In 2025, agent-based artificial intelligence (AI) dominated enterprise IT, and since late 2022, it has been dominated only by generative AI (GenAI). Both have decades of experience in traditional artificial intelligence based on machine learningadvanced statistical analytics, predictive modeling and so on.
What does 2026 portend? The most likely thesis is that all forms of artificial intelligence (AI) will increasingly merge and converge, and the management needs of enterprises will determine the shape of this convergence.
The question is AI bubble The explosion in global stock markets is interesting but of little consequence to enterprise IT in user organizations. The dot-com boom and bust came and went, but the Internet survived and changed every aspect of economic, social and political life. So this is likely to happen with the hyper growth of AI.
One serious problem that agent AI Specifically, in 2025 it was about how organizations manage the identities of agents. Managing the identities of people or the workload of machines is bad enough, but autonomous AI systems have proven to be a particular challenge.
Okta research published in August 2025, pointed to the emergence and growth of new security concerns related to the proliferation of AI agents and non-human entities. Focusing on this topic and aligning the security of personal data with the security of AI agents seems to have paid off by the end of the year, as can be seen. in its third quarter earnings.
But many other vendors have turned their attention to the value of agent-based AI for cybersecurity, as with defender and attacker's point of view. ServiceNow's acquisition of AI-focused identity management platform Veza and agreement to acquire Armis, an Israeli AI security company, both announced in December 2025. talk about this trend. And this is just one of the events.
As agent AI becomes more secure, the technology is likely to flourish in 2026, delivering more of its vaunted benefits in terms of increased economic productivity and making workers' jobs are potentially more creative and enjoyable – the reason that this is a potential rather than a definite cause is that power in organizations is always necessarily contested, and it is not a matter of pure technology in the abstract sense.
The Artificial Intelligence Hype Cycle as the Advent of Internet 2.0
However, addressing cybersecurity issues that may still hinder enterprise applications is more tactical than strategic for enterprises and other organizations.
The question of how the hype around AI, which has increased dramatically with the advent of generative AI in 2022—and its evolution into agent-based AI in 2025—can be productively compared and contrasted with the development of the Internet in the late 1990s, which requires greater strategic attention. It's also a plausible guide to what will happen next in 2026.
In a strict sense, we have been here before. The answers to this comparison question can point to the likely direction of AI development in enterprises in 2026. Computer Weekly posed the question of comparing the two in technology history to a number of top IT industry executives in 2025 during the fourth-quarter conference season.
Steve Miranda, executive vice president of Oracle Application Development at Oracle, said on the eve of the Oracle AI World conference that he would not pick customer experience (CX) as the most likely area for AI development: “Back in 1998 or so, we had a slogan at Oracle: 'The Internet is changing everything.'” If you remember, e-commerce was a big focus back then.
“Well, e-commerce certainly existed, but the Internet changed music and changed the way you listen to it. It changed the way you watch—I'm not even saying watch television, per se. It changed telecommunications and meetings. It changed everything. It changed everything.”
“We believe that AI is going to change everything right now. Right now there are charismatic use cases like customer service and customer engagement. But when we say it's changing everything – and I give examples of invoice scanning, payments and customer experience [a company’s] Ledger – I don't think anyone else realizes the full impact that this will have on all aspects. We are now at the same stage as with the Internet, only it is developing faster.”
At the SAP TechEd conference in Berlin, Philipp Herzig, CTO of SAP, said that how enterprise applications evolve will be more interesting than how the infrastructure itself and large language models (LLMs) evolve. The current growth of AI, he said, is much faster than the Internet, adding: “But I still see that it is very similar to the beginning of the Internet. Back then there was also a lot of attention paid to the infrastructure plans of companies that were creating switches, routers and so on.”
“The discussion then turned to internet providers. [internet service providers] – AOL and telecommunications [companies] it brought the Internet into the home and ensured its adoption among consumers. But it's actually the value of cloud services that we all consume today in streaming services and cloud software.
“We believe that while there is a lot of discussion around hardware, data centers and GPU power, for example, that's not where the end game will be. The end game will be amazing user experiences and AI-powered software as a service that people can just turn on. We'll see a lot of new business models that come out of this.”
This was in Berlin at the beginning of November. Later that month in Barcelona, Gerrit Kazmaier, president of product and technology at Workday, said at Workday Rising EMEA: “The key thing to understand is that there are cycles in technology, and the first thing that always happens is infrastructure innovation. You saw that with Broadcom and so on, with broadband Internet access. Cloud infrastructure, which was built on the availability of bandwidth and the economics that allowed you to do business over the iInternet, in eventually created SaaS [software as a service].
“We're seeing the same cycle. We're now in the second stage. We've had a very powerful computing infrastructure built. Now we have the form factor of large language models that can be used to create a different type of computing – you might call it computational thinking.
“This is something that didn't exist before, and now software vendors who primarily solve business problems have the opportunity to embrace this and change the way business systems are built. Business processes are largely designed around the limitations of human thinking… But now, with computational reasoning, we have a chance to completely rethink.”
At the end of the year, David Richardson, vice president of AgentCore AWS, told Computer Weekly at ReInvent in Las Vegas: “I have a long-term optimism, which is important because there will be bubbles and other events, ups and downs. The beauty of the Internet was that it made possible incredibly flexible communication between a wide range of different computing capabilities and what people have built on top of what no one could ever predict.”
“I first used the Internet in 1987, seven years before I ever used the Web, and it was already exciting. Another thing that happened with the Internet is that it created a huge ecosystem for creating the Internet…[And now] there are so many different players in the ecosystem who see the opportunity to create value through innovation that it feels like we have many years ahead of us to improve the judgment and sustainability of these models.
“We'll continue to find new ways to apply them to problems that we thought we couldn't solve before, or that we couldn't solve economically. So, that's my long-term view of what's going to happen – 12 months from now, I have no idea and won't make a prediction about it.”
What does this fundamental comparison of the AI hype cycle and the birth of the Internet at the turn of the twentieth century tell us about the likely path of AI-powered business applications in 2026 and beyond? This tells us that a new stage is ready to begin. Classic AI – simply put, machine learning – has now joined generative AI and agent AI in a way that will mean enterprise applications will function more like video streaming services that algorithmically deliver content to consumers, rather than the old broadcast TV that Gen X grew up with.
Klaus Jepsen, CTO of Unit4, pointedly calls this “surrounding ERP system“, and him forecasts for the development of AI in ERP in 2026 make for interesting reading. Likewise, Stephen Webb, Director of Technology and Innovation at Capgemini UKAn article predicting 2026 in Computer Weekly makes a compelling case that “AI is no longer an enabling technology. It is quickly becoming the operational fabric of modern enterprises. We are seeing a decisive shift from single-tasking activities to autonomous, adaptive and self-optimizing systems based on multi-agent systems.”
This level of autonomy will require technologies and management processes that have been evolving since late 2022, when generative AI first arrived on the scene, but which will need to catch up. Michelle Eisenberg, Unit4's general counsel, makes interesting points about the construction AI Governance Structurebalancing innovation with discretion that goes beyond its origins with that supplier.
But make no mistake, as DXC Technology's Brad Novak, who has ridden the wave of “mobile, social, cloud and big data” and the rest, told Computer Weekly at the 2025 Boomi World Tour in London: “It's different, it's like we all woke up with 20 more IQ points.”
