As the energy sector evolves, organizations are turning to digital twins, AI and next-generation data models to drive operational excellence and innovation. In this Energy Transition Talks conversation, CGI experts Diane Gutiw, Lukas Krappmann and Peter Warren discuss how digital triplets extend the value of digital twins, helping companies unlock deeper insights, accelerate AI integration and optimize energy systems for a smarter future.

Digital twins and digital triplets are helping companies optimize operations, improve asset management and fast-track AI adoption by building on systems they already have in place. These innovations are quickly becoming essential tools for navigating a more complex, data-driven energy landscape.

Expanding the value of digital twins

Digital twins create living models of assets like infrastructure, equipment or systems by integrating operational and historical data. They improve decision-making and predictive maintenance by giving a full, real-time view of operations.

While already powerful, organizations prioritizing digital transformation are now looking to make this data even more accessible and actionable. That next step is digital triplets.

“The real value of the digital triplet is the ability to access narrative data, images and videos without the heavy lift of traditional modeling” — Diane Gutiw

Introducing digital triplets for enhanced insights

Digital triplets extend digital twins by incorporating generative AI, allowing operators to ask questions and interact naturally with their data. Instead of static dashboards or coding, users engage with systems through intuitive conversations.

At the center of triplets is agentic AI—collaborative AI agents that monitor, analyze and respond to operational data. These agents autonomously identify issues, analyze patterns and generate insights, making complex data ecosystems easier to navigate.

Real-world applications: hydrogen electrolyzer optimization

In Germany’s hydrogen electrolyzer sector, for example, organizations have digitized physical systems and telemetry data into interactive 3D models. Operators can quickly spot anomalies and ask simple questions to diagnose issues or suggest maintenance steps.

This approach improves internal operations while delivering customers a richer digital experience.

“It’s not just about selling a product anymore. It’s about creating a digital service experience for customers” — Lukas Krappmann

Accelerating AI adoption with digital triplets

Digital triplets offer a smart starting point for organizations launching AI initiatives. Rather than building from scratch, organizations extend their current digital twin investments, layering AI to surface faster insights and drive better decisions.

By making AI integration practical and low-disruption, digital triplets accelerate returns and improve business outcomes.

Optimizing energy systems with specialized AI models

Another key innovation is the rise of smaller, specialized language models. Instead of general large-scale models, organizations are training AI specifically on operational datasets like IoT telemetry.

These targeted models are faster, more precise and better tuned for real-world industrial challenges. In the energy sector, they enable quick, actionable insights into grid behavior, equipment performance and operational risks.

A new frontier for energy innovation

Digital twins and triplets are reshaping the energy landscape, from hydrogen electrolyzers to grid optimization and beyond. By making operational data intuitive to access and use, companies can move faster, operate smarter and deliver greater value.

In a complex and rapidly shifting industry, dynamic digital models will be critical to innovation, resilience and sustainable growth.

Listen to other podcasts in this series to learn more about the energy transition

Read the transcript

1. Introduction: From manufacturing to energy markets

Peter Warren:
Hello everyone and welcome back to our ongoing series of conversations about energy transition and how things are changing in industry. We just came back from the Hanover Messe, which is the Hanover Fair for manufacturing, and there's a lot of overlap between manufacturing and the energy markets. We're going to touch on that, but the big dive today is talking about a concept called digital twins and digital triplets. So, with me I have two great experts, Diane and Lukas, and let's start with Diane. Do you want to introduce yourself?

Diane Gutiw:
Thanks, Peter, and thanks for inviting me to the podcast. My name is Diane Gutu. I lead our AI Global Research Center, and a lot of our focus has been on digital twins and extending them to digital triplets. So great to be joining the conversation.

Lukas Krappman:
Yeah, thanks, thanks, Pete. My name is Lukas Krappman, from Germany here and I'm responsible for one of the clients active in the manufacturing and energy and utilities industry. We have already worked on a couple of concepts regarding digital twins and also some ideas and triplets. Thanks, Pete, happy to also be here today.

Peter Warren:
That's a great thing. Thanks for joining me. So, we're covered from the far coast of Canada over to Germany and middle part of Canada. Thanks very much. Diane, since you're the resident expert in all things digital twins and you've come up with the concept of digital triplets, do you want to give us sort of a baseline conversation on what those are?

2. Defining digital twins and digital triplets: Beyond traditional models

Diane Gutiw:
Sure, absolutely. The concept of digital triplet is actually quite simple, but it's a fantastic way to extend an existing investment in data and a data ecosystem. If we look at the different layers, you have your physical assets or a group of assets infrastructure that you're monitoring as your physical layer. The digital twin is that digital representation of those things—people, equipment and infrastructure. It's collecting data from operational systems, historic systems, and edge computing or IoT devices, offering a holistic view of the ecosystem or item being monitored.

Digital twins are not new. They offer a way of looking at operations and interactions between assets and doing AI-based analysis. Digital triplets extend that by leveraging newer technology like generative AI, large language models, and sometimes small language models. Digital triplets allow an operator to converse with the data in natural language. It's a form of agentic AI—a group of collaborative agents monitoring the digital twin data layer, continuously listening, and working autonomously on defined tasks like diagnostics or anomaly detection.

3. Value of digital triplets: Accessibility and new insights

Diane Gutiw:
The real value of the digital triplet to me is twofold.
First, it's the ability to access narrative data, images, videos, and not just discrete operational data without heavy modeling.
Second, it's the accessibility: you can have natural conversations with your data, asking questions like what would happen if a grid component fails, or how to reallocate human and energy resources during maintenance events. It's like having a group of specialist advisors at your fingertips. This approach is rapidly gaining traction in the energy space.

Peter Warren:
I appreciate that definition. So, it's really a case of taking investments in existing digital models and making them more consumable from a business rather than just a technical standpoint. Would that be a fair summary?

Diane Gutiw:
Absolutely. It's a great starting point for organizations to accelerate their AI journeys by extending investments they already have. By adding generative AI not just on documents (like in Retrieval Augmented Generation models), but across the entire data ecosystem, organizations can gain fast insights into complex areas without the historical complexity and expense.

4. Real-world application: Hydrogen electrolyzer digital twin operations

Peter Warren:
That's really interesting. Lukas, you've got a practical application example originally from a manufacturer of energy systems. Could you summarize that story?

Lukas Krappman:
Sure. We started with the classic digital representation of a hydrogen electrolyzer. Many clients in Germany are building PEM electrolyzers. We digitized the physical model, as Diane referred to, and then added telemetry and business processes—like what happens to pressure, temperature, current, and voltage before and after a stack.

We made it interactive. For instance, a stack could turn red or orange if something was wrong. Then users could click, investigate, and even converse with the data, asking questions like "What happened in this stack over the last day?" Was there pressure leakage or a voltage spike? So, it's about more than just monitoring—it's understanding and predicting.

5. The goal: Enhancing digital experience and data-driven manufacturing

Peter Warren:
So this originated from their desire to be more modern and data-driven, optimizing full production rather than just selling physical products. Would that be a good example?

Lukas Krappman:
Exactly. Today, it's not just about selling physical products, but about providing a digital experience as well. We aim to help manufacturers not only improve internal operations but also offer digital services to customers. This includes accessing operational data, integrating with the energy grid, and asking smarter questions like where to allocate leftover electrons. It's about transforming how products are operated and maintained digitally.

6. Unlocking energy system optimization with AI and small language models

Peter Warren:
That's really interesting. Diane, you recently published an article that hits on this point—how to overcome barriers and look ahead to optimizing whole energy systems, not just parts. Could you talk about that?

Diane Gutiw:
Sure. To optimize any system or infrastructure, including energy systems, you must start by identifying the problem you want to solve. A major area is IoT and edge device data. We're collecting massive volumes of real-time data but aren't yet fully leveraging it because it’s too complex to manually monitor.

Using newer tools to see patterns and draw insights without manual monitoring is key. Another big shift is the use of small language models—focused, efficient models trained on very specific domains rather than the entire internet. They improve computer efficiency and precision, giving very relevant answers. For example, a small language model trained specifically on IoT data can extract high-value insights quickly. The path to ROI starts by targeting and refining each part of the value chain.

7. What's next in digital innovation and energy systems?

Peter Warren:
Well, that's outstanding. Thank you for that. We're going to end this first part of a two-part series here. We hope you catch up with us for part two, where we'll dive into more innovation topics, discuss the impact of ongoing geopolitical issues, and envision the path forward. Thank you, Lukas and Diane. We'll talk to you again.

Lukas Krappman:
Thank you, have a nice day, bye-bye. Thanks, Peter.