. Bentley's Teams integration strategy recognizes that new tools must adapt to users, not the other way around.

Conclusion: A New Paradigm for Software Learning

Bentley Copilot represents more than just another AI feature—it signals a fundamental shift in how professionals interact with complex software. By creating an AI that learns the software rather than expecting users to master every feature, Bentley is addressing one of the most persistent challenges in professional software: the gap between software capability and user proficiency.

The technology demonstrated at Bentley Illuminate 2025 suggests a future where engineers spend less time learning software and more time solving engineering problems. As the system evolves and matures, it could fundamentally change how infrastructure professionals develop their skills, collaborate on projects, and leverage technology to create better designs more efficiently.

While widespread adoption will depend on addressing practical concerns around accuracy, security, and cost, the underlying approach—using AI to bridge the gap between software complexity and human capability—represents an important direction for the entire professional software industry. As other developers observe Bentley's progress with Copilot, we can expect to see similar approaches emerging across multiple professional domains where software mastery has traditionally been a barrier to productivity and innovation.