French industrial software company Dassault Systèmes announced the seventh generation of its 3DExperience platform at its annual user conference in Houston, addressing a persistent enterprise problem: the war rooms and Excel spreadsheets that organizations still rely on for complex business decisions.
This announcement focuses on combining data from disconnected enterprise systems into a unified virtual representation that can be queried by artificial intelligence (AI). This approach highlights a fundamental challenge facing enterprise IT, but requires significant architectural changes to be implemented at scale.
Morgan Zimmerman, CEO of 3DExperience at Dassault Systèmes, detailed an example from the electronics supply chain industry. When faced with a parts shortage, a company’s survival depends on speed. “Every morning at 6 a.m. they had a meeting with the CEO, made a €100 million arbitration, and wherever they thought there was a risk, they were pre-purchasing it in front of the OEMs.” [original equipment manufacturers] In fact, I was awake,” he told attendees during a conference session.
Principle: Whoever decides first captures everything available. But making these decisions requires correlating data across products, suppliers, inventory, and market conditions – information scattered across unconnected systems.
“It’s a real nightmare for customers to answer questions like, ‘How will tariffs affect my products and my budget?'” Zimmerman explained to Computer Weekly. “They’re building a war room of people working with Excel spreadsheets and making approximations all over the place.”
This challenge is not new. Organizations invest millions of dollars in digitizing processes and implementing enterprise systems. But when business leaders ask questions that span multiple disciplines, those systems fail to communicate effectively. Teams gather to manually cross-reference data and spend days coming up with approximations rather than final answers.
Manufacturing experts at the conference framed this as decades of incomplete digitalization. “We haven’t really solved the problem because we still need all the pieces to read all the other information,” he said. Departments become information silos, and decision-making slows as complexity increases.
unified data representation
Addressing this requires a fundamental shift in how enterprise data is structured and accessed. This approach involves projecting information from multiple sources into a unified representation that maintains relationships and context, rather than having systems operate independently with occasional data exchange.
Zimmerman used the analogy of a map to explain the concept. “If you have an Excel spreadsheet with restaurant locations and another Excel spreadsheet with florist locations, and you’re trying to find a restaurant near a florist, it’s difficult,” he said. “It’s easier when it’s on a map because the data is inherently correlated.”
Dassault’s “3D Universe” implements this through a virtual twin, which is a digital representation of a physical product, system, or process that serves as a common frame of reference. Click on a component to view quality history, cost data, supplier information, and design specifications, regardless of which system that information was originally stored on.
Power comes from combining these expressions. Mr. Zimmerman’s tariff scenario illustrates this. “To understand how tariffs impact your business, you need to combine multiple virtual twins,” he explained. “We need a virtual twin of the product because we need to know which components are affected. We need a virtual twin of the production system because we need to understand where and in what quantities to assemble. We also need a virtual twin of the supply chain because we need to understand what to buy and from whom.”
Technical challenges include correlating data originating from disconnected systems, such as product specifications from product lifecycle management, production schedules from manufacturing execution systems, and supplier information from enterprise resource planning. [ERP]and quality metrics from the test system. Dassault’s approach uses structured data models that define relationships such as the relationship between a product and its components, the relationship between components and suppliers, and the relationship between suppliers and production facilities.
The architecture must be proven effective across a variety of enterprise environments, especially those with legacy systems and disparate data.
Conversational access
Uniform data representation solves some of the problems. To access these, you need an interface that doesn’t force users to understand complex data structures or work with multiple applications.
Conversational AI approaches are becoming increasingly common across enterprise software and aim to enable users to ask questions naturally, rather than writing database queries or clicking through application menus. Dassault’s implementation includes what the company calls a “virtual companion” and will begin in mid-2026.
The company has deployed three AI agents with expertise in different areas. Aura functions as a business analyst with program management and strategy capabilities. Leo focuses on engineering, design and manufacturing. Marie is responsible for scientific areas such as materials and testing. “They answer the questions you’re asking more accurately or with a different level of accuracy,” Zimmerman says.
For scenarios that require external information, such as rate changes or supply disruptions, Aura can integrate news feeds and market data to identify relevant events. However, the actual impact calculation uses the customer’s own company data. If Aura determines that customs duties will cost €3.3 billion, that figure comes not from external sources but from an analysis of its customers’ product mix, production volumes and relationships with suppliers.
Dassault has built a library of trusted news sources by industry, but organizations can extend these with their own preferred sources.
A demonstration at the conference asked questions such as “What is the status of my order? When will it be shipped? What is the price?” Answer by retrieving data from ERP and manufacturing execution systems. The project manager asked, “Where is my project? What’s stopping me from releasing it?” And I received the summary. Change management processes were handled conversationally rather than forms and approval workflows.
Effectiveness depends on AI models understanding domain-specific terminology and context, recognizing that “rate of supply” has different meanings in manufacturing and agriculture, for example. Dassault claims its Companion leverages decades of industry-specific knowledge encoded in its software, and its capabilities are expanded monthly with new “skills.”
In production, test how well these AI agents handle ambiguous queries, contradictory data, or requests outside of training.
Platform requirements
Zimmerman argues that this approach involves more than simply connecting existing systems. “Product data management is a narrow focus of what we do with the 3DExperience platform,” he said. “Dassault Systèmes’ greatest strength is its ability to abstract customer complexity.”
This distinction includes modeling not only components, but also entire product configurations, production systems, and their relationships. “We believe that what we have positioned in terms of our ability to abstract and represent product complexity is a fundamental baseline predictive system for scaling AI,” Zimmerman said.
This platform strategy aims to “democratize” information, making corporate knowledge accessible across roles and departments without the need for everyone to understand every system. Manufacturing engineers query company standards for specific processes and get answers from the same underlying data, whether programming a machine or designing a component.
Technical challenges include maintaining data consistency when information originates from systems of record that continue to operate independently. Updates to one system must be accurately reflected in a unified representation, raising questions about synchronization delays and conflict resolution.
intellectual property protection barrier
During the conference discussions on intellectual property, practical obstacles emerged. As organizations increasingly share detailed data with suppliers and partners, questions about AI learning privileges arise.
“Data is what matters most in the age of AI,” Zimmerman says. “If you’re a manufacturer and all your suppliers are sharing data with you, the question is: If you start using that data for AI, do you have the right to do so?”
Dassault introduced IP Lifecycle Management to address this issue. This tracks not only data access, but also whether an AI model can be trained on a particular dataset and who owns the IP derived from that learning.
Zimmerman referred to arguments for equipment manufacturers in regulated industries to only share detailed designs that ensure their data will not be used for AI training without explicit consent. “IP protection no longer just means protecting. It means protecting the fact that you don’t learn about data you don’t have rights to.”
The system maintains lineage tracking. When suppliers provide data with consent restrictions, the platform applies those restrictions to AI learning and tracks derived models.
Companies with complex supplier networks require a robust governance framework that defines data usage rights across organizational boundaries. Technical management alone is not enough.
The transition Dassault envisions is from teams manually correlating information to conversational AI querying unified data environments. From days to generate approximations to seconds to calculate scenarios. From departmental silos to what Zimmerman calls “a single point of understanding the data landscape.”
Virtual Companion will launch in mid-2026 in the cloud only due to computing requirements. This approach requires significant architectural changes. It’s not just about implementing new software; it’s about rethinking how enterprise data is structured, accessed, and managed.
Success will depend on factors beyond technology, such as integration complexity, change management, data governance, and proving that speed gains persist in the face of real-world enterprise complexity.
This announcement suggests that at least the enterprise software industry recognizes the need to solve war room problems. Whether integrated virtual representations and conversational AI provide the answers awaits broader implementation. This announcement confirms that the industry is aware of the problem. Now comes the hard part: proving your solution works at enterprise scale.
