Where we’re going, we don’t need roads

In our previous post  (the problem and the framework), we've explored the current state of Digital Twins in AECO, the importance of a use-case driven approach, and introduced our 10 Dimensions framework for defining specific needs. Now, let's look ahead to the future and the technical foundation required to achieve the full potential of Digital Twins.

The AECO industry is striving to improve outcomes on many fronts – time, cost, quality, health & safety, and environmental impact – and digital twin solutions are fast evolving to meet those needs. They are, however, yet to achieve their full potential.

To drive the industry forward strategically we need to take a step back and look at the future more broadly. Over the next five years we can expect significant social and technological change. We have identified five key trends:

1. Pressure to drive ever greater efficiencies in critical infrastructure, optimising at portfolio and state-wide levels, and going beyond net zero.
2. Increased automation of construction, and infrastructure will become hyperconnected with greater real-time monitoring and proactive ‘smart’ asset management. This means fewer people on site and a massively increased volume of data.
3. A greater mandate for having continuity of data between project phases, and for data-sharing between projects on the same site or in close proximity.
4. AI will be everywhere, potentially less in-your-face, but orders of magnitude more capable. Desk jobs will become less administrative and more decision-making.
5. The AECO market will consolidate into a few one-stop-shops or platforms with rich connectivity.

There is no one piece software that can achieve all the expectations of a perfectly comprehensive, “live” digital twin. Instead we should be thinking about a system of systems working together to provide the necessary data, insights, and operational support.

Some vendors are choosing to build a walled-garden suite of tools. In contrast, Sensat’s vision is an open platform with a robust core of end-to-end functionality and rich connectivity to the wider geospatial and AECO ecosystem. This will enable a comprehensive, intelligent, real-time twin to be created far sooner than otherwise possible.

The Digital Twin Architecture

A foundational step for the industry is collaborating using a common language – we need a structured way to break down the technology innovations required for digital twins to achieve their full potential. To that end we have defined a high level architecture comprised of 5 core components:

1. Data Firehose (ingest)
2. Knowledge (model)
3. Intelligence (analytics & prediction)
4. Human Interfaces
5. Machine Interfaces

The architecture serves as a valuable framework for the industry, enabling us to identify current technical limitations, highlight opportunities for innovation and development, and ultimately drive the evolution of more powerful and impactful digital twin technologies. By examining any digital twin solution through the lens of these five architectural components we can effectively determine its level of maturity, and suitability for addressing a set of specific business needs and use cases.

Let’s dive into the components:

1. Data Firehose (ingest) – the interface to the physical world
   This component is responsible for capturing and transforming the full set of real-world information. This includes geospatial data from sensors, traditional and modern remote sensing methods, and surveys; operational data from control and asset management systems; design and engineering data like BIM models and drawings; and non-geospatial documentation. This diverse data, both structured and unstructured, originates from numerous devices and intermediary systems. Key challenges for this component involve ensuring connectivity, managing data volume and throughput, maintaining data quality, and conforming various formats to make the information usable within the digital twin. Without the comprehensive data foundation the rest of the solution is moot.

2. Knowledge (model) – making the ingested data accessible and actionable
   The knowledge model's purpose is to create an accessible representation of the real world, enabling effective understanding, analysis, and reporting. This component enables federation across diverse sources. It covers how data is catalogued and securely stored for use by other components, leveraging technologies like geospatial databases, vector embeddings, and knowledge graphs to represent entities and their relationships. This includes file-oriented attributes and embedded metadata, ensuring industry standards such as ISO19650 and NBIMS-US can be followed.

3. Intelligence (analytics & prediction) – extracting insights and optimising outcomes
   This component encompasses the systems required to extract insights from the knowledge model using techniques ranging from straightforward calculations and descriptive statistics to advanced predictive modelling and physical simulations. It encompasses diagnostic analytics to understand causes and prescriptive analytics to recommend optimal actions. Machine learning techniques and multimodal AI models, such as semantic segmentation and VLMs (vision-language models), operate in this layer to enrich the knowledge model and provide advanced capability to end-users.

4. Human Interfaces – extracting value through human-oriented interactions
   The Human Interfaces translate the digital twin's data and insights into tangible value. This includes rich visualisations like 3D models, gaussian splats, and AR/VR to provide intuitive and immersive environments; collaborative spaces fostering multi-stakeholder engagement; workflow engines embedding structured processes; customisable reporting outputs tailored to different roles; alerts and notifications for critical events; and robust search and query capabilities for easy information retrieval. Effective design of these interfaces is crucial for user adoption and informed decision-making.

5. Machine Interfaces – enabling connectivity with other systems
   The Machine Interfaces component enables the digital twin to connect and interact with external systems through APIs and standard protocols such as A2A (Agent-to-Agent), MCP (Model Context Protocol). This allows for data dissemination to other enterprise platforms, leveraging bespoke external data processing services, and providing hooks for autonomous AI agents. Crucially, it facilitates integration with building automation and industrial control systems for potential closed-loop optimisation. By adhering to interoperability standards, this component transforms the digital twin into a connected ecosystem, extending its value and reach.

Reshaping the Built Environment

At Sensat, we use this framework not only to diagnose gaps but to architect the next stage in the evolution of our digital twin platform. We have pioneered a truly accessible, seamless, cloud-based geospatial environment for working on large scale critical infrastructure. Our platform has effortlessly ingested, federated, and visualised 100s of terabytes of survey, 3D design, GIS, and programme data – facilitating millions of collaborative interactions across thousands of construction professionals.

Today, Sensat’s platform leads the industry in three foundational components of digital twins; Data Firehose, Knowledge and Human Interfaces. Our highly optimised geospatial web renderer gives users across the globe instant access to high quality visuals. The recent launch of Workspaces is transforming the human interface component by pushing the boundaries of collaboration and interface management, whilst simultaneously raising the bar on the technical and UI/UX design aspects of the application. Under the hood we have rebuilt our knowledge engine, introduced a highly flexible security model, and laid the foundations for robust spatio-temporal data orchestration. We have also expanded our APIs, enabling a machine interface to automate data ingest and complex workflows.

While Sensat’s visualisation platform offers powerful geospatial inspection tools for analyses, historically our deep spatial data science offerings have been confined to professional services. That's changing. We are now embedding this intelligence directly into the platform, putting this power into the hands of end-users – building out capabilities and automation within the intelligence component. Early this year, we outlined the potential of vector embeddings and are thrilled to deliver on that vision with project Orion – the first visual search engine for the physical-world.

Looking further forward we recognise that while co-pilots and agentic AI are proving to be very effective at specific micro tasks, they are predominantly still decision support or human-in-the-loop tools. The technology will continue to improve but humans will still be involved, shifting our hands-on activities to higher level tasks, direction, and decision-making. To this end, the future is about evolving machine automation and human interfaces in tandem – a direction that Sensat is uniquely positioned to drive as a leader in visualisation, collaboration, and geospatial intelligence.

‍

Want to be kept updated about Sensat's geospatial data innovations? Sign up here to receive the latest news.