1:31:40
Industry Partners Series: making confident decisions with connected data
Engineers Australia
Overview
This video discusses the critical challenges of managing and connecting data in major infrastructure projects. It highlights how disconnected data across various platforms leads to inefficiencies, risks, and poor decision-making. The panel emphasizes the need for a connected data environment, improved information governance, and a shift in organizational mindset beyond simply adopting new software. The discussion also touches upon the future of AI in infrastructure and the foundational importance of clean, connected data for its successful implementation.
How was this?
Save this permanently with flashcards, quizzes, and AI chat
Chapters
- Major infrastructure projects suffer from a lack of connected project data due to multiple, disconnected proprietary systems (e.g., ProjectWise, Aconex).
- This disconnection hinders project coordination and makes it difficult to gain a unified view of project status.
- There's a tendency to believe more technology or AI will solve data foundation problems, but this is only true if underlying data management practices and schemas are sound.
Understanding this core challenge is crucial because it directly impacts project efficiency, cost, and the ability to make timely, informed decisions.
Projects like Sydney Metro and Suburban Rail Loop involve numerous contractors and JV systems that do not communicate, creating significant coordination issues.
- Infrastructure projects generate vast amounts of data that need to be managed, shared, and reviewed.
- Data is often siloed in proprietary platforms, preventing a holistic project view and hindering AI's ability to reason across different systems.
- Heterogeneous data sources make it difficult for AI to understand and integrate information, leading to unreliable outputs.
As projects become more data-intensive, connecting this data is essential for effective project management, supply chain collaboration, and leveraging advanced analytics like AI.
AI struggles to process data from multiple, disparate systems (like ProjectWise, Aconex, Forma) because it lacks semantic understanding across these different formats and platforms.
- Organizations struggle to establish a single source of truth (SSOT) due to vendor lock-in and a tool-by-tool technology acquisition approach.
- Lack of interoperability between platforms erodes trust when data conflicts arise.
- Semantic differences in how platforms handle data (e.g., naming conventions for documents) make true data aggregation and unification nearly impossible with current technology alone.
Without a trusted single source of truth, conflicting data leads to errors, rework, and a lack of confidence in project information.
One platform might call a document 'drawing number' while another calls it 'document description,' making it difficult for systems to semantically understand and reconcile the same piece of information.
- It is very common for major projects to operate across multiple Common Data Environments (CDEs) and platforms.
- The challenge isn't having multiple platforms, but the lack of clarity on which system owns what information and how data flows between them.
- Ensuring people are working with the current, verified version of information is critical for project success.
This widespread reality necessitates strategies for managing data flow and verification across diverse systems to maintain project integrity.
Even with platforms like Autodesk's Forma or Bentley's suite, which have native integrations, the problem arises when trying to connect these to other systems or when ownership and data flow are unclear.
- Manual data transfer between systems leads to version disparity, increasing the risk of using incorrect information.
- Using wrong information can cascade into significant financial and legal risks due to rework and disputes.
- Metadata, crucial for verification, is often lost or incorrectly transferred, hindering the ability to track design decisions and their implications.
Manual processes introduce human error and create a chain of risks that can severely impact project timelines, budgets, and legal standing.
Constructing based on the wrong version of a design document can lead to costly variations, legal disputes, and difficulties in proving the information's lifecycle and approval status.
- The consequences are felt across the entire supply chain, from on-ground coordinators to executive leadership.
- Ground-level staff spend excessive time manually re-uploading and re-attributing metadata, leading to significant labor costs.
- At the executive level, disconnected data results in a lack of project performance visibility, inability to learn from past projects, and financial hits from rework and inefficiencies.
Recognizing who bears the brunt of disconnected data helps prioritize solutions and understand the true cost of poor information management.
Project managers may lack the latest tunnel or station alignments, leading to rework, while executives face financial losses due to poor visibility and inability to make data-driven future decisions.
- Organizations should aim to 'bring data home' by aggregating data from all vendor platforms into their own controlled environment.
- This is crucial for insurance, cyber resilience, and long-term data ownership, not just for immediate project decision-making.
- A true SSOT is about enabling informed decisions and ensuring the future longevity and security of project information.
Shifting the focus from a single physical location to data aggregation and ownership provides greater control, security, and resilience.
If a company like Carillion had its own aggregated data copy, it could have recovered more quickly from a ransomware attack that crippled their systems, rather than being down for months.
- Standardizing everyone into a single platform is often unrealistic due to existing investments, contractual obligations, and security requirements.
- The better objective is to standardize the *approach* to using platforms and define clear information requirements.
- The focus should shift from 'Common Data Environment' to 'Connected Data Environments' that link existing systems.
Recognizing the impracticality of a single platform allows for more achievable strategies focused on connectivity and standardized information requirements.
Instead of forcing all users onto one platform, organizations can standardize their information requirements and use technology to connect different CDEs, enabling data flow between them.
- A successful strategy enables a connected data environment (CDE) without creating technical debt, using a 'one data language' approach.
- It involves connecting, not replacing, existing authoring tools and CDEs, respecting native integrations.
- Robust governance, consistent information architecture, naming conventions, and versioning are critical, along with a clear plan for data ownership and operational use (e.g., for AI and asset maintenance).
This approach ensures data consistency, usability, and trust across disparate systems, laying the groundwork for effective AI and long-term asset management.
Utopia's CD Sync aims to enable clients to connect their CDEs, creating a unified data language and consistent information architecture without requiring them to become developers or abandon existing tools.
- Establish roles like Enterprise Architecture or Information Architecture, which are becoming essential project functions.
- Organizations must fundamentally rethink their approach to data, moving beyond standardizing authoring tools to embracing semantic design and data stitching.
- Stop 'throwing technology' at problems; focus on establishing frameworks, robust governance, clear information requirements, and the right people before implementing technology.
These foundational steps address the root causes of poor information quality and accessibility, ensuring technology investments yield tangible benefits.
Implementing strong governance and information requirements first, then layering technology on top, ensures that software is utilized effectively to achieve desired outcomes, rather than becoming a costly, underutilized asset.
- Complex projects like Rail Baltica highlight challenges in connecting data across multiple countries, contractors, and sovereign regions.
- A significant challenge is the fragmentation between parties regarding what information is needed and at what level of detail (e.g., BIM models vs. 2D PDFs for site teams).
- Commercial contracts often water down requirements for information exchange, leading to vague terms and disconnects in how projects are built and validated.
Learning from these real-world examples helps identify common pitfalls and refine strategies for better project delivery and information management.
On a rail project, outdated tunnel alignments were used by contractors for months because data wasn't aggregated, leading to station placements clashing with the actual tunnel path, a conflict caught only through data aggregation.
- The biggest misconception is that digital transformation is achieved by simply buying and deploying new software.
- True transformation involves changes in people's behaviors, processes, roles, and standards, not just using new tools.
- Digital transformation is an ongoing capability requiring governance and measurement, not a one-off technology project.
Clarifying these misconceptions is vital for organizations to invest in the right areas and achieve genuine digital capability rather than superficial changes.
Simply installing new software without changing how people create, approve, or exchange information means the organization is just reproducing old processes in a new system, not transforming.
- Treating information management as a technology project rather than a data and people problem is a major mistake.
- Organizations often fall into 'tech joy,' buying technology before defining requirements, governance, or how data will be applied.
- A critical error is 'messy data in, messy data out'; AI cannot fix poor data quality, and a lack of clear ownership across IT, legal, risk, and delivery teams exacerbates the problem.
Avoiding these common mistakes ensures that efforts to improve information management are focused, effective, and address the core issues of data quality and human processes.
Buying AI tools without first ensuring clean, connected data means the AI will operate on flawed information, leading to incorrect outputs and wasted investment.
- Aggregating data from multiple CDEs into a central data lake (e.g., Azure blob storage) allows for analysis of complex commercial claims.
- Connecting systems enables retrieval-augmented generation (RAG) architectures for AI to query and analyze project data for decision-making.
- Early detection of critical clashes, like station alignments not matching tunnel paths, is possible when data from different sources is aggregated and analyzed.
These practical examples demonstrate how connected data directly translates into better decision-making, risk mitigation, and cost savings.
On a rail project, aggregating data revealed that several contractors were using outdated tunnel alignments, which would have caused significant clashes with station placements, a problem caught early through data connection.
- AI is only as good as the data it draws upon; fragmented data leads to unreliable outputs and partial narratives.
- Before pursuing AI, organizations must connect their CDE systems to provide AI with a complete, semantically understood dataset.
- Simply adopting AI tools like Copilot without connected data means they can only perform basic tasks, not leverage comprehensive project information for complex decision-making.
Establishing a connected data foundation is a prerequisite for AI to deliver meaningful insights and drive true transformation in infrastructure projects.
If Copilot only accesses a small portion of project data, it can help write emails but cannot grapple with complex decision frameworks derived from all project CDEs, leading to limited utility.
- AI operating on fragmented data provides partial, potentially incorrect answers with high confidence, leading users astray.
- This lack of context and semantic understanding can result in 'hallucinations' and flawed decision-making.
- A significant risk is that users often don't critically check AI outputs, leading to the propagation of errors and downstream liabilities.
Understanding these risks is crucial to prevent AI from becoming a source of misinformation and costly errors, especially in high-stakes infrastructure projects.
An AI might confidently provide an incorrect calculation or design recommendation based on incomplete data, and a user, trusting the AI's confidence, might act on it without verification, leading to significant downstream problems.
Key takeaways
- The primary information challenge in major infrastructure projects is the lack of connectivity between disparate data systems.
- Connected data is essential for enabling effective project coordination, risk management, and leveraging advanced technologies like AI.
- Achieving a 'single source of truth' requires aggregating data into a controlled environment, not necessarily consolidating into one platform.
- Manual data transfer is a significant source of errors, leading to version disparity, rework, financial losses, and legal risks.
- Digital transformation is about changing people and processes, not just implementing new software; technology is an enabler, not the transformation itself.
- Organizations must prioritize data quality, governance, and clear information requirements before investing in new technology or AI solutions.
- AI's effectiveness is entirely dependent on the quality and completeness of the underlying data; a connected data foundation is a prerequisite for reliable AI outputs.
Key terms
Connected Data Environment (CDE)Single Source of Truth (SSOT)Digital TwinSemantic Artificial IntelligenceInformation GovernanceMetadataInteroperabilityVendor Lock-inRetrieval Augmented Generation (RAG)Common Data Environment (CDE)
Test your understanding
- What is the fundamental information challenge facing major infrastructure projects today, and why does it persist?
- How does the lack of connected data impact the ability of organizations to leverage AI for decision-making?
- What are the key risks associated with manually transferring information between different project systems?
- What realistic goals should organizations set when aiming for a 'single source of truth' in their data management strategy?
- Why is establishing robust information governance and clear requirements more critical than simply adopting new technology for digital transformation?