French industrial software company Dassault Systèmes unveiled Generation 7 of its 3DExperience platform at its annual user conference in Houston, addressing what it sees as a persistent enterprise problem: the war rooms and Excel spreadsheets organisations still depend on for complex business decisions.
The announcement centres on combining data from disconnected enterprise systems into unified virtual representations that artificial intelligence (AI) can query. The approach highlights a fundamental challenge facing enterprise IT – but practical implementation at scale will require significant architectural changes.
Morgan Zimmerman, CEO of 3DExperience at Dassault Systèmes, recounted an example from the electronics supply chain industry. When component shortages hit, one company’s survival depended on speed. “Every morning at 6am, they had a meeting with the CEO, and they were doing €100m arbitrations, buying stuff in advance, wherever they thought they were having risks, before the OEMs [original equipment manufacturers] were actually even waking up,” he told attendees during a conference session.
The principle: whoever decides first captures whatever’s available. But reaching those decisions requires correlating data across products, suppliers, inventory and market conditions – information scattered across disconnected systems.
“For our customers to answer questions like ‘what is the impact of the tariff on our products or on the budget’, it’s an absolute nightmare,” Zimmerman explained to Computer Weekly. “They’re building war rooms with people manipulating Excel spreadsheets and doing approximations everywhere.”
The challenge isn’t new. Organisations invest millions digitising processes and implementing enterprise systems. Yet when business leaders ask questions spanning multiple domains, those systems don’t communicate effectively. Teams assemble to manually cross-reference data, spending days producing approximations rather than definitive answers.
Manufacturing experts at the conference framed this as decades of incomplete digitisation. “We don’t really solve the problem because you still have to have all the bits and pieces to be able to read all of that other information,” he said. Departments become information silos, slowing decision-making as complexity increases.
Unified data representations
Addressing this requires fundamentally changing how enterprise data is structured and accessed. Rather than systems operating independently with occasional data exchanges, the approach involves projecting information from multiple sources onto unified representations that preserve relationships and context.
Zimmerman used a map analogy to explain the concept. “If you take an Excel spreadsheet with location of restaurants and another Excel spreadsheet with location of flower shops, and you try to find a restaurant nearby a flower shop, that’s difficult,” he said. “If it’s on the map, it is simple because the data are correlated by nature.”
Dassault’s “3D Universes” implement this through virtual twins – digital representations of physical products, systems or processes that serve as the common reference frame. Click on a component and see quality history, cost data, supplier information and design specifications, regardless of which systems originally stored that information.
The power comes from combining these representations. Zimmerman’s tariff scenario illustrates this. “To understand the impact of the tariff on your business, you need to combine multiple virtual twins,” he explained. “You need to have the virtual twin of the product because you need to know what components are affected. You need to have the virtual twin of the production system because you need to understand where you assemble and for what volumes. And you need to have the virtual twin of the supply chain because you need to understand what you buy from whom.”
The technical challenge involves correlating data that originates in disconnected systems – product specifications from product lifecycle management, production schedules from manufacturing execution systems, supplier information from enterprise resource planning [ERP], and quality metrics from testing systems. Dassault’s approach uses structured data models that define relationships: how a product relates to its components, how components relate to suppliers, how suppliers relate to production facilities.
The architecture must prove itself across diverse enterprise environments, particularly those with legacy systems and heterogeneous data.
Conversational access
Having unified data representations solves part of the problem. Accessing them requires interfaces that don’t force users to understand complex data structures or navigate multiple applications.
The conversational AI approach – increasingly common across enterprise software – aims to let users ask questions naturally rather than construct database queries or click through application menus. Dassault’s implementation involves what it calls “virtual companions”, launching mid-2026.
The company introduced three AI agents with distinct domain expertise. Aura functions as a business analyst with programme management and strategy capabilities. Leo focuses on engineering, design and manufacturing. Marie handles scientific disciplines including materials and testing. “They will respond with more precision or a different level of precision to the question you are asking for,” said Zimmerman.
For scenarios requiring external information – such as tariff changes or supply disruptions – Aura can integrate news feeds and market data to identify relevant events. But the actual impact calculations use the customer’s own enterprise data. When Aura determines that a tariff will cost €3.3bn, that figure comes from analysing the customer’s product configurations, production volumes and supplier relationships – not from external sources.
Dassault has built libraries of trusted news sources by industry, though organisations can extend these with their own preferred sources.
Conference demonstrations showed queries like, “What’s the order status? When does it ship? What does it cost?” answered by pulling data from ERP and manufacturing execution systems. Project managers asked, “Where is my project? What is blocking the release?” and received summaries. Change management processes were handled conversationally rather than through forms and approval workflows.
The effectiveness depends on AI models understanding domain-specific terminology and context – recognising that “feed rate” means something different in manufacturing than in agriculture, for instance. Dassault claims its companions leverage decades of industry-specific knowledge encoded in its software, with capabilities expanding monthly through new “skills”.
Production environments will test how well these AI agents handle ambiguous queries, conflicting data or requests outside their training.
Platform requirements
Zimmerman argues the approach requires more than connecting existing systems. “Product data management is a narrow view of what we do in the 3DExperience platform,” he said. “The biggest strength of Dassault Systèmes is its ability to abstract and represent the complexity of our customers.”
The distinction involves modelling not just components, but entire product configurations, production systems and their relationships. “We believe that what we have positioned in terms of ability to abstract and represent the complexity of the product is the fundamental baseline projection system to scale AI,” said Zimmerman.
This platform strategy aims to “democratise” information – making enterprise knowledge accessible across roles and departments without requiring everyone to understand every system. A manufacturing engineer queries company standards for specific processes and gets answers whether programming machinery or designing components, drawing from the same underlying data.
The technical challenge involves maintaining data consistency when information originates in systems of record that continue operating independently. Updates in one system must reflect accurately in the unified representation, raising questions about synchronisation latency and conflict resolution.
IP protection barriers
A practical obstacle emerged during conference discussions around intellectual property. As organisations increasingly share detailed data with suppliers and partners, questions arise about AI learning permissions.
“In the age of AI, the most important thing is the data,” said Zimmerman. “If you are a manufacturer and all your suppliers are sharing data with you, the question becomes: if you start using that data for AI, do you have the right to do so?”
Dassault introduced IP lifecycle management to address this – tracking not just data access, but whether AI models can train on specific datasets and who owns derivative IP from that learning.
Zimmerman cited discussions where equipment manufacturers in regulated industries would only share detailed designs with guarantees that data wouldn’t be used for AI training without explicit consent. “IP protection does not mean just securing it anymore,” he said. “It means securing the fact that you’re not going to learn on data on which you do not have the right.”
The system maintains lineage tracking – when suppliers provide data with consent restrictions, the platform enforces those restrictions for AI learning and tracks derivative models.
Enterprises with complex supplier networks will need robust governance frameworks defining data usage rights across organisational boundaries – technical controls alone won’t suffice.
The shift Dassault envisions moves from teams manually correlating information to conversational AI querying unified data environments – from days producing approximations to seconds calculating scenarios; from departmental silos to what Zimmerman calls “a single point of understanding of the data landscape”.
The virtual companions launch mid-2026, cloud-only due to computing requirements. The approach requires significant architectural changes – not simply implementing new software, but rethinking how enterprise data is structured, accessed and governed.
Success depends on factors beyond technology: integration complexity, change management, data governance and proof that speed improvements survive contact with real enterprise complexity.
The announcement suggests, at minimum, that the enterprise software industry recognises the war room problem needs solving. Whether unified virtual representations and conversational AI provide the answer awaits broader implementation. The announcement confirms the industry recognises the problem. Now comes the hard part: proving the service works at enterprise scale.
