When would you trust your ERP system to move from answering a question to carrying out the work?
Recording at SuiteWorld in Las Vegas, I spoke with Brian Chess, who leads AI product and technology for Oracle NetSuite, about what that progression looks like in practice. Brian was an early developer when the business was called NetLedger, later left to build a company, and returned with a different perspective on the needs of small and growing businesses. Our conversation focuses on how AI can help people understand their systems, complete unfamiliar tasks, and gradually delegate work without losing control of the process.
One of Brian's most useful ideas is AI acting as a teacher. He describes users asking Ask Oracle who has logged into their account, or asking it to explain a script they do not fully understand. The benefit is the answer and the explanation of how to find it. Someone who previously needed help from a colleague can begin to understand a part of the system for themselves.
He also shares an example involving NetSuite customer MCNC. When the payroll administrator was unavailable, Brian says a controller who had not been trained on the process used Ask Oracle for guidance. The outcome he describes is straightforward and meaningful: payroll ran on time. It is a customer account from the interview, rather than an independently assessed result, but it illustrates how useful AI assistance can be without requiring a dramatic productivity claim.
Financial close provides another practical example. Brian describes starting with the spreadsheet a finance team already uses, modelling that process inside NetSuite, and introducing automation in stages. I recognise the appeal of that approach from my own time supporting month-end and year-end work in IT. Teams need to understand how the process will run, especially when a mistake can affect reporting or the system of record.
We discuss how permissions apply as software gains the ability to act. Brian says Ask Oracle operates with the permissions of the user, while an agent receives an assigned role governing its access. He describes a progression from read-only analysis to limited changes as users gain confidence. Human approval remains available for transactions, particularly while teams refine instructions and account for cases they may have missed.
Business context and shared knowledge are part of the same discussion. Brian explains how written skills can capture procedures and the intended business outcome, making knowledge available beyond the colleague who usually handles the task. We also talk about accountability. He describes transaction records showing which agent acted and when, together with an activity log. Those records help establish what happened, although they should not be confused with a complete explanation of every internal model decision.
The conversation also covers whether to build AI capabilities or use those embedded in a business platform. Brian's answer depends on the role the capability plays in a company's competitive position, the expertise available, and the time it wants to invest. His view of ROI is similarly specific to the business. Comparing vendor contracts with billing records may reveal savings, but the same result does not necessarily transfer to a company with different suppliers and arrangements.
We finish with the opportunity Brian sees for smaller businesses that have fewer specialists available. AI assistance may help people undertake work they previously needed a consultant or experienced colleague to explain. The question is how to extend that access while retaining appropriate permissions, review, and responsibility.
I would love to hear your thoughts, so please share your experiences. Where would conversational ERP make the biggest difference in your business, and which actions would you still want a person to approve?
[00:00:04] What changes when your business software can explain a process, teach you how to use it and then even carry out the work? Well, this week I'm recording at Sweet World in Las Vegas and I've spoke with Brian Chess who leads AI product and technology for Oracle NetSuite. And we discussed how people move from checking an AI answer to trusting an agent with a business task.
[00:00:32] And Brian will share examples today involving payroll, financial close and the knowledge that usually lives with the person everyone usually calls. Yeah, every department has a go-to person, right? And we will also talk about permissions, human approvals and the records needed to understand which agent changed a transaction.
[00:00:53] Ultimately, for anyone wondering where a conversational ERP becomes useful, today's conversation will put the focus on everyday work and the decisions around it. So, if your finance team's favourite interface is still a spreadsheet, and I know there's more than a few of you out there, you might appreciate the starting point that Brian describes today before asking anybody to hand the process over to an AI agent at work.
[00:01:22] Lots of value in this. And with that scene set, let me introduce you to him now. So, we're recording here today at SuiteWorld in Vegas. And thank you so much for taking the time to sit down with me today. Can you tell everyone listening and watching a little about who you are and what you do? I'm Brian Chess. I run AI product and technology for Oracle NetSuite.
[00:01:46] And more or less, that means I do everything related to how do we come up with our product and how do we take it to market. So, I started at NetSuite around the turn of the century. So, I was one of the early developers of what was then named NetLedger. And then I left, started a company of my own, came back with a much greater appreciation for what small and growing businesses need. And then a few years ago, I got interested in AI.
[00:02:14] And back then, AI was a really interesting topic. And then it became a really important topic. And then it became the only topic. And that's where we are today. And so much has changed. The last 18 months to two years alone, what excites you most about everything that's being announced this week? You've been there from the ground up and seeing it come to where we are now. But what excites you?
[00:02:37] The most exciting thing this year, well, maybe I should back up and say, last year was when we announced NetSuite Next. And so, next was the big word last year. My big word this year is now. Because next has arrived. And it's not the promise of AI anymore. The promise has arrived. It's about putting it to work. And there is so much going on here at the moment.
[00:03:04] And I think it can be overwhelming for the average business leader, wherever they are in the world. So, what technical changes are needed when enterprise software moves from just AI suggestions to agents that complete work? Well, enterprises face a whole range of challenges. They've, in many cases, had humans in the loop because the loop was all humans because that's all there was was humans.
[00:03:29] And now they've got to figure out how do we have a loop that maybe involves a lot of AI and, relatively speaking, fewer humans? The nice thing is for businesses that are running on NetSuite, the data is already organized. The processes are already in place. The roles and permission system already exists. It's we've got a nice setup to make that transition easy on our customers. And we're really proud of what they can now do.
[00:03:56] And before joining me today, I had a quick look at your stuff online. You've created so much great content out there. And I saw a clip of you saying users connecting AI through NetSuite next or exploring work that they have never attempted before and feel empowered doing it. Do you have any examples just to bring that to love? One of the things we see is AI acting as a teacher.
[00:04:19] So someone says, is my account configured for, oh, let's name something. Who's logged in recently? NetSuite knows we have that information and we've also made it accessible to people if they know where to look. But they haven't always felt like that was in their power to do it. Now they can just ask, ask Oracle.
[00:04:45] And it will not only tell them the answer, it will tell them how it got the answer. And now they can do it whenever they want because it taught them how that part of the system works. So AI is a fantastic enabler because it's a great teacher. And so it can take people into doing things. I'll give you another example.
[00:05:06] An administrator frequently might not be a developer, but they can ask, how does this script in my account work? And now they're starting to learn a little bit about the code and they can talk to Ask Oracle about, well, why? What if I wanted to do this? What would that look like? And it can coach them through it. And we see that over and over again. We had an example today with our customer MCNC where all of a sudden the payroll administrator was out.
[00:05:35] Oops, everybody still wants to get paid, but we don't have payroll administrator. So now the boss of the payroll administrator, the controller, needs to step in. Well, they weren't trained on the system. How is everybody going to get paid? Ask Oracle walked her through the process. And that's immensely powerful. It really is. And I'm curious, again, for people listening and watching, which NetSuite workflows are maybe best suited for AI agents today?
[00:06:04] And where is automation probably still too risky just to rely on? Well, we are really happy with what we've done around financial close. Yeah. And this is a process that every company has. Largely, they've had to manage it on spreadsheets. Yeah. And we looked at that and we said, you know, we can automate a whole bunch of this now.
[00:06:31] But they've got their spreadsheet and they want that close to happen the way they want it to work. Why is it on a spreadsheet? Because it's being run by the finance team. And that's their weapon of choice. Yes. So how about we just ask them for their spreadsheet? Show us your spreadsheet. We'll create a model of that. And now you can do exactly what you used to do inside NetSuite and incrementally automate that process.
[00:06:58] That's really important because in order to trust the system, you don't want to take a huge leap all at once. You want to go a step at a time. And so we've enabled that. So financial close is a great example. But you look at supply chain management, a lot of people doing things there, trying to understand what's going on with their vendors. If you look at workforce management, that's another one where I gave you the example of, well, like who's logging in? Who's using this thing?
[00:07:28] There's a lot that's happening there too. So it's really all over the suite. And in a former life, I'm an IT guy and I was helping. I was the resource for month and year and period. And it's the scariest time. There's so much at stake there, isn't there? There's a lot of stress involved. So what kind of feedback have you had from departments that go through that stress month after month, year after year?
[00:07:54] Well, it is a holy grail of sorts to make that a not stressful period. Yeah. To a lot of people, and I think I'd include myself in this camp. I've been working with generative AI since generative AI came around. It still feels a little magical to me. Yes. Some of these things where you have a major problem and then it just disappears feels like magic. Yeah, it really does.
[00:08:22] And how does the platform give an AI system enough business context without exposing data or permissions it shouldn't have? I suppose that is one of the first questions you're asked as well. So we allow Ask Oracle, the conversational interface, the same permissions that the user has. So Ask Oracle, under the covers, cannot do things that that user can't do. And so that's where we start.
[00:08:50] If you're building an agent, then you need to choose those permissions. The way we did that is by modeling them the same way you would model an employee. So you give the agent a role and then it can only access those things. A pattern that we've seen from software development is software developers start skeptical. They say, I'm not sure what this thing is going to do. So maybe you start read only. You can analyze for me, but you're not going to push any buttons for me.
[00:09:19] And then pretty soon that gets old because, oh, look, it analyzes and says I have to take action. It analyzes and says I have to take action over and over and over again. And pretty soon you're like, could you just take care of that for me? OK, now we'll maybe let you write a little bit. We'll let you make some of those modifications, but still very limited. And then over time, trust it more and more. Now, part of your question there also was about, well, what about the context?
[00:09:45] Yeah, it's a really nice part of NetSuite is that we have the context because we're a suite. We're not just managing the financials. We're not just managing the warehouse. You can see all of it. That's really, really good context for AI. The other part, though, is people have always learned how to navigate these systems in particular ways. And they want to be able to impart their knowledge to the AI. That's where skills come in.
[00:10:13] So being able to write down what is the procedure, how do we use this system to get the business outcome that we want is really, really important. After somebody writes it down, then they can also share it. So now instead of just having Joe be the one everybody goes to to get a particular thing done, Joe can say, hey, here's how you do it for yourself, which Joe has been looking forward to for a very long time in many cases.
[00:10:41] Yeah, everyone's got a go to person in their organization. And another big topic over the last 12, 18 months is hallucinations, model drift. So what controls help prevent inaccurate outputs from becoming incorrect financial or even operational action? The primary source of hallucinations is when the model doesn't have the information to go on. Yeah.
[00:11:07] Instead of saying nothing, it confidently makes something up. The best antidote to hallucinations is to give the model the information it needs. Now, there's still no guarantee that the model will always do the thing that you want it to do. And for that, it's really important that you can still keep a human in that loop, particularly as you're gaining trust and refining what goes into your skills.
[00:11:35] Maybe, for instance, there's an important case that you didn't think of because it's the one that doesn't happen every day. Yeah. So you still click approve on do you want to make this transaction part of the system of record? And over time, you make sure that you've actually worked out all of those cases. So it's important, not just because the models are imperfect, but because people, we are imperfect too.
[00:12:05] Yeah. And we've got to mesh. And I think many business leads are also slightly concerned around accountability. So when several agents contribute to a workflow, how can they trace what happened and understand why a decision was made? And it's not just a black box that nobody really knows what happens in there. NetSuite has always been good about accountability for humans. Yeah.
[00:12:29] So it was a relatively small step for us to say, now we need to be able to track the agent that made this decision too. So now you've got on a transaction, which agent touched this and when, as well as the activity log for the agent. And how should a company decide whether to build its own AI capability or use an embedded one that is embedded within its own business platform? Any tips or advice there on how they should make that decision?
[00:12:59] Oh, oh, this is a tough one. Almost a podcast episode on its own. Are you, do you want to make the way you use the AI part of your business advantage? Yeah. Or would you like to take advantage of what you can easily just turn on in NetSuite? Yeah, yeah, yeah. And there's no one answer to that. It depends on what business are you in?
[00:13:28] How much in-house expertise do you have? How much time do you want to invest here? Yeah, yeah. It's a worthwhile question to be asking, but there's no one answer to it. Your agents aren't producing accurate answers because they don't have a complete semantic understanding of your data. And Denodo is solving this and solving it through semantic consistency.
[00:13:56] Through semantic consistency, your agents can start making accurate predictions in real time. So see what else Denodo can do by visiting denodo.com to learn more. There's been a mantra in business and IT for as long as we can both remember, and that is you can only improve what you measure. So which measurements show that an AI feature is saving time, improving accuracy rather than just increasing software activity? And everyone's looking busy. What should they be measuring and how can they tell that?
[00:14:26] We're getting the ROI, and we're seeing a real difference here. We have customers who, well, first of all, one of the amazing things about AI is the range of problems that you can solve with it. So we have customers who come and tell us these just incredible stories about what they've accomplished. Sometimes we already talked about payroll ran on time. Yeah. That is a win.
[00:14:48] We have other customers, though, who say, I gave the AI every one of my vendor contracts as well as the record in NetSuite of what those vendors have been billing us for. And I found all of the places where we could optimize, and then they can add up how much money they save that way, and they come up with some incredible figures. But those don't necessarily transfer.
[00:15:13] If you go to the company next door, they aren't running the same supply chain, so they don't have those same relationships. So I haven't seen yet a lot of transferable where I say, if you adopt this, you're going to save this many dollars. It isn't working like that. It's much more individualized than that. And what should attendees watch for this year's Sweet World announcements if they want to understand NetSuite's direction on AI and what comes after this?
[00:15:42] I know we are focusing on now, but any teaser you could give me on the kind of direction that you're heading on here? Oh, we have announced so much stuff. You started, though, with the number one word for where people should be looking, and that is agents. Yeah. So there's a conversational interface, Ask Oracle. We talk about it like it's ChatCPT or something akin to that, a chat interface. But it really is the very first agent that they're talking to.
[00:16:13] And when they find out that all they have to do to turn that conversation into an agent is say, hey, make this an agent for me. That's a really powerful moment. And that's what I think businesses should be looking for. And you've got access to what people are asking. You've got so many different conversations. I'm curious, when people talk to their ARP system, what kind of questions are they asking? Anything you can share there? Well, we don't go just reading conversations.
[00:16:42] For one thing, there's way too many of them. And it also could be invasive if we did it wrong. But we do analyze their conversations. And what we see is their early questions are about numbers that come straight out of reports in the system. They're asking the kinds of questions where they know the answer because they're trying to figure out, does this thing know what it's talking about? This is actually hooked up to my business. Like, what is this thing? And so they do things there. And then they start to ask for analysis.
[00:17:12] So it's things where they can figure out the answer, but it might be a little more time consuming than just looking it up. Then they graduate to, could you do a little bit of work for me so I don't have to do it? And then that last phase with the agents is about automation. How do you turn that into something where the system is working for you 24-7? And although this is a huge conference here, I want to talk for a moment.
[00:17:40] I'd like to just take a break from talking about the large enterprises and talk about those small to medium-sized enterprises. How is AI helping these businesses operate much like larger ones and give them that competitive advantage? Absolutely. So NetSuite was the first SaaS company. And when we got started with SaaS, we were telling people about how, hey, you're not going to need an IT person who knows which server to buy and how to do a remote backup because we're just going to take care of all that stuff for you.
[00:18:08] So when I look at the, every business has something to gain from AI, but when I look at where does the advantage land, there's more advantage to these smaller businesses because they have fewer specialists. So now that AI can step in and act as doing some of the stuff that you maybe would have had to either have a specialist on staff for or bring in an outside consultant to do, now the AI can help you with.
[00:18:37] And so the advantage goes to the smaller businesses in what's happening with AI. And finally, when you leave here, I mean, we talk about everything that's out there now. We're not talking about next. And you probably had so many different conversations with so many different people. What are you going to be taking away from this event? What are you going to be thinking about on that plane ride home? There's always more to do.
[00:19:01] So one of the great things about Sweet World is we get feedback from people who have been trying things or have an idea about, oh, here's what else we could do. But I'm still really obsessed with what does it take to make every NetSuite user an AI user. So how can we bring the advantages that we see to everyone? Well, I think that's a great moment to end on.
[00:19:28] And for everyone watching and listening, please pop by techtalksnetwork.com. There'll be links to everything we've talked about, including all the big announcements. And please feedback to me. Let me know what you're going to be doing differently and what you're enjoying at the moment. I'd love to hear some of the feedback from you. But more than anything, thank you for joining me today. Thank you, Neil. One of the things I loved about Brian's description of AI today was how he called it a teacher.
[00:19:54] AI can help someone understand a task that can be valuable before the system takes responsibility for completing it. And that example of a financial close also offered a starting point. Understand the process people are already using and then begin automating it gradually. Permissions and approvals remain part of that progression alongside a record of the agents involved.
[00:20:20] And I really appreciated his honesty about return on investment too. A saving found in one company's vendor contracts does not automatically become a promise for the business next door. The outcome needs to fit the work. Whether that means completing payroll on time or finding a cost that deserves attention. So a big thank you to Brian for joining me at Sweet World today. Please share your thoughts, your experiences with me.
[00:20:50] This is why I take the show on the road. This is why I record an episode every day. I want to speak with as many people as I can. And hopefully we can all learn together and all move forward together. So it's time for me to give you some homework. Which task would you first ask your ERP system to explain? And what evidence would persuade you to let it act?
[00:21:15] I don't know about you, but when November comes around, I always start planning my goals for the new year. Between me and you, it's become my little life hack to avoid making those ambitious resolutions in January. I need to abandon them a few weeks later. And regular listeners will know that I rarely promote my own services. But once a year, I open up a handful of places for people who are serious about launching their own podcast,
[00:21:43] but have never actually gotten around to it. And over the last 11 years, I've helped entrepreneurs, business leaders, and Fortune 500 companies all launch their own shows. And here is how it works. You record your conversation on Zoom, Teams, or whatever works best for you. You send me the audio, an episode title, and a short description. And I will handle everything else, from setting up the podcast host and distribution,
[00:22:10] to getting your show on Apple Podcasts, Spotify, and the many, many other podcast platforms out there. I will commit to working alongside you to help you launch your first season of 12 episodes in the bag. So you can start 2027 by ticking a season of podcasts off your bucket list. Think of me as your, not just personal podcast producer, but also your accountability partner,
[00:22:37] to make sure that you actually deliver those 12 episodes into podcast land. Now, I only offer this once a year, with limited places available. So if you want 2027 to be the year you finally launch that podcast, head over to techtalksnetwork.com. Send me a message, send me a DM, or even an audio message. I'd love to help you make it happen. Hop over to techtalksnetwork.com. You can leave me an audio message, DM, follow me on socials, just at Neil C. Hughes.
[00:23:07] But that's it. We're out of time. Time for me to hit the show floor again now. So thanks for listening as always. And I'll be back waiting in your podcast feed tomorrow with another guest. Bye for now.