At SuiteWorld 2026 in Las Vegas, Oracle NetSuite founder Evan Goldberg declared the end of ERP as we know it. His argument was simple. Employees have spent decades learning how to operate business software. Now, software should understand what employees want and help them get it done.
I spent three days at the event, attended Goldberg's keynote, heard from customers, and interviewed both Goldberg and Brian Chess, who leads AI product and technology at Oracle NetSuite.
Several announcements stood out, but I was particularly interested in what customers are already achieving, how NetSuite plans to make ERP easier to implement, and whether AI agents could eventually change how businesses pay for software.
AI as the UI
During my interview with Goldberg, I asked whether employees might eventually use ERP every day without even realizing it. "Absolutely," he replied. He described how NetSuite Next allows employees to create dashboard metrics simply by asking for them. If the results aren't quite right, they can tell Ask Oracle what needs changing.
Anyone who has worked with ERP will recognize the appeal. A finance manager who wants to understand why margins have fallen shouldn't necessarily need to know which report contains the answer or ask an administrator to build one.
At SuiteWorld, NetSuite demonstrated exactly that scenario. A CFO asked why gross margin had fallen by 12.1%. The system connected financial records with marketing campaign information and company discount policies, eventually identifying three orders where discounts exceeded the company's 12% limit without the required approval.
The demo showed how NetSuite wants employees to move from asking questions to investigating the causes and identifying who needs to act. The interesting part was the connection between information stored in different places. A margin report alone would not necessarily tell a finance director that overlapping campaign offers and sales discounts caused the problem. That is where NetSuite believes its connected business data gives its AI capabilities an advantage.
More than 13,000 customers are already connecting AI to NetSuite.
Goldberg revealed that more than 13,000 customers have adopted the NetSuite AI Connector Service. When I asked what surprised him about adoption, he described it as the fastest any feature had ever been adopted in NetSuite.
Customers are using the connector to work with NetSuite through AI services including ChatGPT and Claude. NetSuite also announced more than 45 SuiteSkills covering activities such as financial reconciliation, billing, inventory management and project operations.
One interesting aspect of the strategy is that NetSuite doesn't insist employees use its own interface. Goldberg told me customers will choose different AI models and ways of working according to their needs. NetSuite's priority is making its business information and expertise available consistently.
This raises a commercial question. If employees increasingly ask an AI assistant to complete tasks inside ERP, will businesses still need the same number of conventional software licenses? I put that question directly to Goldberg. "It certainly might," he acknowledged. NetSuite hasn't announced a change to its pricing model. But I found it interesting that Goldberg acknowledged AI could change the economics of enterprise software.
A six-month ERP implementation completed in 60 days
One of the more substantial announcements was SuiteExpress, which uses AI to assist with ERP implementation. NetSuite says the system can take meeting transcripts, documents, spreadsheets, and information from existing applications to build a knowledge base about the customer's business.
When this is combined with NetSuite's product and implementation expertise, it helps configure systems, prepare data migration, and test workflows.
During the keynote, Steve Hehl, chief operating officer at Worksmith, said his company expected its implementation to take six months. It took approximately 60 days. He also described how recording and transcribing implementation meetings helped his team avoid repeatedly explaining requirements and revisiting previous discussions.
For any business that has experienced a lengthy ERP project, that is an interesting result.
A shorter implementation means less management time spent in meetings, fewer months running parallel systems, and an earlier opportunity to benefit from the investment.
However, SuiteExpress is currently available to software industry customers in the United States and Canada. It will be interesting to see whether NetSuite can deliver similar results consistently across more complicated industries and international markets.
What happens when the payroll expert is unavailable?
My conversation with Brian Chess brought the discussion back to something every business understands: the problem of depending on one employee who knows how an important process works.
Chess shared an example from NetSuite customer MCNC, where the payroll administrator was unavailable and the controller needed to step in. The controller wasn't trained to do the task, but Ask Oracle guided her through the process.
It is a relatively ordinary business problem, which is precisely why I found it interesting. Most organizations have employees who know where everything is, understand the exceptions, and can solve problems that nobody else has documented properly. Chess believes AI can help transfer some of that knowledge to colleagues when they need it. He also described how NetSuite is approaching financial close.
Rather than telling finance teams to abandon the spreadsheets they have spent years developing, NetSuite can use those existing documents to model the process and introduce automation gradually. This lets employees keep familiar checks while reducing the manual work involved in completing them.
How much responsibility should businesses give AI agents?
Chess explained that Ask Oracle works within the permissions of the employee using it. For agents, NetSuite assigns roles much like a business assigns permissions to employees. He described a progression that begins with allowing an agent to read information and provide analysis. Once users are satisfied with the results, they may permit it to make limited changes.
NetSuite also records agent activity so businesses can identify which agent modified a transaction and when. Goldberg made a similar point during our interview. Some companies may eventually become comfortable allowing agents to complete routine tasks independently, with employees stepping in when something unusual occurs.
All this still leaves businesses deciding which actions should require approval. An incorrect dashboard calculation is inconvenient. An incorrect payment or inventory adjustment can have immediate financial consequences.
For finance and operations leaders, understanding those permissions and approval processes deserves as much attention as the automation itself.
AI savings won't be the same for every company.
I also asked Chess how businesses should measure the value they're getting from AI. He described customers comparing vendor contracts with billing records to identify where they could reduce costs.
However, he acknowledged that another company can't necessarily repeat one customer's results. Different suppliers, contract terms, and operational processes produce different opportunities. Keep this in mind when evaluating AI investments.
A company reducing its financial close from ten days to five has a measurable improvement. A distributor avoiding stock shortages can assess the effect on lost sales. A finance team identifying incorrect supplier charges can calculate the money recovered.
Chess's payroll example offers another measure. Employees getting paid on time when the usual administrator is unavailable is a meaningful result, even without an impressive productivity percentage attached. Businesses need to establish what they are trying to improve before deciding whether the technology has delivered.
What I brought home from SuiteWorld
Of everything I saw in Las Vegas, three things stood out. First, NetSuite is making a serious attempt to reduce the knowledge employees need to operate ERP. Asking for information, creating dashboards, and building workflows through conversation could remove much of the routine administration surrounding business software.
Second, SuiteExpress could significantly affect implementation costs and the work ERP consultants do. Worksmith's 60-day implementation is an encouraging example, although broader results will tell us more.
Third, the move towards autonomous agents creates decisions that business leaders cannot leave entirely to their technology teams. Someone still needs to decide what agents can do, which transactions require approval, and how errors will be investigated.
Goldberg's description of employees using ERP without realizing it stayed with me because it would represent a substantial change in how people work. Whether that happens will depend on what customers experience beyond the demonstrations.
I've included six video clips from my conversations with Evan Goldberg and Brian Chess in my latest Digital Transformation Stories newsletter, covering AI as the user interface, industry expertise, software pricing, financial close, agent permissions, and measuring ROI.
Read the newsletter and watch all six interview clips
If your business could eliminate one time-consuming ERP process, which would you choose, and how would you measure whether the change was worthwhile?