Oracle NetSuite’s presentations and announcements at its recent user group meeting, SuiteWorld, focused on an initiative called NetSuite Next, a set of artificial intelligence (AI), generative AI (GenAI) and agentic AI capabilities that will be an inherent part of the NetSuite suite of applications starting in 2026. The company’s objective is to facilitate customers’ transition to this AI-centric environment, giving the option to adopt it immediately or defer to a scheduled implementation. The intention is to give customers a relatively simple path to an environment that offers an expanding set of embedded AI features and capabilities. As usual, the caveat, “the devil is in the details,” applies, since there will be many factors determining how fast and how easy the transition will be for each customer.
That noted, the key takeaway from SuiteWorld is that, for most cloud-based applications, embedded AI in all its forms will be a defining feature of business software through the rest
of the decade. ISG research asserts that by 2028, almost all providers of ERP software will have incorporated AI to reduce workloads and speed processes and reduce errors. As has been the case countless times before, the impact of new technology won’t be immediately measurable because of the time it takes for enterprises to adapt and change. There will be steady increases in productivity as software fosters a myriad of small but important cuts in the time and effort required to complete tasks and processes. It will be a major, measurable gain in organizational productivity achieved through thousands of seemingly trivial but, in aggregate, consequential cuts.
It's now trite (and obvious) to state that AI is in the process of changing how people work with business computing applications. Available categories of business software exploded in the 1990s because technology enhanced the ability of systems to mimic how people performed processes, accessed data, analyzed and understood their situation and were able to act. This marked a significant step forward for business software, but mainly in relative terms because there was still much left to be desired. AI in all its forms is in the process of addressing many of the longstanding structural issues people deal with when using applications. Especially with task-and-process-heavy categories like ERP, workloads are characterized by multiple points of friction arising from manual data entry, irregular tasks, fragmented systems and rigid workflows that resist exceptions or local needs. Cross-functional processes are complicated by inconsistent, siloed data that can produce duplicated and conflicting records. Especially with older systems, integrations are fragile, breaking easily with data schema changes. In accounting and finance, static, hard-coded rules are difficult to adjust and adapt to changes in business and market conditions, corporate structure and strategy and regulatory environment volatility.
AI is being embedded in existing business applications. Bolt-on AI use cases may be the only alternative for legacy on-premises systems, but it isn’t ideal. As a rule, NetSuite Next and similar embedded AI systems can more easily and reliably achieve the full potential of AI in business software. These systems work natively with the data schemas, workflows and specific structured and unstructured data in an individual enterprise using a unified data model that allows for explainable and auditable artificial intelligence in all its forms. Interactions are simplified by the use of natural language processing (NLP), which enables systems to understand an individual’s intention rather than having them learn and memorize commands, tasks and process execution steps. The system molds to the individual rather than the other way around. NLP addresses an important latent demand that’s been unmet until now.
NetSuite Next incorporates agentic workflows that are either included as part of the software license, developed in-house or by an implementation partner using the available visual studio or purchased from an agent marketplace. These agents take action autonomously only when it can be done responsibly and reliably. Otherwise, humans make decisions. The agents perform in the context of an individual organization’s operations and workflows while conforming to a specific set of parameters that determine how it behaves under specific conditions.
Instead of static alerts on dashboards when specific conditions or events occur, agents initiate actions when warranted to speed resolution and build organizational agility. When humans are needed to make decisions, the system provides a set of alternatives with likely first-order results to assist in the decision-making process.
Another simple example of NetSuite Next’s capabilities is dynamic dashboards created on the fly. In a meeting, presenters responding to a question or comment can immediately drill into some part of a chart with that question presented in natural language form. The result might be a tableau with several sets of data, charts representing that data, and a short narrative summarizing the answer to the question. The ability to answer questions on the fly and resolve issues quickly builds organizational agility. There’s no more “I’ll get back to you later.” For some enterprises, this capability can contribute to a culture change to one that has a bias for action based on a clearer picture of situations and options.
A major design objective of NetSuite Next is to make an enterprise’s unstructured data more accessible for anyone, but especially analysts and executives. This enables them to quickly absorb the content of unstructured documents and create recommendations for what to do next. It allows users to drag and drop specific documents into a “box” to interrogate them for reference or guidance. Another demonstrated feature is a global search and display of a summary of relevant information about a specific customer. GenAI makes it possible to infer patterns of behaviors and preferences of that organization or individual. Descriptive analytical text as well as interactive displays of charts, graphs and tables are available on a single screen, along with recommendations. Sources are just a click away, and recommendations display the statistical degree of confidence in achieving the outcome. Throughout, an NLP interface reduces frictions in business processes by enabling users to speak or type a simple command, and it will be understood in the context in which it's given. So, an instruction like “post a journal” or “create a sales order” kicks off an organization- and context-specific process managed by an agent.
Whether NetSuite will have all of what was demonstrated in general availability by the end of 2026 is less important than how quickly customers will adopt the technology. Most of what was demonstrated is challenging, but definitely feasible. However, commitment to the new software is highly dependent upon customers having the right environment to make the best use of these capabilities and the willingness to take a leap into the future. It’s likely, based on experience, that only the most technologically competent 10% or so of customers will be heavy users a year from now.
The product roadmap outlined in the keynotes is ambitious, and I wouldn’t be surprised if the depth and scope of what’s initially released has limitations. Experience has shown that the initial stages of a major technology transformation have more than a few gotchas. At the same time, we’ve seen this videotape before, and NetSuite’s product managers are sensibly on guard to anticipate as many issues as possible and have used limited initial releases to identify the most obvious that are uncovered by the earliest adopters.
It's been widely broadcast and well understood that data is the obvious Achilles heel of any AI rollout. Even so, this doesn’t mean that the early stages of adoption will be immune to limitations. Data native to the NetSuite platform likely won’t be a problem, and the more obvious non-native data elements (documents, for example) are likely to be reasonably well covered initially. The underlying Oracle technology infrastructure is solid. What’s unknown so far is the problems that users will encounter as they push the limits of the software.
We also don’t know how costs will affect adoption—in part because it’s not obvious how fast future AI compute costs will decline. Providers lost money in the initial stages of rolling out generative AI, but this was because the quality and usefulness of the results could not justify charging more. As the capabilities of language models have made stair-step advances over the past year, providers have been able to charge more for more valuable and productive results. Moreover, the company has engineered its systems to achieve lower costs, and competition generally is likely to lower compute costs as well. There are reasons to be optimistic that the AI cost-benefit frontier will be continually and rapidly pushed out over the next five years, but this remains unknown.
I recommend that all NetSuite customers make a serious assessment of NetSuite Next and the organizational readiness to adopt its capabilities. Issues that can prevent or limit its adoption should be addressed immediately as a priority. Used effectively, artificial intelligence in all of its forms can provide enterprises with competitive advantages over organizations that are slower to take advantage of the technology. For those considering acquiring a new ERP system, I strongly recommend assessing whether NetSuite meets current and future needs.
Regards,
Robert Kugel
Fill out the form to continue reading.