Automation in the Age of AI: The Stonebranch Vision

Hello everyone, and welcome to Stonebranch UserVerse Online 2026. I'm Lauren Tancini, and I'm excited to kick off the first session of our four-part virtual series. Earlier this year, Stonebranch UserVerse brought the community together in person, and now we're continuing these conversations online where we can explore what's next for enterprise automation with particular focus on automation in the age of AI. As AI becomes more embedded across the enterprise, organizations need a way to connect AI driven work with applications, data, and mission-critical processes that are already running the business while maintaining control, visibility, and reliability. That brings us to today's keynote, Automation in the Age of AI, the Stonebranch Vision. We'll explore Stonebranch's vision for the future of enterprise automation, give you a first look at Universal Automation Center 8.1, including new AI capabilities, Universal Portal enhancements, new integrations, and more. We have lots to cover, so let's go ahead and get started. We're going to kick things off with an industry and Stonebranch update. I'm excited to welcome Peter Baljet, Chief Technology Officer, Gwyn Clay, Chief Product Officer, and Scott Davis, Chief Marketing Officer. Let's go ahead and get started. Peter, Gwyn, and Scott, welcome. Hey, thank you, Lauren. I'd like to extend another warm welcome to everybody across our UserVerse. In one form or another, we've been doing this series for nearly seven years now, and I got to say, this is the year that I am the most excited about, not because we haven't given great updates or product demos in the past, but because this is the first year we're in the middle of a sizable industry shift. Of course, that industry shift I'm talking about is the entry into what we're calling the AI era. Today, you'll learn how AI has evolved Stonebranch Universal Automation Center. You'll see some amazing product demos, many of which are designed to help you navigate this new AI paradigm. With that said, let's go ahead and dive in with our first segment. Peter, Gwyn, welcome. Let's get started. Morning. Hey guys. All right, so let's start off with some questions. The first question I'm going to ask you is how would you describe Stonebranch's momentum right now? Peter, why don't we start with you? Yeah, we've been really down this road for about five years about doing hybrid IT, and we've had tremendous adoption by a number of customers that we get every year. We've got large customers, Fortune 500 customers, as well as smaller customers. And what's exciting to see is that we try to focus on supporting a number of different use cases out there, and we see a wide variety of that, everything from a traditional mainframe, but then you have cloud, DevOps infrastructure, and customers do a combination of those, right? And so seeing that all work together seamlessly on our platform is I think one of the key items. I think one of the things I'm most excited about is seeing that adoption across the customers. This world used to be one where it was largely mainframe, largely on-prem, and this fast-forward leap that we've gone through over the last five years has been all about the cloud, but there's something new that's coming, right? What are we doing now? Now, we've seen this pick up really this year, and that is AI. And companies are now looking... They started doing AI at the front office, helping people do their daily jobs, but now it's shifting back into the IT landscape and looking at how AI is going to be incorporated into systems, infrastructure, automation such as we do. And so today we're going to talk a lot about and see demos around how that is being incorporated into the UAC platform. Yeah. So in addition to all those features, the product momentum we have, there was a big announcement recently and it had to do with the Gartner Magic Quadrant, which is this is the third year the Gartner Magic Quadrant's come out. Gwyn, how did we rank in this? We're a leader for the third straight year, and if you look when the first one got published over the past three releases, we've made significant momentum each and every year and are now one of the top leaders within the quadrant. On top of that, they've also published a critical capabilities report, and in that critical capabilities report, they have 12 categories, and we had a score of excellent in 12 of the 12 categories. So excellent is the highest rank you can get? And that is the highest rank you get, correct. Yeah, it's awesome. But the SOAP Magic Quadrant changed slightly this year, was a little bit focused on AI. Can you tell us a little about that? Yeah, so the last couple of years, AI has been more visible within the Magic Quadrant. In the first year, it was around how AI was used within our products. This year, there was a much greater emphasis on orchestrating AI, right? So how do you integrate with the AI ecosystem of LLMs, models, the specialized AI agent platforms, as well as now AI gateways, which are gaining more visibility within enterprises? Yeah, AI isn't regulated just down to building an agent, it's really an entire ecosystem, and that ecosystem needs to be orchestrated at the end of the day, and that's a lot of what we'll be talking about today. Now, I'd like to take a moment to pause here and talk about this series as well as the broader UserVerse series. Now, like I mentioned, we've been doing this virtual series for nearly seven years, and two years ago, we started doing in-person sessions that were run across the globe, a roadshow that we call UserVerse In Person. Now, this past year in 2026, we had a wonderful series, very exciting. Lots of people came. It was the greatest one that we've done to date, and I want to really quickly call out and thank our hosts of those series. So as many of you know, we typically partner up with one of our customers, and this year we partnered with ING, Munich RE, and JM Family to host these events. So big thank you and shout out to the people that work with us at those host sites. It really was a phenomenal event, so thank you. Next year, we're going to be doing the same thing. So in the April and May timeframe, we're going to have a session in Germany, we're going to have a session in the US, and a session in the broader EMEA area. So get ready, start planning for that, get your budget set, whatever you need to do to show up because it's a lot of fun and you learn a ton. I would like to have a quick call out. We are looking for companies to host these sessions, so customers, partners that would like to jump in and have these sessions at their location. So if you're interested in doing that, please, there's a box that's going to pop up in a second, just click there and let us know and we'll give you a call and we'll work it out. So thank you for that. Thank you for the series and the sessions and all the great interaction we've had with you over time. All right, now let's shift back to some more questions. First, let's dive into some of the industry and marketing updates. Now, there's a lot that's happened in automation orchestration, some of it AI like we've talked about, but some of it not. Why don't we start with the non-AI stuff? Peter, can you tell us some of the market shifts that have happened? Yeah, I mean, we really try to focus and listen to the customers on what their needs are coming from. Nils' group does a great job at really reacting to what customers need today because he can deliver things very quickly. Some of those things have been container-based workloads. We see that shift of going from your traditional on-premise to maybe container based that can be shifted around to the cloud or onto Kubernetes type of clusters. Data movement to the cloud, I think data lakes are huge. We see people adopting a lot of these data lakes that are on Azure, on AWS. And because of that, we've really looked at how we manage data transfers. We have UDMG, which we'll talk about. There's a lot of functionality that's coming there, and also moving data into the cloud or around the cloud. That's becoming very important because of how much data is trying to be managed now. And so if I had to use a hockey analogy, a lot of what you just talked about is where the puck is, right? We are helping our customers. They're asking us for specific pieces about what's happening right now. That's where the puck is, but we're also building towards where the puck's heading, and where the puck's heading is more about AI. Gwyn, can you talk us through what we've seen related to AI since maybe 2024? It's quite remarkable how quickly our customers are learning about AI, and you've seen it accelerate over the past 18 months, where at first they were just asking questions, now they're really asking about capabilities and specific capabilities. How we've seen this play out is that at first they were wanting to see how we were going to be enriching our product with AI, a lot of the work we've done around Robi AI , and now where they're really heading towards is AI agents, and they're building lots of agents and they're building them everywhere. There's not going to be one master platform where all AI agents reside. They're going to be building them within their business applications, within specialized platforms, within their AI tools, and it's really at the point where when they need to start having these agents interact with each other across business applications, across these specialized platforms, across their AI landscape, is really where these are going to break out and they're going to need to be managed properly. That's where a product like UAC comes in that provides security, the governance, the structure, the tracking, to be able to ensure that these processes that are running across their AI landscape and also within their existing automation, the deterministic automation that has been running successfully for years that is being enriched with AI, that's where we're going to really step in and provide that solid foundation for the future for them. Yeah, and we've been talking about it in terms of governed workflows, right? You're going to have agents, they're going to be built everywhere, but how do you control those agents? How do you make sure they run reliably? How do you make sure you have the tracing, the enterprise scalability related to them? And it really requires orchestration, right? I think the key point here is that it's gotten really easy to build AI agents, but it's incredibly difficult to manage and control those agents on an enterprise scale and in a way that you can understand what they're doing and tell auditors what they've done and all those sorts of things, right? So let's break it up now for the audience about what we've built and where we've built it. So there's sort of two sides of AI for us. The first is in the product. Maybe, Gwyn, can you talk to us a little about what's in the product, what we've done, and maybe hint at a little bit of what you're going to be showing later on? Sure, Scott. So in the spring this year, we released Robi AI, and that was an AI service that runs alongside UAC, and in the future, we'll run alongside other products like UDMG as well. This is a separate solution that allows customers to also safeguard their data. If they run on-premise, we can run on premise. They're not forced to move to SaaS. If they run in SaaS, Robi runs in SaaS with them. They can bring their own model. We don't train off of their data, and we also don't keep any of their data as well. So how this has played out in terms of functionality has been around really the AI assistant capability, allowing them to build automation, allowing them to search for answers, and to have assistance on some of the things that they may not do on a daily basis, and then also to be able to issue commands and also do root cause analysis. Yeah. And you mentioned it earlier, but I want to reemphasize it, Robi, Robi AI is what we're calling this feature inside of UAC, but I think that's the important part. It's a feature today that's inside UAC and helps power, AI power UAC functionality, and that's been really important. We put a lot of time to that. That's mostly where, going back to an earlier question, I think that most of our customers and our prospects were asking about that AI powered functionality inside the tool, but 2026 marks a total shift, as Peter talked about. Now it's turned into, how are we going to help people orchestrate AI across the business outside of UAC? And Peter, that's the other side of this. So can you talk to us a little bit about this second side of what we do related to orchestrating AI? Yeah. It's actually using AI agents within execution. So if you think of your traditional automation workflow, which typically has different steps, those steps are now AI based. Maybe it's more of an AI prompt that you're giving, or it could be an agent that does the agentic loop, has tool calling, can reach out to other systems. Since agents and tools and MCPs are being put in all different sorts of applications and infrastructure, that gives the ability to put these in workflows and for them to do functionality that wasn't possible before. What's important is that those aren't completely AI workflows. They're running along with your traditional type of tasks and what we call deterministic type of task. And that's a term I think that we use now, but I don't know if people out there have heard of that, but we kind of use deterministic type of task versus AI type of task with the nature of AI being more probabilistic, right? Customers are getting exposed to that more and more, and we've seen the questions and the ideas increase just over the past six to 12 months. Yeah. Yeah. Like you talked about, 2026 has been the year of early adopters coming in and wanting to see AI inside of workflows or how we integrate with external agent builders or all those sorts of things. So I think the big point here is that there's really two sides of AI for us. The first is what we've built into UAC, Robi AI, and the second is what we can orchestrate with UAC. Now, let's talk a little bit about going back to what's being built inside UAC. There's some really cool features that have just come out, and these features are all about using this AI power to help build automation faster, build automation easier. Gwyn, can you tell us about those two main features, and the ones I'm talking about are Robi AI Workflow Builder and the AI Integration Builder? Yeah, these are incredible capabilities that a few years ago would've been inconceivable. Like magic. Like magic, exactly. If you look at the Workflow Builder from Robi AI, you can interact with it, and through a series of conversations with Robi, Robi AI, you're able to create a workflow and also the tasks associated with it as well. So you put a natural language into Robi as a chatbot and a workflow appears. Correct. And you can see it, you type in a conversational panel on the right and on the left you see your automation appear. And we're going to see that today? We're going to see that today. It's part of the demos. And it's fantastic because there's the interactive mode, but there's also the mode where you can provide a large prompt or a markdown file, and then you can give Robi AI the entire file at once, and it will go through and do the analysis and build the workflow for you and the tasks, and then it'll come back with questions, and that's kind of the next stage of it. And as customers get more and more comfortable with it, they'll likely work from a script approach of asking Robi AI what to build, and then Robi AI will just act as the assistant asking questions. Now, you may want to go in later and look at the definition and make some fine-tuning and some changes, but it's really there and it's there today. So as a dumb marketing guy, I've fooled around with these guys, and I can build workflows now. I mean, I promise you, I'm not technical, but I know how to write English. And I think at a higher level, this is the transformation of how we interact with software, right? It used to be forms and fields, and if you've used UAC, it has a tremendous amount of functionality, so it's got a lot of configuration. Now, you don't have to know all of that configuration because you can give it the intent of what you're trying to do, and Robi understands how to configure that as opposed to you having to go and help and search through the forms. So we're excited because we think it gives you the ability to create and also understand how to use the product a lot better. Okay. Now let's flip, because there's another piece. There's the AI Integration Builder. And as exciting as the Workflow Builder is, the AI Integration Builder may be even more exciting. So we've always had a very lightweight ability to build integrations, right? We've had the concept of an extension, which is where you can build an integration and then plug it into UAC, and you would typically be able to connect to a new business application or a new cloud product, and within a matter of a few days, you could build an integration. Now with the AI Workflow Builder, it's a matter of minutes. Minutes? Minutes, half an hour, hour, you've got an integration not only built, but it also will deploy into your UAC environment for you. And with this, the nice thing about it is it's a no code approach. You provided a description of what you would like to do, and it will go and build the integration for you, and it will also then install it for you, and it's available for you to test with. Yeah, and this isn't just, we call it the Integration Builder, but in essence, it could also write scripts, programs to do certain functions. If you have a, let's say, an ETL task, you don't maybe want to go use a ETL tool, you just want to write a Python script to go change the data. So it can be used for a number of different purposes also outside of just trying to integrate into other systems. And the nice thing is you don't need to be a developer to do it. The extensions used to be written in Python mostly. Now, today, you just basically will have a text file describing what you want to build. It will go off, it will do the analysis, and then build the integration for you. That is incredible. As a layman again, I've seen the demos that were created for the website, and one of the things that struck me, just because I know this is the piece developers hate, is that there's a button you can hit that says, "Create documentation." It's silly, I know, but every developer out there is thinking, "I hate documentation. Why do they..." This just creates a documentation for you too, which is just amazing, right? There's sort of a point that I want to make here. So we're going to be demoing both these obviously today. You're going to see them both running live in action. The Workflow Builder is coming out, had just came out actually in our latest release, 8.1. The Integration Builder is something that is not generally available. It is early access. So what we're looking for are customers who are interested in trying it out. If you're interested, reach out to your account exec or there's a survey that's going to pop up at the end of this session, and there'll be a question asking, "Do you want to try out the early access for the AI Integration Builder? Check yes, and we'll be back in touch with you." The trade-off there is you get early access. We just want some feedback, want to make sure that it's ready for general availability. It's largely built on AI and our domain expertise, and not everything always works exactly the way we want it to with AI the first time around, so we want to make sure we get it right before we roll it out to everybody, but it's really good. And we've been using it internally for how long? Half a year? Yeah, we've used it for a while. The internal teams, this is how they develop in iterations and extensions to the product. Yeah. So when prospects come to us, we used to prototype in days. Now we prototype in minutes, minutes. And when we do the demo here in just a little bit, one of the cool things is that we're going to demo this during the course of this session, and you guys will see it live. Very exciting. Cool. All right, so let's go to another question. Stonebranch has made it easier and faster to build automation, but in terms of automating, orchestrating third-party agents, things that sit outside of UAC, where are people building those agents right now? What's happening in the world? What do we need to account for when we're thinking about orchestration? Mm-hmm. I mean, we're seeing, I think, and this is an area that's changing very quickly. We're seeing agent platforms pop up, but also model gateways pop up, and a combination of both. And so what we're doing is we've realized that agents are going to run in all different places. It could be inside of your ERP system, but it could be in a dedicated platform. So we're building the integrations necessary to be able to reach into these platforms, make it part of a workflow, be able to configure where you want your model to be. So we can go to... I think one of the ones we've heard several customers talk about is the Kong platform. Yup. Agent is going to- AI Gateway. Yup. Typically, customers want to go to the cloud provider instances that they have versus the public ones. So we want to stay very flexible in how you configure and how you integrate in AI functionality so that as that world changes a lot, you still keep your business processes running within our system and they are abstracted from all those changes. Yeah, so the agent or agentic process may run in an agentic system, could be LangChain, could be LangFlow, could be n8n, could be any of the tools that people are using to build agents or applications like SAP or ServiceNow. They're really big on building agents right now. We've built the interoperability to reach out to these places. We don't expect people to stop building there. That's not practical. I mean, at the end of the day, agentic creation has become democratized, right? But that democratization creates mass confusion, mass problems where you're not able to connect and see the holistic view of all the automation that's happening. Because at the end of the day, these agents, they're automators, they're automating things, so they need to be controlled in the same way that the rest of your automation is. Right. All right, so when I think about, and I mentioned LangChain and LangFlow and n8n, why wouldn't you just run all orchestration through those tools? What's the problem there? Do you want to take that one, Gwyn? Well, I think a lot of these platforms are very development centric tools. When you're starting to look at the qualities of service that you need to run these at a very high scale, and then also when you start looking at integrating those agents across business processes that span the last 20, 30, 50 years of IT, you really need to have a platform that knows how to connect to everything and connect everything in a very scalable, safe, reliable way that provides you with the visibility, the auditability, and the logging required to ensure that you are getting the level of success necessary. Your internal IT auditors demand it, right? The compliance initiatives demand it, and we're the ideal platform for people to run those agents that are more process oriented, transaction oriented, workflow oriented, that span outside of the platform that they run on today. Yeah, and I think what we're generally seeing in the market is most of what's being built in the agent orchestrators or the agent frameworks, they're predominantly prototypes, they're experiments. When people put them into production, they may run it for a while, but what I keep hearing is that they eventually fail, and then you don't know that they failed because you don't have the control mechanisms, the SLA processes, the observability to figure out what went wrong. So what we're really getting to here and what we're building up to is something that will help everybody, all of our users across our UserVerse really make a change there. And it's interesting because you're even seeing in some geographies, compliance requirements come down that says, "Hey, if you're doing some sort of critical function, whether it be financial or supply chain related, that there should be some level of deterministic that you're making sure that things happen the way you want them to." Just traceability, just observability, right? You have to be able to prove what happened. Yeah. So it's going to be interesting, but that's why, again, you can't just run these agents. You really have to have what we call a control plane of where they actually run. You're jumping ahead, Peter. You jumping ahead, but we'll get there. We'll get there. Let's talk about all these agents running in silos for a moment. So let's say you have agents running in all these places, what are the problems that having silos of agents, of really anything automation running by itself, why would you not want that? And this isn't something just with agents. I think we've seen this happen in the past when new technology comes up. Everybody started using the cloud, but then they basically had to integrate that back into all of their on-prem systems, and you needed to have tools that help you do that, like UAC. And so I think we see the same pattern here with agents, is that there's a lot of these things popping up, but then it has to come back to, "Hey, these things run our business, and therefore we have to be very structured in how we actually manage them and deploy them." Yeah. So it's great to build the agents someplace, but it follows the same pattern we saw in cloud where people were building automation all over the place in the cloud, and there's another pattern that says it needs to be pulled back to be operationalized, make it comfortable for the I&O teams, the compliance teams, the governance teams. And again, like we said earlier, these agents aren't doing the whole process, the whole picture, right? You're going to have a combination of your traditional automation with agents involved. So you've got to be able to combine that in the same way with the same governance reliability that you have today. Yeah, exactly. Anything to add? I think that pretty much covers it all. All right, cool. All right, so let's... much covers it up. All right, cool. All right, so let's recap just a bit. Enterprises already have multiple automation tools. Now, they have agentic automation that's also being built in silos. These silos are disconnected, ungoverned. Is there anything that our customers can do about it? And it's a leading question to talk about UAC. How can UAC be used? And this is where we want to introduce a new solution within UAC, and what we're calling it is the Hybrid Orchestration Control Plane. And this Hybrid Orchestration Control Plane isn't a replacement. It's not like a whole new UAC or anything else. It's a new set of functionality that allows you to run agentic, deterministic, or a combination of, hybrid, agent and deterministic workflows together. Now, we like to talk about this in terms of meta-orchestration. We've introduced that term a number of times over the years. Meta-orchestration is just we're going to orchestrate the orchestrators, and we saw it play out in data pipelines where we orchestrate airflow or other open-source tools. We saw it play out in the cloud where we orchestrate AWS Lambda or other automation tools within cloud service providers. Is there a difference here? For our customers, is it just adding another workload? Is this adding another thing that we're orchestrating? The UAC platform, it really is adding another capability onto the platform. It is doing what we've always done, but now being able to do that with AI agents, AI tasks, which we'll see today. By doing that, we're orchestrating those two like we've done a lot of other tools. And then it comes into this common layer that does all of the guardrails and control that customers expect. That's awesome. The other thing is that by those agents leveraging UAC as well. So we talk about us orchestrating agents, but there's also the other way around where agents start leveraging the Universal Automation Center to get access to workflows and then also to help interacting with the deterministic automation. Perhaps it's from cloud or data orchestration or SAP or even mainframe to be able to open up a new world for the agents themselves. Yeah. And this goes back to that governed workflow concept I was talking about earlier. What we're really saying is an agent can trigger a governed workflow in UAC and then receive the output that it was completed, and that's part of the tool call that the agent made, right? Yeah. That's awesome. All right. So one point of clarification around terminology. Now, we've often talked about hybrid IT. If you go to the homepage of our website right now, it says Hybrid IT Automation. That is historically referred to the fact that UAC is so purpose-built to automate things across a hybrid infrastructure, could be mainframe, could be distributed servers, could be cloud services, could be Kubernetes, just all over the place. Now, we're talking about a Hybrid Orchestration Control Plane. What's the shift here, Peter? Yeah. Hybrid is used a little bit differently. Before, it mainly referred to the IT landscape. When we talk about the Hybrid Orchestration Control Plane, we're talking about taking deterministic automation, which we've done for a very long time, and now including AI into that picture. And so you've got this hybrid world where we have very deterministic automation with AI and agents that's more probabilistic, which actually brings in new capability. It's not doing the same thing. It actually you can do all these new use cases because of the ability that these AI and generative models provide. Yeah. And we're seeing that in some of the early adopters, the way they're using this, and you'll talk more about that later on. I want to create an easy analogy. I probably should have, but I never heard the word deterministic before 2026. All of a sudden, it's a big word related to our world, and I started to think about it. All right, you have deterministic, you have probabilistic, which is agentic. It reminds me of cars. You never called a car a gas car before electric cars came out. All of a sudden, we have electric cars and then we have hybrid cars. But now referring to regular automation as deterministic feels a little awkward for me, but you'll hear it over and over again. We had to come up with something that made sense that pulled it all together, thus hybrid orchestration, right? Yeah. Thank you, guys. This has been a wonderful talk. I hope everybody got as much out of it as I did. The next section is going to be Peter taking you through a deck that really brings everything we just talked about to life with visualizations and examples. I'm going to pass it over to Lauren to introduce Peter while we get Peter a chance to jump up and run over to his side and get set up to do this. Thank you very much, and we will see you on the other side. As we step into the AI era, the role of enterprise orchestration is evolving. AI agents, models, services, and gateways are merging with systems and processes that are already powering your enterprise. To show how these pieces come together, Peter will introduce Stonebranch's vision for enterprise hybrid AI architecture and Hybrid Orchestration Control Plane. Peter, I'll turn it over to you. Good morning. It's good to be back with everyone. Over the past year, we've talked with a lot of you about how you see AI helping orchestrate your workloads. The difference from the prior year is real interest, people with pilots actually running and ideas of where this fits into their automation landscape. What also came through is a key question. How do we actually do this? And that's what we're here to talk about today. Past technology patterns always help us understand how a new technology evolves and yet stays the same. It started with mainframe. There were distributed systems in the '90s, virtualization came in the 2000s, then came cloud and multi-cloud, and now the agentic era going forward. Every one of these shifts was supposed to replace the one before it, but not a single one did. The mainframe is still there, still running the payment transactions we make online. Cloud didn't replace on-prem. SaaS didn't replace all of your enterprise systems. Every wave has been additive, and the landscape complexity kept growing, but agentic capability is different. The earlier shifts sat alongside what came before. AI sits next to anything. It gets integrated into all of it. It's a new capability showing up inside every application, infrastructure layer, and even new AI platforms. This will make coordination problem bigger, not smaller. The term you'll hear from us during UserVerse is the Hybrid Orchestration Control Plane, the centralized layer that coordinates and governs, secures, monitors, manages AI systems, models, agents, workflows, supporting AI-enabled business processes across the enterprise. But with it comes new challenges. There's going to be thousands of agents across dozens of systems by different vendors using different models doing different business functions. Without a control plane, we know how this goes because we've seen it as new technologies enter the IT landscape. Agents become isolated silos, governance quietly disappears, not because anybody decided to skip it, but because no one layer controlled it. And observability becomes impossible because the evidence is scattered across the platforms that don't talk to each other. We're not building something new and separate for this. We're taking the platform that already runs your business critical processes every night, mainframe, batch, cloud, workloads, file transfers, data pipelines, and pointed it to the fragmentation problem. Same governance, same audit trail, same reliability, now extended to coordinate models, agents, and AI-enabled processes alongside everything else. Every year, we ask thousands of IT practitioners what's working and what they're planning to change. We publish this as the Global State of the IT Automation Report. We ask people planning to change their automation why they're going to. Number one, 69% more functionality, more modern solution. That's up 21 points in a single year. Better customer service was at 53. Cost production was at 47, still important, but no longer a key driver. This indicates people aren't switching to save a few dollars. They're switching because their platform can't get them through the next technology shift. To dive deeper, we asked the follow-up, "Specifically, what capabilities and features are you looking for?" Number one at 40% was self-service portal, third time in four years has taken the top spot. A modern user interface, 38%, SaaS deployment, 37%, and AI workflow creation, 36%. The remainder, container automation, cloud tooling, data pipelines, these are execution features, and they are not what's driving the change. Historically, a platform was judged on one question, can it execute the job reliably? That's what's come to be expected. What people are buying today is enablement. Can my data team, can my ERP team, can my finance team get their work automated without filing the ticket or waiting two weeks for it? The platform has to become a service layer, not just a scheduler. Let's look at the architecture. Start with an honest picture. On the left, work starts, applications, events, people, and now AI agents. On the right are the system where the business actually lives, applications, data, cloud, infrastructure. Both of these ends are good. It's the middle that's the problem. Deterministic workflows over here, AI workflows over there, agents scatter all around. Every one of these was built by a smart team, and every one of them was built in isolation. Nothing here knows what anything else is doing. This is what we call disconnected agent orchestration. And I want to be fair about it, this is not a failure. This is what early adoption of a new technology looks like, pockets of experimentation, but you can't run official business processes on this. There's no end-to-end estate management, no single audit trail, and no way to answer the question, what happened and who approved it? So let's fix the middle. Here is the enterprise hybrid AI architecture, one governed process, multiple ways to execute. And note, these layers describe responsibilities, not a flow. The first layer, sources and triggers. Work starts somewhere, a person, an event, an API, a schedule, and now an AI agent. That's the new one, and it's what caused us to rethink how the platform incorporates AI technology. The second layer, the Hybrid Orchestration Control Plane. This is the heart of it. It owns the end-to-end process, the sequence, the estate, the approvals, the exceptions, the recovery, and it describes how each step executes, deterministic, AI-enhanced, or agentic. The third layer is AI governance and connectivity, model gateways, agent registries, guardrails, cost controls. This is the layer changing the fastest. As an example, MCP protocol went from a proposal to de facto standard within about a year. So we're building UAC to absorb that change without touching your workflows. And note what this layer doesn't do, it doesn't own your business process. You want to isolate that layer from your business from the churn of what's happening here. Layer four is enterprise systems because eventually this has to touch somewhere real, your mainframe, your ERP, your cloud services, and the bar across the bottom, security, observability, compliance, reliability, cost. These have to follow the work end to end. No single vendor has all this. Stonebranch provides the control plane, connecting all of it into processes that are secure, observable, accountable, and reliable. I move through that pretty fast. We're going to have another session on that on October 15th. Both Scott, myself, and Dan Twing from EMA are going to host a session where we go through each of these layers and details live, and we'll give the opportunity to ask Q& A at the end. Let me ground this in what customers are actually doing today. This is what we're actually seeing. First, AI-enhanced data and analytics pipelines. This is an ideal use case for generative AI, analyzing unstructured data, doing sentiment analysis, and even masking data with PII. It moves downstream. Second, orchestration of third-party agents, agents your team's built somewhere else now pulled into workflows managed by UAC. The agent becomes a step in a process that directly has the dependencies, the approvals, and the audit trail. You're connecting the two, and the workflow brings the governance the agent didn't have on its own or were isolated. Third is root-cause analysis and remediation. AI can analyze the failed job and pushes a recommended fix into your ITSM system. This is probably the most common use case that we see, and the reason is simple. It goes straight to resolution time. The slow part of the incident was never the fix. It was the 20 minutes of digging through logs and looking at what actually broke. Compress that and every downstream number improves, your mean time to repair, your SLA adherence, and more sleep time for the person who had to get up at 3: 00 AM or possibly not even get up at all if there's an auto remediation. Now, let's dive into how AI tasks are different. First, deterministic automation, predictable. And I want to say something about the investment you've already made here because it gets overlooked in every AI conversation I'm in. Some of you have workflows that have run nightly for over 15 years. That's not legacy, that's institutional knowledge that's encoded and governed. It runs fast, it runs cheap, and it runs the same way every time. Then there's AI tasks, probabilistic. This is where AI earns its place because there's a category of work that deterministic automation does not solve. You have an invoice in a format that nobody anticipated. An ITSM ticket needs to be interpreted and routed, even a contract clause someone has to interpret. For a long time, the answer was the same. Stop the workflow, wait for a person. That's what a model handles now right inside the workflow. The step that used to require judgment no longer requires a manual queue, so it isn't AI... Instead of deterministic, it's AI filling the gaps, deterministic automation left behind. Then there's agent actions. With an AI task, you describe exactly what to do. With an agent, you hand it a goal and a set of instructions. Go find out where the cheapest place to run this workflow is. The agent reasons, uses tools, and comes back with a result. And you'll be able to build these agents directly in UAC and drop them into your workflows alongside other tasks. The key word here is bounded. The agent chooses how, but it is governed by a set of limitations that you define. Here's that evolution in an actual workflow, how we see processes evolving over time. One, pre-gen AI, scheduled, triggered, extract, validate, business logic runs, updates, and a notification is sent, all deterministic. Most of what turns in your environment tonight looks similar to this. The second, AI-enhanced. An LLM task performs analysis on data retrieved and sends results downstream. Now, the third is agentic. An agent conducts research before the business logic, possibly enhancing the output from the previous steps or sending data to other systems through tool calling. The key point is the evolution isn't replacing automation with agents, it's running agents alongside what you already have. So the fast, cheap, predictable steps remain, and AI is introduced where it makes sense. Agents and workflows work together in many ways. The first path, the agent is a step inside the workflow. The workflow runs, reaches an agentic step, hands off, the agent reasons, does its work, and returns with a structured result, and the workflow carries on to the outcome. The second pattern, the workflow is a tool the agent can call. Now, the agent is driving and needs to act in a real system. Instead of giving it access to critical system, the agent requests the action. The governed workflow validates, routes for approval, executes, and returns the result. The agent gets a scoped interface, no direct access to infrastructure, and you get the reliability, predictability, and audit trail automatically. The second pattern is an important one. Connecting an agent to a system is easy now. MCP makes it so you can do that in a few minutes. Give agents governed workflows as their tools instead. They get the power to act and you get the control of what happens. I typically get asked, "Won't agents just handle all of this?" The question isn't whether agents get adopted, it's what they're going to run on and how are they best used. Gartner in June of this year indicated, by 2030, 90% of infrastructure and operations will be using hybrid automation. Today, it's under 10%. That 90% hybrid, not all agentic. The market is not converging on replacement, it's converging on a mix, and here's why. Determinism is what makes autonomy safe. Restart from the point of failure, dependency resolution, SLA adherence, and auditable record, every one of those is a deterministic property. You can't compromise an SLA on something that may work differently from time to time, so the more autonomy you grant, the more you need a deterministic layer underneath to catch it when it goes wrong. I'd like to leave you with one final thought. Stonebranch is a control plane for business critical automation. It does not replace the entire automation tools that you have. Your data team likes their tools, and your DevOps teams likes theirs. What we do is three things. Connect the automation islands so those tools, teams, and agent pilots become one coordinated picture, coordinate end-to-end execution, deterministic jobs, agent task, agent actions, human approvals, all the way from the mainframe to the SaaS app, and govern change, risk, and recovery because something will fail at 2:00 in the morning and you'll need to know what broke, what it touched, and who approved it. Here's what I'd like to ask you to take away. Everybody talks about what agents can do. Very few are talking about what it takes to run them. The second question decides whether your AI program is a set of impressive pilots or a part of how your business actually operates. And UAC is the Hybrid Orchestration Control Plane to provide the backbone for safety and reliably deploying your AI and agents. Thank you. And now, to look at some tangible demos, I'm going to pass it over to Nils who's going to show you some exciting stuff. Thanks, Peter. Now let's bring Hybrid Orchestration Control Plane to life. I will give you a demo with the different scenarios that you were talking about. In this demo, I will show how Universal Automation Center connects file transfers and orchestrates the applications that use the data. We will follow sales data through AI processing, SAP analytics, and plannings. We will also see AI investigator a failure and open an incident ticket automatically in ServiceNow, then use Robi AI to help fix it with my confirmation. Finally, a Kong AI task, we create an execution report and post it to Slack. Let me explain the demo workflow. Uploading sales data to SharePoint triggers a workflow. UAC downloads the file to an SAP application server on AWS, then starts two parallel streams. The first stream marks sensitive fields and waits for manual approval before starting the model training pipeline through the Google AI Agent Pipeline task. UAC triggers and monitors that pipeline. The second stream loads the sales data into SAP through an iFlow. ABAP programs then calculate commissions. After that, the workflow branches again. OnePath runs Snowflake's analytics while the other harmonizes data and SAP Datasphere and loads sales data quantities into SAP IBP for planning. For analytics, SAP Business Data Cloud Connect for Snowflake makes published SAP data products available to Snowflake through zero copy sharing. Think of a data product as a prepared data set with business context. Snowflake can use that shared data without a separate copy for this connection. UAC checks that sales data product is ready before starting the analytics tasks. Once analytics and planning path finishes, the report task brings the results together for the team. Let me upload the sales file to SharePoint. The file is now on the SAP Linux server, and the downstream task have started. Let's look at the AI masking step. This task uses Azure OpenAI to master data before it reaches the Google pipeline. I choose an authentication method, such as service principle, and configure the tenant and credentials. Then I select the model available for this connection. For this demo, I'm using a smaller model to keep costs low. The system prompt defines the masking rules. The conversational prompts supplies the input data. Here's the output with the selected field masks. Before the Google AI pipeline starts, I've edited a manual approval step, Assigned to Nils. I can handle this in Universal Portal, including any mobile device. The new Universal Portal supports many languages and runs on any device. You can even use it with a mobile with one hand because it has a responsive design. Let me approve the task. The approval is complete and the Google AI Agent Pipeline has started. Meanwhile, one task has failed. Any error in Universal Automation Center automatically triggers a Universal Event, which is captured by our configured webhook. This webhook launches the Kong AI task. The task retrieves the failed instance logs through the Stonebranch MCP server, uses Dynatrace tools for additional context, and asks the model to analyze the issue. It then creates automatically a ServiceNow incident through the ServiceNow MCP server. So three MCP servers are involved in this scenario. As you can see here, the task can connect to multiple MCP servers subject to the access configured in Kong Konnect. Its model request passed through the central Kong AI gateway. It can also limit which tools are available to the task. I can enter prompts directly or use version script on the UAC script library with variables for the current run. Here, we describe the instructions how to investigate a failure and create the incident. Here, you see the incident, including the AI analysis. The SAP task uses the wrong ABAP variant, List D instead of List P. A mistake like this can happen when variant names differ between development and production. This happened to me already several times in projects. Let me fix the issue with Robi. I will ask Robi to verify the issue, then update the variant to List P and restart the instance. Robi asked for confirmation before proceeding. I will confirm the change and check the result. The SAP task has now completed successfully. I will close the ServiceNow incident and verify its status. The incident is closed and the workflow has continued through Datasphere and SAP IBP. The final Kong AI task retrieves the workflow task log through the Stonebranch MCP server and compiles a structured execution report. The report is then posted automatically to our workload automation Slack channel, giving the team a summary of the run. That's a complete flow from a SharePoint upload through AI processing, SAP analytics and planning with human approval, AI assistant recovery, and automatic reporting coordinated by UAC. Now, let's take a closer look at how the vision is taking shape across the Stonebranch product portfolio. Peter and Gwen will share the latest developments across UAC, Universal Portal, BatchMan, UDMG, and our growing portfolio of integrations. We'll also see some of our newest AI innovations in action. Joining us for these demonstrations are Huseyin Gomleksizoglu, lead solution engineer... are Huseyin Gomleksioglu, lead solution engineer, and Nils Buer, vice president of solution management. They'll take us through a closer look of the AI Workflow Builder and AI Integration Builder. All right. Gwyn and Peter, take us through what's new in the Stonebranch product portfolio. So, I think we're getting into the real fun stuff here where we're talking about some of the things that we've just recently released and what we've been up to. Yeah. There's been a lot. So, let's get into it. Yeah. I think that if we really look at the past 24 months, it's been pretty spectacular what we've been able- Yeah. to accomplish. Like, I think it really starts with the AI path. Right? So, everywhere from releasing Robi AI, which is really the core of what we do and powers all of our AI capabilities through to this new workflow builder, the extension builder, and, of course, the entire ecosystem of tasks that we build to hook into LLMs and- Yeah. Yeah. application-based AI agentic platforms and other agentic-based platforms. It's really been a great path forward. Yeah. So, if you look at where we are with the whole portfolio today a lot of it is the same, but it's also changed quite a bit as well. Yeah. So, it starts with the Universal Controller and the Universal Agents, which have been the mainstays of the entire platform for many years. Yeah. Along with the Universal Data Mover, which is integrated in the agent, which provides a very robust high-volume scalable transfer protocol. The UIP, Universal Integration Platform, was a real turning point for us in our ability to really rapidly build out into all of the various automations that we need to hook into. And now we have the Universal Data Mover Gateway. And, of course, Robi AI. Maybe it would be good to explain what the cadence is on how often we release these products and how we try to keep everything- Yeah. positively aligned. Yeah. We're very focused on the cadence with which we release out there, because of how fast everything's changing. That core platform, we like to be very careful with the changes. Usually, those are every six months. And that's the very cadence of... Most of our customers know we do it the same time every year. We don't change it. UDMG, given that it's a newer product and not as complex as UAC, that is more of a three to four month type of timeframe. The integrations, of course, like, we discussed is really on demand when they're ready. And we don't have to build those. Those can be built by our partners. Those can also be built by customers themselves. And then we have BatchMan, which we're releasing also and keeping updated. And that's how we like to think about the whole platform, really having that set cadence, so, that it manages the change that we're putting out. Right. And then, of course, the integrations, we released those... Yeah. to the Stonebranch Integration Hub. Right? And they're immediately available on demand for people to download rather, and then implement into their system and start working with it right away. Good. And these are just some of the highlights that we're releasing... Yeah. this quarter. Right? So, Universal Automation Center 8.1 will be out along with Robi AI 2.0. There'll be a new Universal Portal, and this is a great release with the addition of workflow visualizations. And then, also, the data mover gateway we're releasing in conjunction with UAC. Plus, also, there's a new BatchMan coming in November. And, of course, we're going to have countless new integrations that... And, of course, one thing I like to point out is that the Robi AI really is used across our products. So, it provides that fundamental capability to embed AI into our products. So, that's something that, again, can be used across our portfolio. Yeah. Excellent. And in terms of Robi, maybe walk us through... Yeah. Robi AI. Sure. So, one of the most important things that we look at when we look at putting in AI functionality is making sure that it's private and it's secure. That's top of mind with most customers and most of the questions we get. So, the models we use are, typically, pointed to that are private, we see customers using the model gateways. Right now we see customers going through the Azure OpenAI and we actually actively test against that. No customer data is collected. So, when you use these private models, typically, there's guarantees that they're not using your data in any bad ways, let's say. And then traceability. We are tracing all the things that go through the AI engine, Robi AI. And that's also something that the model gateways provide too. So, there's almost different layers of governance that you see now. And then full deployment control. So, you're able to configure Robi AI. You're able to deploy it with UAC or with UDMG, but you don't have to. So, it's not, like, it's shipped within the product. And we did that very specifically, because we knew that customers were going to be at different stages of adoption and we didn't want to force them to have to install the AI component. So, it is something that that is optionally, that you can use or not. Excellent. And, of course, one of the great things about Robi AI is that Robi AI can only do what the users permissions allow it to do. Right. Right? So, it respects full permissions within UAC as well. And when we look at 8 .1 and Robi AI and Universal Portal highlights, these are some of the capabilities. We've talked about it a few times already. And Huseyin is going to be up right after this. Yeah. I think with an awesome workflow builder demonstration that really shows. Yeah. the full capabilities of what we've delivered. The MCP client and server. Yeah. MCP client and server. Yeah. MCP has become, as I talked about it, a de facto standard. Everybody is putting an MCP layer on their products. Essentially, it, typically, wraps your API. So, we have a rest API. And you can almost think of the MCP as a front end to the rest API that allows AI clients to actually hook into UAC, if you give the authorization to do so. And so, we've seen where we can hook up different OpenAI clients, Claude clients, and you can sit there at prompts and be able to interact with the system. It's also the way that agents will probably interact with the system, because MCPs have what we call tool calling and you can give that access to agents. So, we see that as a very powerful capability to enable companies to leverage their AI, AI clients, or AI agents to do things within UAC. But, again, there's very much a security protocol that goes in front of that. And then, also, with our jobs as code capability, this works with both UAC, 8. 1, as well as the jobs as code extension that we have where we've added some capabilities that makes it much more robust in terms of being able to respond to. Right. more of an event-based process as well as being able to synchronize the unique user IDs. Yeah. of jobs as code back and forth as it moves between Git and also the UAC database. Now we've also heightened security. We've gone to AES 256 for our credential store. And then with the Universal Portal, we've now taken it a step forward. We had a really good visualization at a task level, but we really wanted to show visually the diagrams of workflows. So, in terms of either the definition or live running view, you can now see a workflow from within Universal Portal. And, of course, we have many more enhancements. We try to do dozens of small enhancements that our customers find important for their everyday use of the product. That simply just makes the product better. Now UDMG 3.4. We continue to move forward with this. A couple of key things was really the container support. Yeah. This was important. Again, we see customers really wanting to deploy in a containerized environment. And so, trying to package UDMG, so, that you can do that, you can have pre-built images that you can use to deploy the product. It also scales very well when you can deploy it as containers too. So, that's coming out. And we went through that process over the past few years with UAC. So, it's great to also see this with UDMG 3.4. In addition, we've added versioning, especially, for configuration differences. So, if the configuration has shifted. Yeah. it's a diagnostic point to be able to find out what's changed and to be able to go back and revert the changes. Additional features is really the quota piece. And this is really important for partners, especially, for partners who may not be using the system properly, maybe abusing sending too many files, too many big files. This is now completely controllable. And then, also, the CLI for administrators. And then one of the most innovative changes, our universal secure proxy is a pretty sophisticated solution. Maybe take us through the forward proxy. Yeah. UDMG is enterprise grade, and it's used by financial institutions. And so, we needed to have a very strong security layer. And so, one of the unique things we do there is we don't just proxy the connection straight through to UDMG. We actually break the connections, can do some inspection there to make sure it's valid, and then there's a separate connection back to the back end. So, it's a much more stringent security layer than you would with just a normal type of proxy. Right. So, we offer both the session break and also the standard reverse proxy and then also the forward proxy. Right. And you can also do virus scanning at the proxy level as well. Excellent. And then, also, maybe I'll just cover off BatchMan here as well. One of the things that we've found important when HONICO joined the Stonebranch family and brought BatchMan into the fold was really that we get on a regular release cadence and deliver features to customers regularly. So, we started this off this spring in February. We released 5. 3. And also this November we'll be releasing 5. 4. On top of that, these marquee features, we're delivering a lot of other customer-driven enhancements. But we've also done something a little different in terms of from a support and compliance perspective. Yeah. One of the benefits to the BatchMan customers is our capability to do broader support. We have locations all over the globe. So, we're able to do that seven by 24 support. Also, we're very up-to-date on our compliances. There's a lot of regulatory compliances that we have, SOC 2, ISO 27001 and there's also up and coming ones in Europe when you look at NIS2 or DORA. And so, we're making sure that the BatchMan product comes under those programs too. Excellent. Excellent. Yeah. And I think the other thing is UAC, obviously, has a much broader reach than BatchMan. So, we also have that integration that if the BatchMan customers need it, they now can hook into UAC to provide that extended hybrid IT type capability. Right. Right. And now I think finally it's time to talk about our integrations. So, these are the Universal Extensions, which are available through the Stonebranch Integration Hub. I'll start off with the center column. So, with SAP, we've been building SAP-certified integrations for over 25 years. And we've been certified for SAP R/ 3, S/ 4HANA, Rise With SAP. We really cover the gamut. And there isn't a level of SAP certification that anyone else has that we don't have as well. What we've also been doing is we've been adding more and more integrations into the SAP ecosystem as this list here indicates. On the left-hand side is Oracle Fusion. So, this is a little bit new for us. So, if you look at enterprises, large enterprises, they, typically, have one of two ERP systems. They either have SAP or they have Oracle. So, what we've been doing is we've been building out over the past year a wide array of Oracle Fusion, which was the Oracle E-Business, Oracle financial suite. And we've been building out a broad range of integrations to cover off all of the business use cases our customers who are running Oracle ERP systems require on a day-to-day basis from an automation perspective. So, we're really at a point where it's gotten to be quite complete. Every single one of these are running in production with large customers, and if there's any customers out there who are interested in talking to us about our Oracle integrations we'd be happy to have a conversation with them. Now let's talk about some of these AI integrations we've been working on. Yeah. This is the things that I talked about during the keynote and we talked about upfront, which is it's important that we have AI capability within our product, but we realize that agents are going to be running in all different places. And we've seen customers start to build agents, some of these platforms like Kong, on Google Vertex. And so, we're making sure that we have integrations to those platforms, so, that as these agents proliferate everywhere and you have one that you want to integrate into a workflow, that you can do it through these integrations. And there's also SAP with their... What's their agent platform called? Joule. Joule. Yes. So, SAP Joule, Nils has done a great job at building some integrations into SAP Joule as well. So, again, these agents will live in different places and we want to make sure that we can integrate in no matter where they live. Yeah. And we're waiting for Joule 2.0... Yes. so, we can even take it that next step forward and have some really fantastic integrations there too as well. But beyond AI and ERP we're also making sure that we integrate with the rest of the ecosystem. So, that would include if other integrations in terms of Cloud platforms, in terms of observability. Yeah. and also for self-service with things with ServiceNow. Now we've been working on something a little bit special with ServiceNow. Maybe walk us through what we're doing in the whole infrastructure space along with our integration strategy. Well, the ServiceNow, that was something that was actually built in the ServiceNow platform. So, it actually is a component that you can go to their app store and download. But you can actually have a ServiceNow ITSM workflow that can actually trigger things in UAC. So, SAP jobs, different tasks, workflows. And that's a very tight integration that's there. Yeah. And then, also, speaking of, yeah, marketplaces, we also now have a certified integration within the Dynatrace. Yeah. Yeah. for observability as well. Yeah. What we see is we had looked at putting a very sophisticated dashboard of capability within UAC, but when we talked to customers they have these solutions already where they aggregate from different systems and you see Dynatrace, Datadog out there. And so, that was one of the drivers for OpenTelemetry. But what we've done is actually had things in their stores that are purpose-built for UAC. So, that you can go to their app centers and use a UAC dashboard that's already there. And we've seen customers go that route and adopt those types of capabilities. So, we've covered off a lot and most of this is really in the past six months. Yeah. since we released 8.0. Right? And so, the pace is going to just continue to move faster and faster, and with the new integration builder, of course, we may need to be even more selective on what we... Yeah. ...display out of all of the ones we can possibly build for the future as well. So, I think it's now time for us to pass this over to Huseyin, because we've been talking a lot about this... Yeah. ...AI Workflow Builder. Right? And I'm just dying to see his demo in real life. Yeah. So, over to you, Huseyin. Thank you for Gwyn and Peter. I know you're excited for the AI Workflow Builder. Today I will use the AI Workflow Builder to build a workflow that will do bulk exports, compress, encrypt, upload the SFTP, and then send notification. So, let's start the demo. So, now I will ask Robi to create me a workflow. I need to describe what I want. So, I already have this script here. So, I'm telling Robi saying that, "I want to create an export and they can process." And I also give some information about the processes. For example, I want to export, compress, encrypt, upload, and clean up. So, by that kind of prompt, Robi will understand that we want to create a workflow and a workflow engine will start. And then based on this one, so, it suggested some task, but I don't like this task. So, I want to create, I want to use this UserVerse task. And Robi will give you option to select your task or you can give the information or you can just give some part of the task you want. So, I don't like this task, I don't like this one. So, I want to use a compress, export, encrypt, upload. I don't like this one as well. Clean up and notify. Yeah. So, I select the task I want to use. So, based on these task Robi will try to predict what's the flow in the workflow. So, it says that export, compress, encrypt, upload, clean up, and also it asks me," Do you want to have the notification? "Notification will be the last one. Notification will be the last task. So, in this case, it will update the flow. Then this is the final flow. Now export, compress, encrypt, upload, clean up, and notify. This looks good. And then it asks me if I want to create it. Yes. I want. So, the workflow will be created here. So, let's see the workflow. And, also, I can ask Robi to launch it. So, Robi will launch the workflow for us and it will ask us two questions. One is what's the reason to launch it? And do we want to use some variables? The reason is... If you don't want to give any reason you can just say none. So, the workflow it created is export, backup, workflow. This is the name of the workflow. So, also, Robi decided a workflow name for us. And then it asks me if you want any variables? I said," No. "And it will launch the workflow. So, there's a task in the workflow will fail. Let's see it. So, the first task is using the CLI comment to do a bulk export on a server side, and then the second task, when it tried to compress the file it failed. So, I can also use Robi to analyze this. Robi will do the analyzation of the failure. And so, it will tell us what was failed, what is the error, and also it can also give us some recommendation. So, it says that it recommended a fix, but it didn't fix for us. So, let's say that," Can you fix it? "As AI gives different responses, different times. So, sometimes it directly asks. Otherwise, you can also ask Robi to fix it for you. So, in this case, it suggests me to update the script and rerun it. This is exactly what I want. So, now Robi's fixing the script for us, and then it will rerun the task and we will see it online. Okay. And you see the task is completed. The workflow is completed. Fine. So, now I want to create one more task here, because if one of the process fails in the middle... So, there will be some leftover files. And we don't want to keep the leftover files there. So, I want to create a failure path. So, in this case, instead of creating just a Linux task, I want to do a little more fancy, I want to do a safe delete. So, I want to use some script. But I don't want to write it. I will ask Robi to write it for me. So, I will say that," I need to create a script. I want to create a Linux task. "No. Not this one. I will say that," I want to create a script. "And that will do a safe delete. But, also, I want to say," At the end, if there is no error, echo a message." So, I can give that kind of direction for Robi. So, Robi will create a script for us. Robi creates the script here. So, it's using some RF, but it's also validating the output and everything. So, let's say, "This is good." Now I will tell Robi to use the script, and create a Linux task for me. So, in this case, I said that, "Create me a Linux task using this agent, this credential, and use the safe script." So, at the end of the script we have a message, like, a safe delete completed successfully. "So, we can say something, like, this," If you get this message, we will consider this task a success. "So, I will also give that kind of information. Let's see. So, now Robi will create a Linux task for me. And then it's asking a name for a task now. So, I will give a name, something, like, this. I said that," It's, like, a safe delete task. "We create the task. Now I have the task here. So, this agent credential using that script. Everything is good. Now I want to add this task to the workflow. So, Robi can also update the existing workflows. So, here I have the workflow here. I will say that," I want to add the task to the workflow. "And what was the name of the workflow? Export. Export. Backup. And workflow. Then Robi will ask me where I will put it. So, it put the task here. So, this is the task I want to add. Then it will ask me in which part I will put this task. And this task will be executed when the encrypt task is failed. So, you don't need to write the full name of the task. So, you just give some information, so, Robi will find which task to use. And then it will create the connection. And you will see it in real time here. So, when it's updated you will see the update on the workflow as well here. So, it's added. And the last time... Let's launch it again. So, let's see if everything is good. So, what's the reason? And the reason is none. And the variables is none. And launch it. So, this time when it's successfully completed it will connect with the success path, the failed task will be skipped, and everything works. So, this is the AI Workflow Builder. And as you can see, we can create task, we can update scripts, we can create scripts, we can create workflows, we can update workflows with AI Workflow Builder. And now we will pass to Nils. Thanks, Huseyin for that great workflow builder demo. What I notice is that you can still enhance your workflow by adding another file transfer to an Amazon S3 bucket. Let me show you how you can build a complete new integration with our new AI Integration Builder without any coding. For this, we created the AI Integration Builder. As input, I only provide a simple markdown file describing what the new task type should do. In this demo, I want to create an AWS object storage task with two actions, list objects in the AWS S3 bucket and upload a file from a Linux server to an AWS S3 bucket. Then I choose the AI model and provide my API keys. I can now also ask the integration builder to generate the documentation and upload the created code into my Git repository. Optionally, I can auto-generate test cases. Under the hood, the integration builder uses Claude code together with Stonebranch-specific skills and specialized agents. Different agents take care of requirement, extension design, implementation, dependency management, testing, debugging, packaging, and documentation. The integration builder also uses agentic looping. This means that if an issue occurs during automated testing the agent analyze the problem, automatically adjust the generated code, and run the test cases again. Once everything is working, the extension, this again. Once everything is working, the extension is packaged and imported into Universal Automation Center. The generated source code and documentation can also be published to a source code repository such as GitHub. You saw that I entered my Git credentials. So I only describe what I want. The integration builder takes care of how it is implemented according to our Stonebranch development and testing standards. I will now launch the task. Now the AI agent will take the coding work. This will take a couple of minutes. I will hand over to Lauren for the product update and come back once the agent have finished their work and we can look at the finalized integration. Thank you, everyone. We've covered a lot from the latest product updates to some exciting new AI capabilities and demos. Now, let's look ahead. We'll walk through the Stonebranch product roadmap and see the AI workbench in action. Most importantly, we'll discuss what these innovations mean for the Stonebranch users, from building automation faster to simplifying your day-to-day work and getting even more value out of Stonebranch. Gwyn, let's get started with the roadmap. So now it's time to take a look at the road forward. And we've had a hypothesis for some time on the possibilities and evolution of how AI will be able to help you create automation quickly, effectively, and at a higher level of quality than ever before. Imagine the following. You log into UAC, you go to Robi AI, and not only can you create your tasks and your workflows like you saw earlier with Hussein, but you can also create quality code. There's so many small applications that exist in your enterprise, everything from scripts that will zip things, applications that will do scans for you, and a lot of that can actually be done through programs developed from UAC that then you can keep in your code repository. And then from there, of course, there's the integration builder where from a no code perspective, you're able to create integrations directly within UAC and have them installed in place. In addition, if you are a more advanced developer, we also have the code version where you're able to go in and create also very advanced integrations and integrate it with some of your applications. So let's look at the AI capabilities that we're building across our portfolio. Robi AI, which is available today that we're going to continue to enhance and augment over the next release, and really also focus it on the agentic orchestration piece. The AI Integration Builder, which is currently in early access, which we will be taking forward to a commercially available product, and of course, the AI agent task itself, which will do the agentic orchestration on the UAC platform and then also help us connect and do greater orchestration than ever before. That's the product side. The integration side, all of these extensions that we create are going to enable us to encompass the entire AI ecosystem from LLM task, AI platforms, and also AI gateways which are emerging in prominence and helping enterprises to govern and control access to AI models and MCP servers. And in terms of AI orchestration, today, UAC definitely can do AI orchestration. Right now we are orchestrating the AI ecosystem. With 8. 2, we will do all of that, plus we will have our own AI platform for agentic orchestration, which will allow you to build, manage, and run in agentic runtime agents, as well as orchestrate those across the enterprise. With the UAC AI native task, we're going to be providing an AI agent runtime. As mentioned earlier, we'll be able to interact with LLMs and gateways, interact with MCP servers and MCP clients, and then also create your own agents just like you do with AI integrations today in a no code way. And with the AI Integration Builder, from simple instructions, you'll be able to build the extensions automatically, the no-code approach, the automated deployment, which is something that you'll be able to take and implement in your test and dev environments and then be able to promote into your production environments. And then also we'll have this available via coding assistant. And then to wrap up the portfolio, we'll also be doing a new release of UDMG, UDMG 3. 5. Here we will also be introducing Robi AI into our managed trial file transfer, our B2B file transfer world, making it easier than ever to set up new configurations, find documentation in how to solve problems in a fraction of time, and also an AI assistant that you can interact with just like you can within UAC, within UDMG. We'll also be beefing up our integrations with notification channels. We'll be adding multihomed virtual file servers, and we'll be implementing deeper and deeper data retention policies and purging controls, which are even more important in the world of compliance. So there's a lot on our way. But before we wrap up, we're going to show you one last thing that we're planning on rolling out in 8. 2. We would like to introduce you to the newest member of the UAC Family, Universal AI Workbench. So AI Workbench is a new innovative user interface that is AI native and GitOps native. It is really designed to be something that will work for traditional UAC users as well as UAC users who wish to use as code or GitOps approaches. So what is it? At the center of it is an AI Copilot, but it also has a workflow builder that is graphical, a task library where you can either work in forms or in code, a code library with a formatted coding editor, a runbook library, which is a whole new concept we'll discuss later, and then access to all of your UAC data that you can then take a look at and import into Git or GitHub. So you are able to switch effortlessly between working with workflows and forms to working with code and then working with an AI assistant. It has UAC connectivity, it's multi-environment, it is Git as a repository, and of course, you can publish and sync with UAC. So you're probably wondering how this all works, right? So we have UAC, we have Git or GitLab or GitHub, whichever your favorite Git is, and of course, in the center we have AI Workbench. So you can build in UAC, which many people love to do on a daily basis. And of course, with Robi AI, we make this even easier. And then you can sync with Workbench and then it'll be stored, and then you can also store it and version it within Git. You can build anywhere using any code editor you want, save this to Git, and then access it via Workbench and then sync, publish, and import into UAC. Or you could work in Workbench and you can go both ways, publish, sync, and export it to UAC. And then also it's natively stored and versioned within Git. And those versions are available to you and you're able to recover from prior versions using Git and then also sync the versioning within UAC. Now, I think it's time for us to switch to a browser and let's take a look at this thing. First of all, when you log in, the first thing you're going to see is a workflow. And within this workflow view, it's a traditional UAC workflow where you're able to use all of the traditional workflow commands you can use while editing. You can expand. You can zoom. You can change a layout, add tasks, add workflows. You can import from UAC. You can build from scratch. Over on the right-hand side, you'll see the Copilot, which is powered by Robi AI. In this example here, you'll see that I've asked the questions about the workflow on the left. Moving from left to right, you'll also see that we have a task library. And within the task library, it's very easy to search. And then from there, you'd be able to see it from a form based view or switch into a JSON view as well. So you can go from a form view to a JSON view. You can publish it to UAC. And then from here, we also have a new capability called task templates. So this is something that doesn't exist in UAC today. This allows you to create templates of tasks and those templates will allow you to create new tasks based on that task template. From here, we also have the code library, which is based on the script library which exists within UAC, but we gave it the name of a code library to give it a little bit more visibility in terms of what you're really able to do now, especially with AI. So I'm going to pull up a script that we have stored and here we'll see that it has all the formatting that you would expect if you're using a code editor for Python. From here, you're able to naturally save it to the GitRepo, and then you're also able to publish it from UAC. And then of course, if you want to import from UAC, you can simply click here and it'll pull up a list of the existing scripts that exist within UAC. Being able to pull from UAC is a universal capability between the workflow builder, the task builder, and the code library. We have a few new items as well to look at. One is the runbook library. So what is a runbook? If you think of all of the traditional manuals and documentation you would have for procedures within the data center, many of those are automated, yet the runbook and the definition itself may exist in tools like ServiceNow, they may exist in Word documents or PDFs somewhere on a SharePoint server, or they may exist even in Excel tables. What we've done here is we've enabled the capability of pulling up an existing runbook, which will describe all of the steps. You'll be able to go in, understand what each step of the runbook does. You'll also be able to enrich it with references and attachments as well. More importantly, if you have a runbook, there's a button here called Build Workflow and it'll automatically build a workflow based on that runbook. Not only that, but if you feel that you need to start doing a better job of documenting your existing workflows, you can select an existing workflow and through the AI Copilot, ask it to create a runbook from scratch based on an existing workflow. And of course, UAC data. This allows you to go in and connect to a UAC instance and navigate through the data itself. From there, it'll bring it up in JSON. You can then easily go in and copy that JSON and then paste that within your IDE, and then you can start working as code. So these are just some of the many capabilities that are going to be available within AI Workbench. But in the spirit of how we've been making things easier with AI, there's one last thing I'd like to show you that AI Workbench is capable of doing. You're going to look here on this tab and you're going to see a diagram. Now, please don't mock my whiteboard skills. This was a bit of a messy whiteboard. There were shadows, as you can see. My penmanship never got me a top mark in public school, but you'll see that I've gone in here and said one task's a Linux task, one's a manual, and I kind of scrawl below each one of them a name, and then I kind of put an agent in also the same pretty lousy scrawl as well on where I'd like it to run. So what do you think we can do with this? Let's see. So we're going to go back to AI Workbench, and I'm going to ask AI Workbench if it can actually take that whiteboard diagram and go and build me tasks and a workflow. So here I go and I'm going to import this image. There's no tricks up my sleeve, it's the same image. And I'm going to ask it, "Can you please create a workflow and a set of tasks based on this whiteboard diagram and show me the workflow?" And it's thinking. This might take a few seconds. And as it's working, it's going to be analyzing and it's going to be figuring out exactly how to build the tasks and how to build the workflow. Oh, here we go. It's gone and it's created a workflow for me. And it's gone and it's created five tasks and it's built the workflow. So let's go over to the workflow builder and here's the workflow. This workflow was built from this diagram. Imagine being able to sit in a cafe with the back of a napkin and design your automation, work with team members on a whiteboard just like I did to figure out a process that you want to create in UAC. Imagine being able to sit with something like Microsoft Visio or Miro or one of the other countless diagramming tools, even PowerPoint, and being able to create automations, really making this accessible to any user out there. I've been in automation for more than two decades designing automation products. And with the products that we've seen today with Robi AI, with agentic orchestration, and with tools like AI Workbench, this is the most exciting time it's been since I've been involved in this industry. Now, it's time to go back and check what Nils is doing. If you remember earlier, Nils started creating an AI integration, and I think it's probably done by now. So over to you, Nils. Thanks, Gwyn, for the nice Workbench demo. That is really cool stuff. Let's have a look if the AI agents have also done their work. As you can see, the integration builder completed successfully. Let me open the log file. In the log file, you can follow the different steps it performed to generate the integration. At the end, you also get a summary of the build costs. In our case, the total cost was less than $5. Now let's have a look at the task type it generated. Because we enabled automatic test case generation, you can already see the generated test cases for the list objects and upload file actions. Let's open our new AWS object storage task type. It contains exactly the two actions we requested, list objects and upload file. Now let's try it out. First, I will use upload file to copy a file from my Linux server to an AWS S3 bucket. I will enter the name AWS S3, select the agent and credentials so I can access the source file. I enter the source file and enter the target bucket, sales commission, the target object key, and I select the AWS S3 credentials, so secret key and private and key. Now I run the task. As you can see, the transfer completed successful. This is also shown in the log. Now let's verify that the file really arrived in S3. I run the same task type again, but this time I use the list object action. So I create a new task type, which I call list objects action, by just copying the existing task. And I run the list object task. Here you can see the file we just uploaded is available in the list output. So starting with only a small markdown requirement file, the integration builder created a complete working AWS S3 task type. It's implemented the integration, generated and executed tests, automatically corrected issues where necessary, packaged the extension and deployed it into Universal Automation Center. I also asked the integration builder to generate the documentation. If we now look at GitHub, we can see the generated extension source code together with the automatically generated documentation and markdown. From here, I can go one step further and add customer specific integration test cases. For these tests, I can provide the required credentials and environment specific information. The integration builder can then generate the corresponding test cases automatically as task and Universal Automation Center and execute them directly against the target environment. This means we can validate a generated integration not only with generic test cases, but also with real customer specific scenarios. The final result is much more than just a working task type. We also get the source code documentation and reusable integration tests under version control ready for further enhancements, testing, and maintenance. Let me now add the newly generated task to the workflow that Hussein generated with the Robi AI Workflow Builder. I want to add the new AWS S3 task to the export backup workflow right between the upload SFTP task and the cleanup task. And in Robi AI, I can simply describe this exactly the way I just said it. Robi, add the AWS S3 task to the export backup workflow between the upload SFTP task and the cleanup task. As you can see, the new AWS S3 task has been edited right between the upload SFTP task and the cleanup task automatically by the workflow builder. So in this demo, you have seen how easy it is to create a new integration with a new AI Integration Builder and then add that integration to an existing workflow using the Robi AI Workflow Builder. All right, this is all about the integration builder for today. And that brings today's keynote to a close. I want to thank Scott, Peter, Gwyn, Nils, and Hussein for sharing their insights and giving us a look at what's next for Stonebranch and the future of enterprise automation. And of course, thank you for joining our first session of Stonebranch UserVerse Online 2026. Today we looked at hybrid orchestration control plane, the evolving role of orchestration in the AI era, and the latest developments across our products, roadmap, and new AI capabilities. This is just the beginning. We have three more UserVerse sessions ahead. We'll go into a deeper dive of the topics and technologies we introduced today with more insights, demonstrations, and practical use cases. We hope that you'll join us and continue the conversation around AI and enterprise orchestration. Thank you so much and we'll see you at our next UserVerse Online session.