Digital Twin Consortium Members Develop and Deploy Multi-Agent Gen AI Systems
July 23 2024 - 9:00AM
Today, Digital Twin Consortium® (DTC) announced that members are
developing and deploying Multi-agent GenAI Systems (MAGS) that are
redefining the boundaries of how product design, services, and
processes can be realized, born of efficiency and optimizations.
Use cases include automotive, infrastructure, and manufacturing,
where MAGS is utilized to drive significant productivity
improvements, streamline operations, and maximize efficiency.
Digital twins are providing advanced levels of automation
infused with GEN AI, not only integrating copilots but now
utilizing MAGS to perform a multitude of tasks either operating
independently, self-organizing, self-optimizing and
orchestrated—with or without a traditional human in the loop for
decision-making guided by human oversight that is free from
conventional repetitive routine activities.
MAGS are composed of multiple interacting GenAI-based agents
that perform various tasks, often in parallel. MAGS can now provide
decentralized, autonomous, self-organizing, and self-optimizing
capabilities. Through interaction with each other and their
environment, agents can independently achieve individual or
collective goals through reflection, memorization, and continuous
improvement.
Infused with Gen AI, each agent can perceive its environment,
including multiple modalities, make decisions, and independently
act while coordinating and communicating with other agents that may
or may not be orchestrated/managed. Some key attributes of a
digital twin-based MAGS include interaction, coordination and
control, reflection memorization, and execution.
“MAGS provide the next phase of the evolution of digital twin
systems and continue to increase business values,” said Dan Isaacs,
GM & CTO of Digital Twin Consortium. “Digital twin MAGS are
evolving to address challenges such as increasing trusted autonomy
and operating with trusted digital twins. Future applications, such
as life-critical operations, will require significant testing
across many different areas with extensive validation for
trustworthiness.”
"XMPro's work on Multi-Agent Generative Systems (MAGS)
extends intelligent digital twin capabilities to include more
complex decision-making processes. Early implementations for
an Infrastructure Water Utilities Management application have shown
potential for enhancing process optimization in real-time,” said
Pieter van Schalkwyk, CEO of XMPro. “As this technology evolves,
we're focusing on ensuring its alignment with established safety
protocols and ethical guidelines to address the challenges of
increased system autonomy.”
“Sev1Tech is leveraging the power of multi-agent generative AI
within our advanced Digital Twin platform. This integration allows
us to harness the full potential of the digital thread. This
comprehensive data framework spans the entire lifecycle of a
product,” said Greg Porter, Principal Solution Architect,
Sev1Tech. “Our innovative approach is transforming various
aspects of our manufacturing operations.”
“SODA has pioneered an advanced multi-agent system that
revolutionizes the entire automotive development lifecycle. MAGS
can autonomously implement improvements to build times, test
efficiency, and resource allocation (at an early stage),” said
Sergey Malygin, CEO of SODA. “The system learns from these
optimizations, becoming increasingly efficient without human
intervention. MAGS seamlessly integrates with Digital Twin
technology and Software Defined Vehicle (SDV) approach, creating a
dynamic, intelligent ecosystem for automotive innovation that spans
from concept to certification and after-sales.”
About Digital Twin ConsortiumDigital Twin Consortium is
The Authority in Digital Twin. It coalesces industry, government,
and academia to drive consistency in vocabulary, architecture,
security, and interoperability of digital twin technology. It
advances digital twin technology in many industries. Digital Twin
Consortium is a program of Object Management Group. For more
information, visit https://www.digitaltwinconsortium.org.
Note to editors: Digital Twin Consortium is a registered
trademark of OMG. See the listing of all OMG trademarks. All other
trademarks are the property of their respective owners.
Karen Quatromoni
Digital Twin Consortium
978-855-0412
Karen@omg.org