“Generative AI is a usability revolution,” says Brian Chess, Oracle NetSuite’s AI chief. And he’s not just talking about writing product descriptions. Chess believes AI is about to blow open ERP for a much wider audience. No more specialist operators. No more developers required for customisation. Just business users talking to their software to get what they need.
(For the implications for market share, check out the interview with founder and CEO Evan Goldberg about how AI could help NetSuite compete directly with Xero and QuickBooks – and our opinion piece, Will generative AI kill small business software?)
In this interview with Scale100, Chess explains how embedded AI is reengineering NetSuite from the inside out. Tools like NetSuite SuiteAnalytics Assistant are already turning ERP into something a finance team can navigate and even customise without calling IT. He makes the case that built-in AI beats bolt-on every time, and that low-code, conversational interfaces will unlock ERP for everyone from founders to frontline staff.
The upshot? ERP gets simpler, faster, and more powerful – and smaller companies get a seat at the table much earlier than before.
The interview below has been edited for clarity and length. It was recorded in person at the Sydney SuiteConnect event in Sydney, Australia on 4 April, 2025.
Scale100: NetSuite is already using gen AI to fill in text fields like item descriptions. Do you think that there is any opportunity to use gen AI to rethink how ERPs work, or is generative AI always going to sit in text fields as opposed to the numbers field?
Chess: Oh, there is so much room for generative AI to contribute beyond just writing text. And I think writing text was the right place to start. Then we said, how about if we start to have the AI drive the application?
For example, are you familiar with how deep the menus are in Excel? So you know Excel can do it, but how do you tell Excel to do it? And some of these ERP systems can be the same way. You know it's super powerful, right? But how do you find the button to push to get you the thing you want? So how would you talk to something that can speak your language and have it translate that into driving the application?
Some of the stuff I was talking about this morning in the keynote – SuiteAnalytics Assistant is exactly that. We've got a tremendous amount of data in NetSuite. We've got tremendous reporting capabilities, yes. How do you find this stuff? If this is your career, you know exactly which buttons to push, and generative AI is not going to be too helpful to you.
But if you're a business owner, and you know the data is in there, then just ask the system and it tells you. So I think there's this revolution that's going to happen in terms of usability.
Scale100: What about using gen AI for coding in NetSuite?
Chess: I started at NetSuite a long time ago, when I was a software developer. They don't let me check in code into NetSuite anymore, but I do still write software on my own just to keep myself entertained.
One of the things that I've found with code assisting tools is I can write software in languages I don't know. I've never learned any Visual Basic, but I've written Visual Basic with AI assistance.
Scale100: Which app did you use for that?
Chess: I was using [Meta's Code] Llama and asking it, how do I express this in Visual Basic? It just tells me what to do. In fact, we have customers where the finance team has started to write code (using Oracle Code Assist, another gen AI tool). And it's very, very powerful for them. Finance teams ask for software all the time, but if it's a custom application they have to hire a consultant who has to do a discovery phase, and it's a very long and very expensive process.
But if they can say, “Hey AI, how do I do this?” And then get some software developers to tell them how to take that code and run it, then they can all of a sudden start doing this really powerful stuff.
Now, we haven't tried pitching it this way, but we announced Oracle Code Assist for Suitescript at SuiteWorld last year. I think one of the things we're going to find is people who don't really identify as software developers will start doing customisation of NetSuite.
And so it's going to move from the realm of just improving usability – where you say, “I know you've got the data, show me the data” – to making the system do things that you weren't going to be able to do without assistance from another human, right? Like, no amount of poking at the manual is going to turn a finance person into a coder.
Scale100: Interesting that you focus on usability. Are you talking about, say, a NetSuite GPT on your phone where the business owner can say, “Show me this” and it comes back with detailed numbers?
Chess: So I mentioned SuiteAnalytics Assistant, and that has a chat-style interface. But there's another component there. It's not just you asking questions and getting the AI to answer it. The AI is actually manipulating the application to show you what you asked for in the application. Let's say you get a report on the previous year, and you say, I want to report on the previous quarter. But if there's a button right there that says quarter instead of year, just push the button. So I don't think particularly finance people are just wanting to talk to their phones.
Scale100: But a business owner might?
Chess: Like a business business owner, yes, and I will tell you who definitely will – sales.
Scale100: In the world of Xero and QuickBooks there have been some early gen AI applications that have been doing this for about 18 months or so. They'll pull the data from the accounting software and then it will produce the report and create text explanations for the financial activity in the report.
Chess: Yes, except this is built on all of the NetSuite capability that already exists. It's not NetSuite and then there's some AI over here. It's NetSuite with AI built in, which I think is a critical part of what we're doing and sets us apart from a lot of other customers.
Scale100: Are any other competitors building in AI to the same degree?
Chess: I don't study our competition in any great detail, but here's something that I have noticed about the competition. In many cases, they're saying, here's our product and here's how much extra we'll charge you to put AI on it. And to me, that says they've got this AI thing that goes on top of the product, and that's just not the way we're thinking about it. We're thinking about the AI as part of the product, which means we don't charge extra for adding AI to it because there's no way to take the AI out from it.
Scale100: This idea about finance teams generating their own code for an ERP. Obviously you’re going to need some kind of guardrails around that, like say creating macros in Excel to know exactly what's happening and that the right thing is happening. How do you make sure that it happens in a way that is predictable and safe?
Chess: Well, I'll tell you who is the most interested in this question – the auditors. “How did this number come to be here?” And if the answer is, “I don't know, it's what the AI said”, it's not going to fly. But it's actually another reason why I like this idea of AI taking somebody's query and turning it into code, because now you've got something that you can audit.
So I don't think that we're about to enter a world in which we can do without rigorous, systematic thinking. But actually, finance teams are pretty good at that. And when they've written code, they can establish a process for how we want to trust this code. If it's just a "what if" question, maybe we don't need to worry too much about how much we trust it. If it's going to go into an audited financial statement, then we need more process around it.
But coming back to the connection to humans, humans need to do that too. In other words, why did you trust the code the developer created? Generally, you go through some sort of quality assurance process. You also do a sniff test – did it produce a sane result? Did it produce a result that bears some resemblance to the result yesterday? So we're going to need all of those same kinds of processes.
I used Excel as an analogy earlier. I think we're all very much used to relating to software that way, where you tell Excel to add up a column of numbers and it adds up the column of numbers – there's no question of that. Then we want our computers to get a little more creative. And that's kind of where AI comes in. Now a couple of years ago, Chat GPT struggled with that quite a bit. Now not so much, and I think it's going to get more and more rigorous.
Scale100: I still wouldn't trust Chat GPT to add up numbers in a column just now. You're saying that sufficiently advanced AI will have fewer problems. But it really needs to have zero problems, like Excel has zero problems when you tell it to add up the numbers.
Chess: Well, here's an experiment you can try with Chat GPT. Just start asking it to add numbers, and start making those numbers bigger and bigger and bigger. And eventually it will write a Python program to do it. And you can go audit that program. Maybe you've got to ask it to divide or multiply, and I don't know how many zeros you need to add, but eventually you can click on the little arrow and it'll show you the code that it wrote. And you can say, okay, now I know why I would trust this answer.
Scale100: So the short answer is, yes, the AI can be relied on to add up a column of numbers as well as Excel, as long as there's oversight of the code it's making to do that.
Chess: I think it is much more trustworthy than it used to be. If my life depended on it, I think I would still put my faith in Excel over Chat GPT. But again if you show me why you think this is the result – as soon as you even have a “why” I'm trusting it more.
Scale100: So you've been playing with code-writing tools. To what extent has NetSuite adopted AI for writing code internally? Have you seen these claims by Microsoft and Google about reducing the amount of time it takes them to produce code? Has that been true for you as well?
Chess: I mentioned Oracle Code Assist, and we are adopting it internally as well as working on getting it to our customers. That's for both writing Suitescript, the external language, as well as Java internally. Now, one thing that I'll say about these applications that write code is if you ask it to write a line or two, it's amazing. If you ask it to write an application from scratch, it can do some incredible things, going from nothing to something. If you drop it down in 10 million lines of code that it hasn't seen before, it struggles more.
I'll tell you who else does that? Humans. It takes time to ramp up on a really large internal code base. So I think the frontier right now on software development is not about writing an application from scratch. It's about solving a complicated problem in a really big landscape.
Scale100: So you haven't been using Oracle Code Assist yet, or you just haven’t seen any results from using it?
Chess: We have started to see results, but we have not rolled it out to absolutely everybody yet. We are still struggling with how we make it work in this really big, complex code base.
Scale100: But Microsoft and Google are both making these claims and Microsoft's code base must be one of the worst in the world given the age of some of their applications. So have they solved the problems yet?
Chess: I don't think so. You know, Microsoft has some very old, very complicated code, and also some pretty small, new stuff. Same at Google. It's not like it's one thing. And so when I look at what we're doing with Code Assist, I think it's going to be super useful to our customers first. Because they're more likely to be writing small things, they just need to adapt NetSuite for their use case. Whereas when we're looking at the full world of code inside, that's a harder problem to solve. It's an area of intense interest for me, though.
Scale100: How will AI change the way we interact with complex software like ERPs?
Chess: If you start from the idea that AI needs to be built into the foundation, it takes you to some interesting places. I talked about pricing, but here's another one. You need a user interface that now not only is going to be manipulated by a human, it's going to be manipulated by the AI at the same time. So there are all kinds of user experience ramifications to building in AI, and we've actually increased our investment in user interface because of AI. I don't think I would have predicted this a couple of years ago.
Scale100: If the upshot is that as ERP interfaces become easier to use for business users, do you think it will also have a material impact on the age at which businesses will move to ERP? Businesses typically think, “I can use QuickBooks because it's simple and easy. ERP is really scary, and I need a massive manual to use it.” But if that changes and it becomes easy to use, do you think you'll see NetSuite extending further down into the market?
Chess: Now that is a really interesting proposition. I'll tell you where I've thought about that the most. What do big businesses have that smaller businesses don't have? Bigger businesses have more specialists. That’s part of the nature of the beast, that a small company just can't afford to have specialists in a bunch of different areas.
One of the big contributors of AI is going to be helping a generalist act in a specialist role. And so it's going to make those smaller organisations more effective. Now, what I had not thought about is, does that mean that smaller organisations can now benefit? It makes sense to me. It makes sense you can say, well, if I don't have to read the big, thick manual anymore, then I can start taking advantage of this earlier. And certainly people have talked a lot about how we're going to see, you know, the sole proprietor billion-dollar business before long because of the power of AI. And it makes sense that that sole proprietor is going to want a structured way to manage that business. So I think you're on to something good there.
Scale100: Well, it's almost like it's the Wizard of Oz behind the curtain, right? If the interface becomes so simple, if you've got one person who can use it to run their florist shop and they just have absolutely no idea that it is actually NetSuite ERP because that's not the interface of that use.
Chess: So I think this ease of use is going to happen, starting now. It is happening. If you say something like, show me my sales last quarter, you can just get an answer to that. But the next phase is, why did you want to know about your sales last quarter? Maybe it's because you're trying to decide where to invest. So how about the system says, here are some ideas about where you should invest. And so every job starts to become more like an executive job, because the system is taking care of a lot of those lower level operations.
Scale100: Are you going to personify that thing? So say you have three people in the finance department, will you start replacing them with AI agents that work within NetSuite to do those roles?
Chess: So everybody's excited about agents, and maybe this is the perfect time to be excited about agents, because until we nail down what the definition is they can be everything to everyone.
Scale100: Sure.
Chess: I’m absolutely excited about it. And I think maybe the lowest common denominator for answering me, what is an agent? It's something you can talk to, but instead of just talking back, it can be pulling levers. Now what levers is it pulling? There's quite a bit of variety there. Maybe it's talking to a database, maybe it's talking to the application. Maybe it's going out and searching the web. I think that's all fair game. And we certainly see a big future.
Image credit: Supplied by Oracle NetSuite from the Sydney SuiteConnect event in Sydney on 4 April, 2025.
