Dassault Systèmes has spent years selling a platform story. Gian Paolo Bassi, one of its most recognisable product voices, now talks about it less as a roadmap and more as a habit formed under pressure. I visited the company mid-way through a recent start-up accelerator event in London.
The pressure, in his telling, comes from startups that move too quickly to tolerate friction, and from manufacturers learning the hard way that fragile supply chains and porous data environments can turn engineering into a liability. This is a pitch tightening around one idea: engineering teams move faster when design, simulation, manufacturing, and documentation behave like one system, and when the data behind it can be trusted.
Bassi, a senior vice president at Dassault Systèmes, frames the company’s startup work as something closer to a reciprocal arrangement than a benevolent programme. “It is a two-way type of collaboration,” he says. The obvious value runs one way: access to tools, networking, and introductions that can matter when a company is trying to move from prototype to product. Yet he insists the exchange changes Dassault as well, because early-stage firms are blunt about what slows them down.
Customisation falls out of favour
“We do not approach clients with long propositions of ROI. Our ROI has to be measured in weeks and actually in minutes,” he says. The point is not marketing. It is an operating doctrine. If a new hire cannot be productive immediately, the product is failing. If a workflow requires months of tailoring, it risks becoming the next legacy trap. Bassi is unusually direct about customisation, a feature once treated as a badge of enterprise seriousness. He describes deep bespoke work as a short-term boost that becomes a long-term penalty, leaving customers stuck on old configurations and unable to adopt newer capabilities without pain.
The startups, he argues, force a different expectation. They do not arrive asking for a single licence, a point tool, or a stack of file converters. They arrive with a deadline and a product to ship. They want design, simulation, data management, and manufacturing preparation to function as a single system. Translating files, shuttling formats, and managing disconnected standards feels to them like a wasteful tax on velocity.
Simulation moves forward in the workflow
This is where Bassi keeps returning to simulation. “Simulation cannot be an afterthought or a downstream process. Simulation has to be upstream,” he says, describing what Dassault labels “mod-sim”: treating analysis as something that sits beside design from the first sketches. The logic is familiar to anyone working with additive manufacturing. The geometry is more ambitious, the tolerances can be tight, and the consequences of getting build behaviour wrong are expensive. Metal additive has pushed this fastest, he suggests, because it is both a prototyping enabler and an area where many startups are trying to change the economics of machines, materials, and processes.
The company’s startup programme has shifted in less glamorous ways too. One of the programme leaders notes that the duration and support structure changed after repeated feedback that the journey from idea to shipped product takes longer than the original schedule assumed. “We are listening to the ecosystem,” they say. That change, Bassi implies, was not a concession. It was an acknowledgement that the bottlenecks in modern engineering are rarely in CAD itself. They sit downstream in manufacturing readiness, supply chain choices, compliance documentation and the handoffs where errors multiply.
Marketplace rethink: partner first, platform second
That downstream focus appears again when Bassi turns to Dassault’s marketplace initiative. The company launched its marketplace years ago, attached to the same integrated-platform pitch. If you can design in SolidWorks and manage data in one environment, why not connect to manufacturing services in the same place? In his retelling, the effort underperformed for a basic reason. “We did not dedicate enough attention. It did not grow the way it should,” he admits. The plan now is to return to it with a different structure: less a monolithic Dassault-owned marketplace and more a federation of specialist platforms.
Bassi speaks about existing marketplaces with a pragmatism that is not always common among software chiefs. These platforms already work, he says, because they focus tightly on one job: manufacturing procurement, product sourcing, capital, engineering services. Dassault’s role, as he sees it, is to bring a common language and a common environment that lets these specialist networks interoperate, rather than trying to replace them.
Xometry as an amplifier, not a competitor
Xometry is his favoured example. The US-based manufacturing marketplace could be cast as a rival to a platform that wants to own the design-to-production path. Bassi rejects that framing outright. “It actually is a partner. We are very close to that company,” he says. The reason is simple and commercial. A large share of the work flowing through these distributed production networks begins as a SolidWorks file. The more seamlessly designers can move from model to quote to manufacture, the more SolidWorks becomes the invisible default.
Asked whether that relationship might one day become an acquisition, he dismisses it as a live topic. “It is not on the table right now,” he says, before adding that it “could be a good idea” in theory. His preference is to let such firms scale independently. Dassault benefits if the ecosystem thrives, particularly if the ecosystem uses Dassault’s tools as its shared vocabulary.
Buying ideas, not market share
When Bassi does talk about acquisitions, he chooses a line that sounds rehearsed and still carries weight. “We do not buy market share. We only buy amazing ideas,” he says. It is a tidy way of stating an approach that prioritises intellectual property and technical depth. Dassault’s confidence, he suggests, comes from breadth. The company already spans design through high-end simulation, including electromagnetic analysis. What it wants to buy, when it buys, is the sort of technology that expands its scientific and engineering base, not a rival’s customer list.
Materials data becomes a competitive asset
Materials data is a good example of where Dassault has been building quietly. Bassi says the company has a dedicated materials team that has developed relationships with more than 1,000 material manufacturers. The result is a growing internal materials characterisation database used across simulation workflows, from stress and strain to manufacturability. He describes capabilities that matter in additive manufacturing in particular: predicting thermal stress, simulating sintering behaviour, and modelling process parameters such as speed. The database does not have a consumer-facing name, he says. It is embedded in the platform rather than sold as a product in its own right.
If the marketplace is the outward-facing attempt to connect design to production, AI is the inward-facing attempt to make the platform usable at speed. Bassi’s examples are not grand visions of autonomous engineering. They are the unglamorous irritations that add up in an eight-million-user ecosystem. He recalls the original observation that placing a bolt in an assembly required multiple clicks. The fix became machine-learning-assisted suggestions and automated placement logic. “If we save one click per user, that has a big impact. We save 8 million clicks,” he says. It lands as a quick line. It is also a statement about scale. Tiny frictions matter when your product is work.
Aura targets documentation and embedded know-how
Over the past year, that AI effort has moved into a more visible phase with Aura, a conversational assistant built with French large language model developer Mistral. Aura is designed to let engineers interact with the system in natural language and, crucially, to train on a customer’s intellectual property without pushing that knowledge outside their controlled environment. Bassi points to an unromantic problem that almost every manufacturer recognises: documentation.
Companies sit on vast piles of specifications, internal standards, and regulatory obligations. Finding the relevant clause can be harder than writing the design. Proving that the final product matches the specification often remains manual, expensive, and error-prone. Bassi describes Aura as a way to query those documents directly, pulling answers such as minimum draft angles for moulding or internal rules that sit in a forgotten PDF. That, he argues, is the first step. The next step, still in development, is to extend Aura from text into engineering models. If the rule was never written down, the designs may still contain it. Dassault wants to extract that embedded knowledge by analysing geometry, features, and historic decisions across a company’s archive.
When pressed for measurable AI impact, Bassi returns to two areas: transparency and exploration. The first is about reducing time wasted translating intent into a digital model. Ask the assistant how to create a loft, use the right command, and avoid the frustration of hunting menus. The second is generative and simulation-led design. Engineers already know how to model a bracket, he says. What they cannot usually afford is the time to test many versions of it. AI-backed optimisation opens that design space, letting engineers explore weight, cost and material trade-offs faster than manual iteration allows.
He offers a startup example rather than a blue-chip case study. A company called Raeon, which designs custom batteries, uses simulation augmented by AI to build confidence around safety and reliability. Bassi says the startup’s head of engineering, Marisa Kurimbokus, told him that wider simulation use increased customer confidence and helped adoption. It is a reminder that AI’s value in engineering can be indirect. It changes the sales conversation by making the technical argument easier to evidence.
The orchestration gap
Dassault’s current limits are revealing. Bassi is clear that the company has many narrowly focused AI “companions” that handle specific tasks: extracting sketches, turning a picture into 3D, converting tessellated data into parametric surfaces, and improving rendering through AI denoising. What it does not yet have, he says, is orchestration: a single system that can chain these capabilities into an end-to-end workflow from “A to Z” without manual stitching. “We have a lot of good AI companions that do very vertical tasks,” he says. Making them work together as a coherent assistant is the next challenge.
That challenge sits alongside a more basic one: users can feel intimidated by the platform’s sheer breadth. “Usability,” Bassi says, is the main friction point. He describes a dilemma. The platform contains a huge range of capabilities, yet arriving on it can feel like landing in a cockpit. Dassault is using machine learning to simplify what each user sees based on patterns and industry context, and to guide them toward a smaller subset of tools aligned with their immediate goal. Simulation, he adds, should not require a PhD. Engineers should be able to describe what they are trying to test, and the system should narrow the choices.
Model-based definition pushes drawings aside
Asked what he would rebuild from scratch, Bassi points to drawings, a subject that still divides engineering organisations. He wants the stack to lean harder on model-based definition, where design intent is expressed in machine-readable form rather than as a 2D artefact. Dassault has already released automated drawing generation from 3D models, reaching around 80 per cent completion for prismatic parts before human refinement. The aim is not to abolish drawings overnight. It is to remove the labour of making them and shift the centre of gravity toward the model.
The competitive gap he identifies is not technological. It is narrative. “We have a gap in our storytelling,” Bassi says, arguing that Dassault’s integrated platform and common language are underappreciated compared with point competitors that stop at file management. He returns repeatedly to intellectual property: protecting it, extending it, and ensuring customers do not become prisoners of their tools.
Security and sovereignty are part of that pitch. Dassault operates Outscale data centres and emphasises contractual commitments around data location and protection. Bassi says the company has roughly 2,000 engineers dedicated to cybersecurity. He describes the procurement realities facing smaller suppliers: major OEMs now require credible cybersecurity strategies as a condition of doing business. Buying into a secured platform becomes, in many cases, the only practical way to meet that standard.
2030: manufacturing shifts to resilient networks
By the end of the conversation, Bassi has widened the frame again, away from CAD features and toward industrial structure. His 2030 prediction is not about a single modelling breakthrough. It is about production organisation under geopolitical strain. He argues that concentrating manufacturing capability and know-how in a single geography has proved dangerous, not only politically but operationally. A ship stuck in the Suez Canal can expose how thin the margin of resilience has become.
“I believe the time of vertically integrated companies has ended. It is the time of vertically integrated networks,” he says. The distinction matters in his mind. Companies will still want integration, but achieved across a network of suppliers, materials sources, and manufacturing nodes rather than inside one corporate boundary. The winning organisations, he suggests, will be those that can choose and reconfigure these networks quickly, building resilience into design choices, supplier qualification and manufacturing strategy.
That is where Dassault wants the platform to sit, acting as connective tissue for engineers, entrepreneurs, and manufacturers. Bassi compares the ambition to a social network, though he leans into the analogy carefully. The goal, he says, is to “grease” the circulation of ideas that drives innovation, and to connect people across borders and firewalls. In his view, the network effect is already visible in miniature, with entrepreneurs exchanging knowledge across different industries inside the startup ecosystem.
The claim is large. The evidence, so far, is more grounded: startups demanding less friction, simulation pulled forward into design, marketplaces treated as partners, and AI deployed in ways that shave time from real work. It is a platform strategy measured not by slogans, but by whether an engineer can move from an idea to a manufacturable, documented, secure product with fewer wasted steps.
The interview sketches a company shifting from CAD vendor to infrastructure provider for product development, where the prizes are resilience and speed rather than feature parity. Dassault’s bet is that integrated simulation, materials intelligence, secure data environments, and partner marketplaces will matter more as manufacturing fragments into networks, and that AI will be judged by how much it shortens the path from intent to verified production.
The 3D Printing Industry Awards are back. Make your nominations now.
Do you operate a 3D printing start-up? Reach readers, potential investors, and customers with the 3D Printing Industry Start-up of Year competition.
To stay up to date with the latest 3D printing news, don’t forget to subscribe to the 3D Printing Industry newsletter or follow us on Linkedin.
