AI for CAD Tools
What Dassault's 3DEXPERIENCE Virtual Companions actually do for CAD and PLM, where the native AI stops, and how Leo closes the gap for engineering teams.
·
⏱
9 min read
Michelle Ben-David
Michelle Ben-David is a mechanical engineer and Technion graduate. She served in an IDF elite technology and intelligence unit, where she developed multidisciplinary systems integrating mechanics, electronics, and advanced algorithms. Her engineering background spans robotics, medical devices, and automotive systems.
BOTTOM LINE
3DEXPERIENCE unifies CATIA, ENOVIA, SIMULIA and DELMIA on one data model, and in 2026 Dassault added native, agentic Virtual Companions on top of it for program management, engineering and simulation work. Those companions are genuinely capable inside the platform's own data, but their reach ends at its boundary: legacy CATIA V5 files, SOLIDWORKS assemblies, a second PDM/PLM system, or specifications sitting on a network share are all outside their scope by design, not by oversight. A tool like Leo AI addresses that gap directly, connecting to PDM, PLM, ERP and file systems at once so retrieval does not stop where the platform does. For most mechanical engineering teams, the practical answer is not choosing between the two, it is using the native companions for what is inside 3DEXPERIENCE and a cross-system layer for everything that is not.
3DEXPERIENCE is the one major CAD and PLM platform most "best AI tools" roundups skip, mostly because it does not fit neatly into either category. It is CATIA and ENOVIA and SIMULIA and DELMIA running on one shared data model, which means the AI question is not "which plugin do I add" but "what does the platform already do, and where does a team still need to look elsewhere." In 2026, Dassault Systemes gave that question a real answer with a set of native, agentic AI companions built directly into the platform. This post covers what those companions actually do, where their reach ends at the edge of the 3DEXPERIENCE data model, and what a team running CATIA V5, SOLIDWORKS or a second PDM/PLM system alongside it still has to solve for itself.
What 3DEXPERIENCE Actually Is
Most mechanical engineers know CATIA as CAD and ENOVIA as PDM/PLM, but on the 3DEXPERIENCE platform they are not separate products stitched together after the fact. They share one underlying data model, alongside SIMULIA for simulation and DELMIA for manufacturing. A change made to a part in CATIA and a change order tracked in ENOVIA reference the same product structure rather than two structures kept in sync by an integration. Dassault also connects legacy CATIA V5 data and SOLIDWORKS models into this same environment, so a team migrating in stages does not have to convert everything on day one. The pattern is not unique to Dassault: Siemens NX runs a similar single-platform argument, and the CAD choice underneath it still shapes what a team can do years after the decision is made.
That single data model is also why 3DEXPERIENCE is harder to summarize than a point tool. ENOVIA on the platform handles bills of materials, change and configuration management, requirements, quality and compliance, and design review, and it does so as one layer under CATIA rather than as a separate system a design has to be checked into. For a team already standardized on the platform, that removes a category of integration work other PDM/PLM setups still carry, and it is a large part of why enterprises in aerospace, defense and automotive adopted the platform in the first place.
IN PRACTICE
It integrates directly with PLM and existing workflows, making past designs, standards, and calculations instantly available. The result is fewer errors, faster decision-making, and a more consistent process across teams.
- Sergey G., Board Member
The Native AI: Virtual Companions
At 3DEXPERIENCE World 2026, Dassault introduced Virtual Companions: named, agentic AI assistants built into the platform rather than bolted on as a chat window. Three were announced with distinct scopes, one for program management, one described as handling complex engineering work, and one aimed at deep science and simulation. Dassault frames them as designed to "co-create and cooperate with users, to co-develop products, assets and services, solving complex industrial challenges in a faster and more efficient way," and says they draw on "structured industry knowledge and know-how to interpret user intent, reason in an industrial context, take decisions and ultimately generate outcomes grounded in science."
Functionally, that means the companions are meant to work across the whole workflow the platform already spans: design, engineering, simulation, manufacturing and operations, not just one stage of it. Dassault also says the companions run on OUTSCALE's sovereign AI cloud infrastructure across three continents, which is the kind of detail that matters if a team's compliance requirements dictate where model inference physically happens. One naming note worth flagging directly: Dassault's own engineering-focused companion is named LEO. That is Dassault's name for its own tool and has no connection to Leo AI, the platform discussed later in this post.
Where the Built-In AI Reaches, and Where It Stops
The honest scope of a native AI companion is the data model it was built on top of. Virtual Companions reason over the structured product, program and simulation data that already lives inside 3DEXPERIENCE. That is a genuinely large amount of ground when a team is fully standardized on the platform. It is a smaller amount of ground the moment any part of the engineering record lives outside it: legacy CATIA V5 files that have not been migrated, SOLIDWORKS assemblies from an acquired team, a second PDM/PLM system running in parallel after a merger, standards and specification PDFs on a network share, or supplier documentation that never made it into ENOVIA in the first place. None of that is a criticism of the companions themselves, it is simply what "built into the platform" means for any vendor's native AI, on 3DEXPERIENCE or elsewhere, the same way Windchill's own built-in AI stops at its own data model.
The other honest note is timing. Virtual Companions were unveiled in February 2026 and expanded with new skills mid-year, which makes this a first-generation agentic capability rather than a mature, years-old feature set. Coverage of specific workflows, and how deep the reasoning goes on any one of them, should be expected to keep changing release over release, which is normal for a first-generation platform feature and not a reason to wait on evaluating it.
Leo AI: An Intelligence Layer That Sits Across the Whole Stack
This is where a tool like Leo AI fits, and it is worth being precise about the fit. Leo is not a replacement for ENOVIA or a competitor to Dassault's own companions, it is an intelligence layer that sits on top of an organization's full knowledge base, connecting to PDM and PLM platforms, local and network directories, and ERP systems at once. Leo offers integrations with leading PDM and PLM platforms, including SolidWorks PDM, Autodesk Vault, PTC Windchill, Siemens Teamcenter and Arena PLM, among others, alongside whatever a team already runs on 3DEXPERIENCE.
The value driver is retrieval that does not stop at a platform boundary. A mechanical engineer searching for how a similar bracket was toleranced two product generations ago, or which supplier part can replace a twenty-year-old assembly, is usually not asking a question that lives entirely inside one system. Leo is trained on more than a million pages of standards, books and articles on top of an organization's own data, so the same search surfaces an internal design decision, the governing ASME or ISO standard behind it, and a citation the engineer can click through and verify, regardless of whether the underlying file sits in ENOVIA, a second PLM system, or a folder that predates all of them. That kind of cross-system search is the same problem covered from the CATIA side in this look at what actually works for CATIA users today. For a team on 3DEXPERIENCE specifically, that means the native companions and Leo are not an either-or choice: the companions cover what is inside the platform, and Leo covers the search that has to cross out of it.
How to Decide What You Actually Need
The practical question for an engineering team is less "which AI is better" and more "how much of our engineering record actually lives in one place." A team that is fully migrated to 3DEXPERIENCE, with CATIA, ENOVIA and SIMULIA all current, is closer to getting full value from the native companions as they mature. A team mid-migration, running CATIA V5 alongside 3DEXPERIENCE, or maintaining a second PDM/PLM system after an acquisition, or storing decades of legacy designs on a network drive, has a retrieval problem the platform's own AI was never built to solve, because that data sits outside its model by definition. The same migration reality shows up across the CAD landscape, including among teams still deciding what a platform like Solid Edge does and does not cover.
Most mechanical engineering organizations, including ones standardized on Dassault tools, are somewhere in that second category rather than the first. Security is not a reason to hesitate here: any layer added on top should be SOC-2 certified and GDPR compliant, should not train on customer data, and should keep IP protected regardless of which CAD or PLM system it connects to. The realistic 2026 answer for most 3DEXPERIENCE users is to let the platform's own companions handle what is native to it, and add a retrieval layer that spans everything the platform boundary does not reach.
FAQ
See What Leo Finds Across Your Systems
One search across CAD, PLM, ERP and network drives, wherever data lives.
Leo connects to your CAD, PLM and ERP systems plus network directories, so engineers get one answer instead of five separate searches.
Schedule a Demo →
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams
