Enterprise artificial intelligence moved from a narrow focus on generative models to a broader agentic paradigm in 2025, raising new questions about identity, autonomy and control in business systems. Behind this shift sits a longer history of traditional artificial intelligence techniques, including machine learning, advanced analytics and predictive modelling, that are now beginning to combine with generative and agentic approaches. The emerging consensus for 2026 is that these strands of artificial intelligence will converge inside enterprise applications, with the pace and shape of that convergence determined by how effectively organisations implement governance frameworks.
One of the most immediate challenges highlighted in 2025 was the management of non-human identities as autonomous agents proliferated across corporate environments. Okta research published in August 2025 pointed to novel security problems associated with artificial intelligence agents and non-human identities, and its strategic focus on this area appeared to be rewarded in its Q3 earnings. Other suppliers have also acted, with ServiceNow’s acquisition of artificial intelligence first identity management platform Veza and its agreement to acquire Israeli artificial intelligence security company Armis in December 2025 signalling a broader industry drive to harden cyber security around agentic systems. As agentic artificial intelligence becomes more securable, advocates argue that it will be better able in 2026 to deliver promised gains in productivity and more creative roles for workers, although organisational power dynamics will continue to shape how benefits are distributed.
Industry executives increasingly compare the current artificial intelligence hype cycle to the early internet era, while stressing that the new wave is moving faster. Oracle’s Steve Miranda recalled the late 1990s slogan that “the internet changes everything” and argued that artificial intelligence will have a similarly pervasive impact across back office processes as well as customer experience. SAP’s Philipp Herzig and Workday’s Gerrit Kazmaier both suggested that infrastructure and large language models are only the first phase, and that long-term value will accrue in artificial intelligence native software as a service, new business models and redesigned business processes that exploit what Kazmaier calls “computational reasoning”. AWS’s David Richardson framed the current moment as the start of a many year build-out of an artificial intelligence ecosystem, while declining to make precise predictions “12 months from now”.
Looking into 2026, the article argues that classical artificial intelligence, generative artificial intelligence and agentic artificial intelligence are starting to blend so that business applications operate more like algorithmic video streaming services and less like traditional broadcast systems. Unit4 chief technology officer Claus Jepsen describes this direction as “ambient ERP”, while Capgemini’s Steven Webb contends that “artificial intelligence is no longer a supporting technology” but is becoming “the operational fabric of modern enterprises” through autonomous, adaptive and self-optimising multi-agent systems. This rising autonomy raises the stakes for governance, a theme taken up by Unit4 general counsel Michelle Eisenberg, who focuses on frameworks that balance innovation with caution. Veteran executive Brad Novak of DXC Technology captures the step change in capability by saying that the current wave of artificial intelligence is “different, it is as if we all woke up with another 20 IQ points”.
