CGI’s digital triplet approach extends the digital twin model by applying AI-driven analysis to increase the usability and interpretability of the insights. By synthesising data quickly and communicating outcomes clearly, digital triplets help organisations enhance decision-making.
Through advanced AI capabilities and human-AI interaction, the digital triplet serves as a third-party advisor to the digital twin end user, such as decision-makers in a health system, manufacturer, utility, or other enterprise.
This capability enables organisations to vary scenarios, ask validating or alternative criteria questions, and explore options in partnership with AI using real-time natural language through text, voice or both. This innovative approach enables more personalised, evidence-based, and transparent decision-making and recommendations to create more business opportunities and reduce risks. This solution also increases the value and return on investment for digital twin capabilities.
The digital triplet is CGI’s solution to providing more insight to decision-makers through an agent that can interrogate and explore the digital twin. By adding generative AI and explainable AI to digital twin solutions, organisations can use natural language to explore the insights, gain in-context recommendations, and investigate alternate scenarios in real time.
Our digital triplet approach consists of three components:
The physical entity can be a person, physical asset, or a physical architecture (building or complex integrated system such as a utility grid). The physical entity is the subject that is assessed, monitored, and evaluated (the “as-is”) to improve functioning and/or outcomes.
The digital twin models the physical entity’s state and dynamics using data-driven and knowledge-based methods. It is a virtual representation (the digital representation of the “as-is”) of the physical entity that can be monitored and evaluated. It may include a single system or a full virtual blueprint of all systems and interactions. In the case of a physical asset, a digital twin includes the asset’s technical specifications and operational history. It processes real-time data collected from edge devices and operations, as well as from diagnostic tests or monitoring devices attached to the physical entity. In addition to complex monitoring, the digital twin can incorporate causal inference techniques to identify cause and effect relationships among variables and the potential effects of interventions (the digital “could-be”).
The digital triplet: Intelligent advisor that allows the user to interrogate the digital twin. This functionality is provided through three AI technologies: GenAI to generate scenarios, XAI to communicate GenAI results to a human expert in an understandable way, and a multi-agent ecosystem to seamlessly enable secure, reliable, and robust AI output.
- Generative AI (GenAI) uses information provided by a digital twin to help interpret the data and explore additional and more complex hypothetical scenarios through a scenario generator. Its methods may include reinforcement learning, simulation, or optimisation to generate and evaluate scenarios according to different objectives and constraints, such as maximising efficacy, minimising side effects, visualising unintended impacts, reducing costs, or aligning with policy guidelines and organisational policies and preferences. The digital triplet can also use extended capabilities to generate realistic and high-quality images of assets or ecosystems in addition to text responses.
- Explainable AI (XAI) communicates the results of GenAI scenario generation to a human expert in a transparent and understandable way. It uses natural language (text or speech), visualisation, or interactive dialogue to present scenarios, highlight differences and similarities, justify recommendations, and answer questions from the human expert. Pairing this ability with routing capabilities and mobile device data, such as location, moves solutions from monitoring of stationary screens, to provide actionable insights directly to the most appropriate individuals. Explainable AI can also elicit feedback from the expert and use it to refine the scenarios or generate new ones.
- Multi-agent ‘worker bees' are seamless to the users that interact with the solution through speech or text, but provide a powerful approach to ensuring the accuracy, trustworthiness and reliability of the delivered information. These agents help complete all of the tasks required to retrieve, generate, synthesise and validate information across discrete and narrative data, documents and images. By treating the foundation model agents as services, CGI can match the right agent to the use case or user need and configure a digital triplet that enhances existing investments in data and infrastructure.
By considering multiple factors and outcomes, examining different options, and explaining reasons and trade-offs, digital triplets:
- Deliver customised, evidence-based, and proactive advice.
- Significantly boost the capabilities of human experts by increasing their knowledge, independence, and creativity.
- Facilitate synergistic collaboration between humans and AI, driving responsible use and trusted outcomes.
- Enable industry experts to create new hypotheses, test new interventions, and discover new insights from data and models, increasing the value of their digital twin investments.
- Increase the value and ROI of digital twin investments and implementations.
