Last week I attended the IFIP 23rd International Conference on Product Lifecycle Management (PLM26) in Lecce, Italy, hosted by the University of Salento and its Department of Innovation Engineering. Three days (July 6-8), a doctoral workshop, industry day, keynotes from academia and industry, a digital twin contest, and, honestly, one of the best settings a PLM conference could ask for — the baroque streets of Lecce in July.
In this article, I want to share my impressions of the conference overall — background, agenda, welcome sessions, and keynotes. Two follow-up articles will go deeper: one about Martin Eigner’s PLM Pioneer keynote “40 Years of PLM: From Documents to Product Intelligence,” and one about my own keynote, “From CAD Files to Product Memory,” where I also announced my upcoming book.
A Bit of History: 23 Years of the IFIP PLM Conference
The IFIP PLM conference is not a vendor event and not a marketing show. It is an academic and industry research conference running under the umbrella of IFIP — the International Federation for Information Processing. The conference was founded in 2003 by Prof. Debasish Dutta, who chaired the first edition at the Indian Institute of Science in Bangalore. Since then, the conference has traveled the world every year, deliberately building bridges between research communities in different countries.
Prof. Dutta, now honorary chair, returned to the conference after about ten years, and his opening remarks set a tone about AI and democratization of the tech: AI is impacting every company and every country and aligning capabilities of people to innovate and accelerate their research. His call to the PLM community was to think about how to democratize the positive impact of AI in a responsible and ethical use, with special attention to SMEs and under-resourced companies. I found this was a refreshing and important framing.
The Theme: Complexity, Intelligence, Resilience, Sustainability
The official theme of PLM26 was “Frontiers in PLM: Managing the Complexity of Industry through Intelligent, Resilient and Sustainable Technologies and Methods.” The welcome addresses from the University of Salento leadership made the framing clear: PLM is no longer just a technical framework for managing product information. It has become a scientific and technological paradigm that integrates design, manufacturing, operation, maintenance, and end-of-life into a continuous digital ecosystem.
Three concepts anchored almost every welcome speech:
The digital thread is a persistent flow of information connecting data, models, processes, and decisions across the full lifecycle, ensuring traceability and consistency of engineering knowledge.
The digital twin is a dynamic virtual representation that evolves alongside the physical system, combining data from design, production, and operation.
And AI is positioned as the layer that transforms lifecycle data into actionable intelligence. The formula I heard in the opening stuck with me: the digital thread provides the knowledge infrastructure, the digital twin provides the dynamic representation, and AI turns the data into intelligence. The vision presented was aligned with Industry 5.0 – technology serving not only efficiency, but sustainability, resilience, and human-centered innovation.
One more line from the University of Salento leadership worth repeating: “True growth happens when international knowledge meets local energy.” Hosting a global conference in Lecce, Italy with local companies and students in the room, was exactly that.
The Conference as an Ecosystem
Prof. Abdelaziz Bouras, chair of IFIP Working Group 5.1, made a point I liked a lot: don’t think of this as a conference, think of it as an ecosystem. Researchers, PhD students, and industry people in the same rooms, supported by Springer proceedings and journal special issues. The ecosystem includes a dedicated industry day, sessions focused on SMEs and local companies, factory visits, the long-standing best paper award, the Digital Twin contest (with open voting during the conference), and the PLM Pioneer award honoring the founding figures of our field — past recipients include Francis Bernard, co-founder of CATIA. A new special interest group on AI in PLM was also announced and it designed as a focused 1-2 year activity feeding results back to industry.
The doctoral workshop, held the day before the main conference and coordinated by Monica Rossi and Yacine Ouzrout, brought together PhD students from Brazil, Finland, South Africa, France, and Italy. Each student presented their research, received peer reviews, and got feedback from senior researchers. As Prof. Dutta put it, these students will be the leaders of tomorrow, and dedicating a special day to them matters.
Keynotes: From AI Reality Check to Aerospace Practice
The keynote program balanced academic perspective and industrial reality.
The opening keynote on day one came from Prof. Ernesto Damiani, representing the AI side of IFIP (a different technical committee — TC12), with a talk titled “Agentic AI: Transforming Hybrid Products.” It offered something valuable: an outside perspective on the “apply AI everywhere” pressure that every organization is feeling. His observation matched what I hear from manufacturing companies constantly: the push to adopt agentic AI is coming from above from CEOs, boards, and even governments, often before engineering teams have any framework for understanding what the technology actually does. His core concept was the “hybrid product”: a bundle of a physical product plus a continuous AI-driven services platform, illustrated by digital therapies in pharma where a supplement is validated through real-world evidence and an agentic platform interacts with each patient individually. The same logic, he argued, applies to automotive, avionics, and defense: increasingly, the services platform is worth more than the physical product, which becomes obsolete without its ecosystem. And his punchline resonated most with the PLM audience: the agent platform cannot be a separate thing, it has to know the product, its configuration, its packaging, its sales channels. Retrieval augmented generation over PLM data is the channel that makes this possible. Engineers, as he noted, are usually the last to know about these initiatives. That has to change.
On industry day, Danilo Cannoletta, Senior Vice President of Digital Transformation & Sustainability at Leonardo Aeronautics, delivered what I’d call a masterclass in pragmatic digital transformation. Leonardo is a company with 60,000+ employees, close to €20B in 2025 revenue, and a supply chain that is roughly 70% external.
A few things stood out to me:
Three PLM pillars are model-based definition (single 3D models carrying all tolerances and specs instead of 2D drawings), automation applied selectively where it delivers value (their “one-shot” riveting process performs drilling, countersinking, swarf removal, and rivet installation about 400,000 times per fuselage), and a human-centered smart factory bringing digitalization to the shop floor.
Vertical AI, not horizontal AI. Leonardo applies AI in specific verticals: multi-disciplinary design optimization, computer vision for fuselage completeness checks (30,000+ parts per fuselage), automated work-cycle generation, non-conformity disposition, and predictive maintenance through a customer portal. No “AI everywhere” — AI where it pays.
Data sovereignty by design. Their DaVinci-1 HPC platform is on-premises deliberately — as a strategic asset for the Italian government, Leonardo cannot expose classified data to public cloud. A useful reminder that for a large part of the industrial world, the cloud-only conversation is simply not applicable.
And maybe most importantly, his repeated statement that the main challenge is not technology but change management and culture. Their academy programs in Naples produce graduates with essentially 100% employment. Digitalization and sustainability, in his words, are “two sides of the same coin.”
The industry roundtable that followed brought together leaders from Ansaldo Energia, CNH, AvioAero, Cristal Optimisations, and Leonardo.
The final day belonged to two keynotes looking at where PLM goes next: Martin Eigner‘s PLM Pioneer talk “40 Years of PLM: From Documents to Intelligence,” and my own keynote “From CAD Files to Product Memory.” I will cover both in detail in the next two articles, but here is the short version: coming from very different starting points – Martin from 40 years of building PDM/PLM systems and academic research, me from the trajectory of CAD files, BOMs, and the data management problem – we arrived at remarkably similar conclusions about semantic data layers, graph-based architectures, AI orchestration, and the limits of the legacy PLM paradigm. When two independent lines of reasoning converge, it usually means something.
Papers, Proceedings, and What’s Next
As always, the accepted papers will be published by Springer, with special issues planned in associated journals. The scientific sessions covered an enormous range — from digital twins and MBSE to sustainability frameworks, dark data and cyber risk in PLM systems, AI-based benchmarking, and additive manufacturing process data. The Digital Twin contest brought energy (last year’s winner was an undergraduate student competing against PhDs — I love that), and the closing ceremony recognized the doctoral workshop winner working with Capgemini on the Future Combat Air System project.
People
These days, the main reason I attend conferences is to meet people. You can find the slides online and watch presentations on YouTube, but no online medium can replace human communication and personal connections.
It was wonderful to reconnect with old friends and meet new people from both industry and academia. These moments are priceless.
I also want to extend a big thank you to the University of Salento, the IFIP organizing committee, and Mariangela Lazoi for inviting me to attend the event.
What Is My Conclusion?
The PLM research community is at an interesting moment. For years, academic PLM conferences and industrial PLM practice have lived in somewhat parallel universes. What I saw in Lecce was a good sign of convergence: academics talking about industrial deployment realities, industry leaders talking about semantic models and knowledge graphs, and everyone talking about AI, but with much more nuance than a year or two ago. The conversation has shifted from “should we apply AI?” to “what data foundations does AI actually need?” and that is exactly the right question. The digital thread gives you connected data. But as I argued in my keynote, connectivity is not comprehension.
The next decade of this field will be about capturing not just what happened to a product, but why. More on that in the next two articles. Just my thoughts…
Best, Oleg
Disclaimer: I’m the co-founder and CEO of OpenBOM, an AI-native collaborative digital thread platform connecting engineers and manufacturing teams. My opinion can be unintentionally biased.
