Cloud and SaaS is around for the last 20-25 years. Product Lifecycle Management (PLM) industry came late to the SaaS journey with the early examples of hosted systems and then expansion started in the late 2010s. Salesforce.com became a foundational story of SaaS with famous “No Software” slogan by Marc Benioff. PLM software was later to the party, but the demand for streamline product development, enhanced product quality and service lifecycle management was not associated with development of new technologies, but mostly oriented on transformation of business processes.
My attention was caught by an interesting story of Sebastian Siemiatkowski speaking about their experience of shutting down Salesforce.com together with other 1,200 SaaS application. Instead they encouraged their employees to use ChatGPT and they discovered Neo4j and they discovered a beautiful world of graphs. The most interesting passage in this article was the following:
“…we decided to start consolidating; to put things together, connect our knowledge, and remove the silos. The side consequence of this was the liquidation of SaaS—not all of them, but a lot of them. And not for the license fees, even though those savings have been nice, but for the unification and standardisation of our knowledge and data”.
A few months ago, Microsoft CEO Satya Nadella made a headline-grabbing statement during a discussion on the future of business applications.
“The end of SaaS is coming.”
Check my article End of SaaS and AI agents. It wasn’t a dramatic throwaway line. Nadella was articulating a deep shift in enterprise software. The age of individual cloud applications—browser-based forms, siloed workflows, and human-driven interfaces—is beginning to dissolve. In its place, we’re seeing the rise of AI agents that operate across systems, understand business context, and deliver outcomes through reasoning and orchestration.
In other words, we’re moving from apps to agents. The interface becomes invisible. The business logic becomes dynamic. And the software? It becomes something that works for us, not something we work in.
Ironically, while this transformation signals the end of an era for most SaaS platforms, it could be a new beginning for PLM vendors—who, for the most part, never really started the SaaS journey at all.
PLM’s SaaS Journey: A Missed Opportunity Disguised as Progress
Let’s rewind. Over the last 15–20 years, the enterprise world embraced SaaS as the dominant model for delivering software. Multi-tenant platforms. Subscription pricing. Agile deployments. Fast onboarding. APIs, integrations, and user-first experiences.
But not in PLM…
Despite countless “cloud” announcements, most traditional PLM vendors remained the same. There are not many new PLM platforms and all existing platforms that exist for 20+ years lifted and shifted old, monolithic software into hosted environments—usually on AWS or Azure—and rebranded them as “SaaS.” I’m not criticizing them. I think, it was the right step for all those vendors both technically and marketing wise by providing continuity of the product lifecycle management and PLM software as it is today. ll existing product data management tools, business processes and each PLM solution that manages product data and exists for years remained the same.
What customers got was hosted legacy databases instead of scalable multi-tenant infrastructure, browser-based UIs sitting on top of existing workflows, expensive, complex implementations with layers of consulting, and configurations and customization making future upgrades complex and expensive.
As I wrote in How PLM Will Move to SaaS in the 2020s? It Is Not What You Think…, the PLM industry’s approach was cautious at best, cynical at worst. It produced the illusion of progress without the substance of transformation.
In 2025, the result is sobering: while many industries made their transition to SaaS with new platform that were built for the last 20 years, PLM is often arguing about what “PLM” means and what terminology to use for different BOM types.
And Just As PLM Starts to Catch Up… The Platform Shifts Again
Here comes the twist.
Just as manufacturing companies finally started to accept the ideas of SaaS, online services and changing their processes from documents to data, focusing on openness of API and data, looking at modern SaaS services and multi-tenant platform —the software world is evolving again.
And this time, the change is even more fundamental.
The arrival of AI agents is set to redefine how business software operates. These agents don’t need menus or workflows—they understand intent, access data, perform tasks, and improve continuously through learning. In this model, traditional applications fade into the background. The interface becomes a conversation. The architecture becomes event-driven and graph-based. The value shifts from “tools” to “outcomes.”
And here’s the harsh reality: AI agents won’t run on legacy platforms. They need data that is accessible, structured, and connected. They require semantic understanding, real-time feedback, and the ability to work across domains. The need to be agile and nimble integrated in a new type of workflows. This is not something a file vault wrapped in a web UI can deliver.
PLM’s Current Role: The Smart Vault That Stuck In Engineering Departments
At the recent CIMdata Market & Industry Forum, I shared one of the most repeated insights I’ve heard from manufacturing companies:
“PLM is still viewed primarily as engineering document management.”
PLM software and its implementation is focusing on engineering data, change management, and extension of what we know today as a “PLM technology”. PLM introduced themself as an platforms to develop innovative products and product lifecycle, but mostly focused on how to hold an up to date data for engineering PLM processes. It was not a connected enterprise platform. It was not a holistic product lfiecycle management (PLM) from idea to retirement. Not the foundation of digital transformation. Not even a system of record for products data that can be used across the company. Just a vault—for CAD files, revisions, ECOs, and often a set of derivative files – STEPs, PDFs, DXFs.
And if we’re not careful, that’s where PLM will stay—as a back-office tool for engineering documentation, increasingly sidelined as AI takes over planning, sourcing, and collaboration tasks across the enterprise.
And because of the well developed modern infrastructure, companies can be find a very affordable new systems to host those “released” files across the company instead of relying on complex “engineering PLM software oriented databases”.
The Existential Risk: Becoming the Forgotten Backend
The risk isn’t that PLM systems will go away. It’s that they’ll become irrelevant. As AI reshapes how companies operate—from supply chains to production to service—PLM may remain stuck in the basement, managing CAD files and PDM workflows, while the real digital intelligence is built somewhere else.
If PLM systems don’t evolve to become part of the AI and automation layer, they risk being abstracted out of the conversation entirely. Just one more silo to plug into an integration bus—eventually minimized, ignored, or replaced.
The Real Opportunity: PLM as the Intelligence Layer for the Digital Thread
Despite the challenges, there’s a powerful opportunity in front of the PLM industry—perhaps the most significant since its inception. But it requires a fundamental rethinking of what PLM is and what it can become.
If PLM vendors embrace this moment, they can build something entirely new:
A platform that is no longer limited to document control or part structures, but instead becomes the semantic and operational backbone of the digital enterprise.
That means shifting away from rigid schemas and siloed workflows, and instead building open, graph-based data models that reflect the true relationships between systems, parts, people, and processes. It means enabling federated data access that spans design, sourcing, compliance, manufacturing, and service—not trying to centralize it all into one stack. And it means exposing that data and logic in ways that AI agents can understand, reason with, and act on.
I explored a glimpse of this in my article, AI-Powered CCB Agent: Transforming Engineering Change Management with Intelligent Automation. In it, I discussed how AI agents could not only assist with but also orchestrate complex workflows like engineering change—contextually analyzing data, suggesting actions, enforcing rules, and engaging the right stakeholders. This is not science fiction; it’s a realistic direction for where PLM systems can—and must—go.
The cloud was a stepping stone.
SaaS was a warm-up.
AI is the main event.
And the real value of PLM isn’t in managing CAD files.
It’s in providing the intelligence layer for how products are designed, built, delivered, and evolved.
What is my conclusion?
Satya Nadella’s announcement about “the end of SaaS” is more than a headline—it’s a clear signal that the software industry is entering a new phase, where intelligent agents—not traditional apps—will lead the way.
For most enterprise software, this marks a transition.
For PLM, it exposes a gap that’s been widening for two decades.
While other industries moved into true SaaS platforms, PLM clung to legacy architectures and superficial cloud rebrands. As a result, many companies still view PLM as little more than a document management system for engineers.
Now, with AI and intelligent agents on the horizon, there’s no time left to waste.
If PLM wants to stay relevant, it must do more than catch up—it must leap ahead. That means embracing modern data models, graph-based architecture, federated access, and AI orchestration.
The digital thread is no longer a vision—it’s a necessity. And PLM can either become the platform that enables it… or the bottleneck that holds it back.
The future doesn’t care about your applications – it cares about the data, what your system knows, product quality, supply chain management, connection to contract manufacturers, business value of products and services, status of product development projects, optimization of a product development process, connecting between business systems.
The core of future product lifecycle system is a product knowledge graph that connects siloes. The outcome of this process is unification of information, consolidation of knowledge and creating a model that orchestrate other tools that are open and connected. The important element of this transformation is how fast we can build a core model that can learn and how well it connects with other systems.
Just my thoughts…
Best, Oleg
Disclaimer: I’m the co-founder and CEO of OpenBOM, a digital-thread platform providing cloud-native collaborative services including PDM, PLM, and ERP capabilities. With extensive experience in federated CAD-PDM and PLM architecture, I’m advocates for agile, open product models and cloud technologies in manufacturing. My opinion can be unintentionally biased
