Data Science Talent Logo
Call Now

Digital Twins: The First Fruit of the Metaverse, Powered by Generative AI By Aakash Shirodkar

Aakash Shirodkar is Senior Director of AI & Analytics at Cognizant. Aakash has over 20 years of experience helping businesses achieve global success through the application of AI. Aakash’s previous roles include Data Science Portfolio Leader at IBM and Senior Manager at Idea Cellular.
In this post, Aakash discusses the ways in which the metaverse is set to transform business. Digital twins are integral to this transformation, allowing businesses to flourish in the virtual space.

Aakash demystifies the concept of the digital twin, and explains how you can harness its power to revolutionise your business in the metaverse:

From Invisible Struggles to Empowering Discoveries:

When you hear the word ‘Metaverse’, what images pop into your head?
Likely a 3D world, an immersive video game, or virtual reality. But the Metaverse isn’t confined to science fiction or gaming anymore; it’s steadily infiltrating the business world through ‘digital twins.’ These progenies of the Metaverse are revolutionising enterprises by optimising design, processes, systems, and assets. They’re enabling businesses to lessen environmental impact, enhance customer experiences, and streamline operational costs.
Allow me to demystify the concept of digital twins and explain how you can get started with a digital twin for your enterprise.

Hello, and welcome to this fascinating tale of the enterprise metaverse.


The term ‘digital twins’ originated at NASA in 2010 as an effort to improve the simulation of physical models of spacecraft. John Vickers, who worked at NASA at the time, coined the term. Digital twins are a far cry from their early days and have evolved beyond static, 2D replicas.

Today, they’re dynamic 3D digital clones that can learn, adapt, and predict. They mimic their physical counterparts so accurately that they are, in essence, creating a bridge between our world and the digital realm of the Metaverse.
A common misconception is that a digital twin is nothing more than a glorified CAD 3D model, a simple simulation model, a common data environment, or an eye-catching telemetry visualisation. The reality of digital twins, on the other hand, is far more complex and infinitely more exciting.

In essence, a digital twin is a symphony of data, models, and real-time information. It’s a digital entity that breathes and evolves, fed by numerous data sources and dynamically processing live data. It’s not a mere snapshot frozen in time but a living, evolving replica of its physical counterpart.


Enterprises have invested in IoT by installing sensors in the real world, and data is now being collected from thousands of devices. They can now use IoT data to sense physical events in real time.

As you install an increasing number of IoT sensors, it becomes necessary to have context about their location and the structures to which they belong.

If we take a smart skyscraper as an example, and your goal is to monitor its occupancy, simply installing a thousand sensors in the building is insufficient. Instead, a virtual model of the skyscraper with rooms, elevators, lobbies, and other relevant areas must be created, and each sensor must be placed in the appropriate context.

In this sense, a digital twin is a digital replica of a physical object that provides context for all IoT devices reporting on it. In our skyscraper example, using the sensors contextualised by the digital replica, we can determine that there are six occupants in room sixteen on the sixty-sixth floor.


Generative AI (a term you may not be familiar with, ha) is an umbrella term used to describe any type of AI that can be used to create new text, images, video, audio, code, or synthetic data. Generative AI models use deep learning to identify patterns and structures within existing data in order to generate new or desired content.

In enterprises, generative AI can perform a variety of tasks such as classifying, editing, summarising, answering questions, and creating new content.

There are many foundational models, also known as Large Language Models (LLMs), that can serve as a starting point. These LLMs can be used as a base for AI systems capable of performing multiple tasks. However, these LLM models require fine-tuning and learning through human feedback. Once they have undergone this process, they can contextually understand input at an enterprise level, enabling them to perform multiple tasks and reducing hallucinatory outputs.

The evolution of generative AI is accelerating. With proper guardrails in place, it can be used to automate, augment, and accelerate work. Each of these actions has the potential to add value by altering how work is performed at the activity level across enterprise business functions and workflows.

Generative AI is now poised to play a pivotal role in both the creation and operation of digital twins.

How, you ask?

Read on to find out.

Therefore, digital twins are defined as:

A dynamic virtual mirror representation of a physical asset, process, system, or environment that looks and behaves exactly like its physical counterpart.

A digital twin, by ingesting data and replicatin processes, enables real-time simulation, analysis, and prediction of the performance, outcomes, and issues that a real-world environment may encounter.

These aren’t static models; they evolve and learn from data to accurately similate the behaviour of their physical counterpars over time.


An everyday simplistic example of a digital twin is Google Maps. It is a virtual representation of the Earth’s surface that uses real-time data on traffic to optimise driving routes and share other relevant information.

However, it is the world of Formula 1 that offers an unadulterated glimpse into the pinnacle of digital twin maturity, providing an emblematic illustration of the remarkable strides we have made in bridging the chasm between the real and the virtual.

To fully comprehend these technologies’ transformative potential, let’s delve into their application in the fast-paced, high-stakes domain of Formula 1 racing.

In Formula 1, every millisecond counts. From car design to simulators, efficiency to analysis, and real-time decision making, digital twins have transformed the world of Formula 1. They are so critical to the success of Formula 1 teams that they are used in every facet of the Formula 1 value chain.

Formula 1 teams and drivers use digital twins to produce high-performing cars, as simulators for drivers practice, to optimise efficiency, a tool for scenario planning as well as real-time decision making in high-pressure racing conditions and lastly for post-race analysis to better calibrate feedback and understanding of the data coming from the car.




In the realm of Formula 1, we now understand the profound impact of digital twins. However, there are challenges that arise when attempting to implement this cutting-edge technology.

Formula 1 teams also collect external data non-car related data to optimise the car’s performance.


While the advantages of implementing digital twins are evident, it’s crucial to recognise the challenges.

  1. Executive endorsement: To ensure cooperation at all levels across the enterprise, which is essential to facilitate the adoption of digital twins.
  2. Upfront investment: The creation of a digital twin requires people, tools, technology, and processes. Although the digital twin can be developed in stages, an initial budget is necessary to get the project started.
  3. Digital maturity: A robust data infrastructure and access to high-quality data that is ingested onto a data platform are the cornerstones of a digital twin. Fundamentally, a higher level of digital maturity will make implementation easier.
  4. Tools and technology: Choosing the right technology is crucial to avoid limiting the enterprise to a single, inflexible technology solution with limited integration capabilities. To build a long-term strategy, a combination of different tools and technologies that offer a high degree of flexibility in integration and scalability is essential.
  5. Talent: To fully benefit from digital twins, you need more than just the right tools and technology; you also need a team of skilled resources, including data engineers, ML engineers, 3D modellers, and data scientists.
  6. Compute: In complex digital twin scenarios, compute power is necessary to speed up the inference process. For instance, creating a twin of a manufacturing facility would require significantly more computing power than twinning a single asset within that facility. Firming up your cloud strategy and evaluating cloud partners is therefore key.
  7. Security: To prevent unauthorised access that could potentially cause damage and disruption, enterprises must take a proactive approach to securing their digital twins. This includes implementing a robust role-based authentication, authorisation, access, and management policy as part of basic hygiene practices. It is also important to encrypt data and APIs end-to-end.
  8. Change management: Adoption is an important consideration when starting with digital twins. If the digital twin is not adopted within the enterprise workflow, the investment will not see any ROI, and the enterprise will not benefit in any way. Including a strong change management process to accelerate adoption and provide a feedback loop back into the digital twin is crucial.
  9. Learning curve: Successfully integrating a digital twin into the operations of a company requires not only learning how to operate it but also understanding how it will impact the organisation’s workflow and processes. To prepare for potential setbacks during the implementation process, it is recommended to allocate enough time for training and familiarisation. Therefore, enterprises should not underestimate the amount of time and effort required for the successful integration of a digital twin into their operations.
  10. Partnering with the experts: While the digital twin can be considered intellectual property, which justifies allocating internal resources for its build and operation, the process can be time-consuming and tedious. Additionally, learning while building can carry risks. Choosing not to seek assistance may not be the best approach and could be counterintuitive. An ideal approach is a hybrid one where enterprises use a core of internal resources supplemented by a partner.
  11. Ethical considerations: Data privacy and security are critical issues that require careful consideration, especially in scenarios involving healthcare, personal information, and other sensitive data. However, it’s also important to evaluate other aspects, such as data and model bias. Additionally, ethical considerations should be made regarding the identified use cases, as they should maintain confidentiality and sensitivity. As a result, a comprehensive enterprise digital twin policy and proper guardrails are necessary to prevent misuse.
  12. Regulatory and Legal Considerations: It is important to comply with relevant laws and regulations when dealing with data and data storage, especially in digital twin scenarios with societal applications. This applies to both input and output data.

Enterprises must address several questions, including: Who owns the data generated by a digital twin? Who is responsible if a digital twin’s predictive model fails, causing harm or financial loss? Is the twin connected to any licensed or proprietary systems? Are there any contractual obligations?

When embarking on the transformative journey of implementing a digital twin, these are some of the challenges that enterprises need to take into consideration.


To overcome these challenges, enterprises should start small by a) identifying a suitable use case and b) defining the goal they want to achieve.

To put it simply, enterprises should ask themselves, what are we trying to accomplish? Are we aiming to improve efficiency or reduce something? If we are reducing, what specifically are we trying to decrease?

Complex or dynamic environments that can greatly benefit from real-time optimisation have emerged as prime candidates for the implementation of digital twin use cases.

Once you have defined your use case and objective, it is important to have a clear business case tied to value. This is critical for maintaining your focus on achieving tangible results.

Then understand your starting position, such as your current digital maturity. Evaluate your strengths and weaknesses to determine if you’re ready to adopt digital twins based on your objectives or if you need to build up certain data, infrastructure, skillsets, cyber security, or other policy and ethical protocols.

Then focus on building a Minimum Viable Product (MVP). This pragmatic approach allows for learning and expansion because you can gradually add features to your digital twin setup rather than building the twin in one fell swoop, which could prove to be a minefield.

Learn from your MVP, set improvement goals and let it grow organically.


STEP 1: Identify what you want to achieve (use case and objective).

STEP 2: Tie down a clear business case to the objective.

STEP 3: Understand your current state of digital maturity.

STEP 4: Take an incremental MVP approach and model the environment with the help of data.

STEP 5: Gain operational awareness.

STEP 6: Establish a maturity arc in which your digital twin capabilities are incrementally improved to meet your enterprise objectives.

The implementation of digital twins should fit seamlessly into your broader digital strategy. It’s important to consider how this technology can be used to outpace competitors and secure a unique market position. For example, if a digital twin can be used to improve product quality faster than your competitors, it should be prioritised.

As enterprises continue to adopt digital twins, understanding the different stages of maturity is crucial to developing a successful strategy that enables them to unlock the full potential of this technology.

In conclusion, integrating digital twins and Generative AI into your Enterprise’s strategy can unlock a wealth of opportunities, helping you navigate the digital landscape with agility and innovation.


A convergence is taking place right before our eyes, as the capabilities of generative AI make it a prime candidate for integration into an enterprise digital twin strategy.

The potential for generative AI to democratise design, data, information, and insight is remarkable. It can accomplish this in two ways:

1. During its “Build Time,” it has the potential to democratise the creation of the digital twin.

2. During its “Run Time,” it can democratise the digital twin’s data, insights, and operations.

Creation of a digital twin, the “Build Time”:
Imagine the savings that can be realised during the creative stage by exploiting the multi-modal capabilities of LLMs to develop real-time 3D digital designs, interactive experiences, and environments. As LLMs improve, these digital models are likely to get richer and more immersive, and designers might do this by merely describing what they want to build rather than painstakingly creating everything from the bottom up. Generative AI serves as a co-pilot, augmenting designers, saving enterprises time, optimising costs, and increasing productivity.

This is resulting in an increase in the number of digital twins created before new initiatives are launched. Before a shovel ever touches the ground, the digital twins of a facility or infrastructure are developed. The digital twin of Vancouver Airport had been developed before construction began, and simulations were utilised to complete the final design aspects.

Operation of a digital twin- the “Run Time”:
At the operations stage, the twin can be queried for insights, goal-seeking objectives, or to augment, allowing the human to do the job more quickly and effectively.

Imagine a digital twin connected to real-time data, resulting in a digital 3D engine with the ability to emulate and simulate. This allows enterprises to simulate future time periods under specific conditions based on real-time data. The twin can be queried using generative AI. For example, imagine being able to ask the twin, “What does this data tell us, and where should we take action?”. This type of information flow throughout the enterprise enables everyone to make faster and more informed decisions.

Generative AI has the potential to transform the digital twin value chain. By automating, augmenting, and

accelerating from the creation stage to the operational stage, generative AI can unlock new levels of creativity and problem-solving.

Essentially, Generative AI has the potential to make English the lingua franca through which humans engage with digital twins, reducing most barriers to adoption.


For products , digital twins allow virtual simulation of the manufacturing process, identifying potential flaws in design prior to production. Real-time analysis and adjustments thereby enhance the product’s quality while accelerating its entry into the market.

Similarly, Service Twins serve as valuable tools for examining design functionality and implementing real-time redesigns where necessary. This approach yields a higher-quality product, meets customer needs more effectively, and provides an enterprise with a competitive edge.

Meanwhile, Customer Twins have revolutionised customer engagement by providing fully immersive product interactions, contributing to significant revenue boosts. A notable illustration of this can be seen in the automotive sector, where virtual test drives have amplified sales volumes.

In the pursuit of sustainability, digital twins aid in reducing material use and route optimisation, thus mitigating environmental impact. This technology has resulted in significant cost reductions and has aided the circular economy across all industries.

The reach of digital twins and Generative AI extends far beyond a single industry.

In manufacturing , they allow real-time monitoring and management of processes, and these insights can then be used to improve production efficiency and shop floor performance. Through predictive maintenance, it is possible to reduce asset downtime.

Healthcare applications include enhancing operational efficiency of healthcare operations, which is the foundation for offering personalised care, more precise treatment plans, and disease management, thereby substantially improving the patient experience.

Digital twins of human bodies or organ systems are being developed, with revolutionary implications for medical education and training.

In supply chain and logistics, a virtual model of the entire supply chain or logistics network can be developed to anticipate performance and optimise routes and resource allocation, lowering costs. Another application could be optimising warehouse design for better operational effectiveness.

In construction, a digital twin can help construction firms with building design elements, enabling better planning by modelling human footfall effects, light, wind, and other aspects. Digital twins for constructed infrastructure can comprehend how a facility is doing in real-time, allowing them to stay ahead of the curve and manage potential events before they occur.

In retail, people twins are used for customer modelling and simulations, enabling retailers to create customer personas to improve the experience they deliver.

Aerospace, automotive, education etc… I could go on, but I am sure you get the picture. Digital twins have gained widespread use across industries.


Digital Twins are growing in capability, performance, and ease of use. As technology continues to advance, the future of digital twins is bright, with endless possibilities for new applications and capabilities.

One trend that we may anticipate seeing as a result of substantial developments in process mining and process capture is how enterprises will create simulations for entire business functions or clusters of business processes rather than a single business process. To achieve superior outcomes, leaders will explore ways to incorporate multiple technologies into digital twins, such as machine learning, process mining, risk analysis, and compliance monitoring.

As a result, more efficient methods of connecting data across these organisations will be pioneered, connecting digital twins and digital threads across various enterprise activities. The glTF file format is gaining popularity for exchanging 3D models across tools.

Another area where digital twins may evolve is in the realm of virtual reality. Enterprises have started using the confluence of these technologies to augment their engineers with holographic lenses or other similar devices so that they can interact with faulty assets in a more realistic and intuitive way when they are in the process of fixing them.

Digital twins will increasingly transform how we run companies. This necessitates a high level of specialisation, and no one provider provides an end-toend digital twin solution. To produce the best fit-forpurpose solution for their organisation, enterprises will need to integrate numerous capabilities. Part of that strategy will necessitate more modular, open architectures as well as the flexibility to create an ecosystem-based system.

For enterprises, the digital twin is more of a mindset than a tool. They are a formidable instrument with the ability to transform how we build, run, and maintain complex physical assets, processes, systems, or environments.

We’ve seen how digital twins can improve the performance of everything from skyscrapers to Formula 1 cars, industrial assets, airports, and even cities.

Lastly, as we navigate through the dawn of the Metaverse era, digital twins and Generative AI are poised to be vital players in bridging our physical and digital realities. The enterprise metaverse will surpass the gaming metaverse in the coming years, and digital twins will be the first fruits of this metaverse.

“Digital Twins are growing in capability, performance, and ease of use. As technology continues to advance, the future of digital twins is bright, with endless possibilities for new applications and capabilities.”

Back to blogs
Share this:
© Data Science Talent Ltd, 2024. All Rights Reserved.