AI is redefining how enterprises design, deploy, and interact with software, leaving traditional digital experience platforms struggling to stay relevant. As the boundaries between developer, user, and agent blur, the idea of a static platform is being replaced by something faster, smarter, and more adaptive.
A new generation of artificial intelligence is challenging the very foundations of enterprise software. As generative models begin to generate front-end interfaces, workflows, and even architecture-level code, the question is no longer how platforms evolve but whether they remain necessary at all. A recent IQM white paper explored the technological upheaval of next-gen enterprise systems.
The current generation of low-code and no-code digital experience platforms (DXPs) helped democratise software development, reducing technical barriers and accelerating delivery. But their value was always constrained by what they could not do. These platforms offered modular, interoperable components, a form of pre-configured configuration. Businesses often had to adapt their workflows to suit the platform rather than the other way around.
“Low-code democratised software in the sense that it allowed non-technical users to build out some kinds of functionality, but it was very much a Hobson’s choice,” Mike MacAuley, General Manager at Liferay, explains. “You are stuck with a minimal number of options. You are not actually building it out. You are taking pre-built components and sticking them together like you are a five-year-old with Lego.”
From disposal to design
That rigidity is disappearing fast. With AI-generated code, the cost and complexity of custom development are collapsing. Platforms once promised to get you 80 per cent of the way toward a solution. Now, enterprises can build disposable applications for a single use case, launch them in days, and discard them weeks later.
“You are running an event next week, and you need a website, a customer portal, and a ticketing system that is going to last for two or three weeks,” MacAuley says. “You can just shut it down and throw it away at the end and start again the next time. That is radically different. Why would you need a platform if AI can generate software at that level of speed and specificity?”
The implications go far beyond cost. As large language models (LLMs) improve, AI is beginning to outperform humans in backend coding, sales engineering, and front-end prototyping. Even if current models are not yet enterprise-ready for safety-critical domains, the pace of improvement is remarkable.
According to MacAuley, there is little reason to think this momentum will stall. “Anybody who is comforted by the level and quality of what AI can do now is living in a dream world,” he says. “Think about where it is going to be in 18 months from where it was 18 months ago. Currently, it makes mistakes, and there are use cases where it cannot be used; however, the improvement rate is incredible. There is a pretty good chance it will be good enough for a lot more soon.”
The emergence of custom, use-case-specific software is eroding the very premise of a platform. Enterprises do not want standardised building blocks; they want code written in their language, tailored to their needs, immediately. That is what generative AI delivers. The power is not just in automation but in alignment: software that reflects the business, not just enables it.
Agents not users
As code becomes commoditised, the user interface is evolving just as rapidly. Agentic AI, enabled by frameworks such as Anthropic’s MCP or Google’s A2A, is reshaping how enterprises approach experience. The future is not just mobile-first or cloud-first; it’s both. It is conversation-first.
“If you are not using the screen, you are using a conversational interface,” MacAuley says. “You have something very similar to a personal assistant working on your behalf, and you can instruct that. The way that interface connects with your services will be key to whether people can access your offer through agentic AI. That is not something most DXPs are set up to deliver today.”
The shift mirrors earlier digital transitions. For MacAuley, the rise of agentic AI feels like 2009 all over again, when mobile suddenly became a critical digital channel. Audio-driven, semi-autonomous interfaces will demand new integration standards, new forms of service orchestration, and entirely new delivery pipelines.
“Agentic AI is a new endpoint for your digital estate,” he says. “You have kiosks, desktops, mobiles, and now this. If your offer is not accessible through this kind of channel, you will lose customers. You need to get on board with that as soon as you can.”
This transition will not only alter user behaviour, but it will also fragment the notion of a digital experience itself. Customers may no longer interact directly with an app or site.
Instead, they will instruct AI agents to navigate those layers on their behalf. This will force businesses to rethink not just interfaces but also how service logic and content are exposed programmatically.
Developers under pressure
Ironically, this new flexibility poses a threat to the very talent it depends on. Junior developers risk being bypassed by AI-generated code, leaving a future skills gap that no one has yet solved.
“To have senior engineers, you have to have engineers. To have engineers, you must first have junior engineers. If you replace all the junior engineers, then in a few years, you will not have any senior engineers,” MacAuley says. “At the same time, developers will still be required to supervise, debug, and guide AI output. Generative systems may accelerate productivity, but they cannot yet be trusted to operate unsupervised.
“You have to understand what it is doing, what each aspect of the code is fulfilling, and where it might go wrong. With a little bit of guidance, it produces something decent, but it still needs a human in the loop.”
The result is not a reduced workforce but a different one. Future developers will act less like coders and more like team leads, supervising AI collaborators and iterating at a faster pace than ever before. For some, that will be empowering. For others, it will be exhausting.
It is not only junior talent under threat. Even mid-level developers risk being outpaced by agents that can generate boilerplate, test functionality, and recommend architecture, all within seconds. Human roles are evolving from builders to validators, with higher cognitive demands and faster release cycles. Without organisational change, burnout will follow.
The edge of the platform
This new reality also raises questions about the role and relevance of digital platforms themselves. As AI systems become more specialised and fast-moving, few vendors will be able to keep up. “Our strategy is to remain as open as possible to a rapidly evolving ecosystem,” MacAuley explains. “Trying to compete by investing heavily in AI ourselves makes no sense. The ecosystem is evolving so rapidly that it is impossible to predict where it will be in a year. So staying open is key.”
That openness will extend not just to AI services but to how digital channels are defined. The ability to integrate, not just generate, will be the new measure of platform value. The winners will be those who understand what is coming and adapt accordingly.
“The structure of the industry is going to be massively disrupted,” MacAuley concludes. “People who get on board with that change and take advantage of it will be the winners. Those who are slow to adapt will be the ones who lose. It really is that stark.”
From disposable code to agentic interfaces, the old boundaries between vendor, developer, and enterprise are dissolving. The next generation of digital platforms may not be platforms at all, but fluid, intelligent systems defined by their adaptability, rather than their architecture.
