AI Agent Builder Summit | Jesse Shiah, AgilePoint

Clips
News
More from AI Agent Builder Summit

Jesse Shiah

CEO & Co-Founder

AgilePoint

What to expect at the AI Agent Builder Summit: Join theCUBE April 16

Causal AI is gaining momentum, with seven in 10 enterprise artificial intelligence professionals either using, experimenting with or planning to adopt it by 2026. That surge in interest reflects a growing demand for systems that don’t just respond, but explain: AI that can reason, justify its decisions and build trust with the humans it serves. As companies evolve beyond generative AI’s headline-grabbing outputs toward more functional, dependable AI architectures, agentic AI systems and causal intelligence emerge as essential components of enterprise-grade solutions

AI Agent Builder Summit preview: From coding to composability – the rise of agentic AI

AgilePoint Inc., a premier sponsor of the AI Agent Builder Summit hosted by theCUBE Research, is at the forefront of empowering businesses to swiftly operationalize robust, agentic workflows across various vendors and platforms.The rapidly evolving field of agentic AI is driving a paradigm shift in how enterprises approach automation and orchestration. At the heart of this transformation lies the concept of “holistic abstraction,” which enables businesses to achieve resilience and agility by orchestrating complex workflows without traditional coding constraints

play_circle_outline Introducing AI digital summit by SiliconANGLE and AgilePoint

play_circle_outline Importance of agentic AI in business operations

play_circle_outline History and vision behind AgilePoint's agentic workflows

play_circle_outline Importance of abstraction and AI Control Tower framework

play_circle_outline Composable enterprises and business models with agentic AI

play_circle_outline Technology should be programmable for agility and future-proofing

play_circle_outline AgilePoint AI Control Tower framework for holistic abstraction and harmonization

play_circle_outline Key success factors in building AI agents and agentic systems

Info
Transcript

Jesse Shiah, AgilePoint

Jesse Shiah

CEO & Co-Founder AgilePoint

TheCUBE Research’s Scott Hebner was joined for a preview featuring Jesse Shiah, CEO of AgilePoint, as Shiah discusses AgilePoint’s capabilities to be showcased at the AI Agent Builder Summit. AgilePoint's innovations assist organizations in operationalizing enterprise-scale multi-vendor workflows at the intersection of human and digital workers.

Hebner has a conversation with Shiah about AgilePoint's mission and advancements, including the company’s no-code solutions and how they empower enterprises to create adaptable, AI-driven operations. This dis... Read more

explore Keep Exploring

What is the digital summit hosted by SiliconANGLE and theCUBE Research on April 16th dedicated to exploring best practices in building AI agents and agentic systems? add

What is the top prime candidate for overcoming ROI challenges for AI initiatives? add

What is the importance of harmonizing tech stacks and overcoming siloed business operations? add

What is the concept of utilizing a separate layer of abstraction, such as the AI Control Tower framework, to enable the collaboration and integration of various agents, like IBM, into existing workflows without the need for migration or extensive training for business users? add

What can help organizations achieve the notion of composable enterprises and composable business models with the evolution of agentic AI and autonomous agents? add

What is the importance of making technology programmable and how can organizations leverage it to create future-proof applications and solutions? add

What are the four key attributes of the AgilePoint AI Control Tower framework? add

What key success factors are explored in building AI agents and agentic systems? add

bolt Powered by CUBE AI

Jesse Shiah, AgilePoint

search

>> Hello, I am Scott Hebner, the principal analyst for AI here at SiliconANGLE Media in theCUBE Research. Thank you for tuning in today. We here at SiliconANGLE and in theCUBE Research are so excited to be hosting on April 16th in industry unique digital summit. Dedicated to exploring best practices in building AI agents and agentic systems. Helping organizational leaders speed their way to ROI and competitive differentiation. Today, we'll preview one of the summit sponsors AgilePoint. They are an industry pioneer that is applying new innovations to help organizations operationalize enterprise scale, multivendor agentic workflows involving both human and digital workers. AgilePoint gives you the mature capabilities needed to create resilient operations within your organization. I had the opportunity along with my colleague George Gilbert here at theCUBE Research to get a deep dive 45 minute demonstration of their solution. Let me tell you, it's quite impressive and you'll learn more about that at the actual summit. With that, I'm so happy to have Jesse Shiah, the co-founder and CEO of AgilePoint, join us to preview what you'll learn from the summit from them and discuss these mature capabilities that I mentioned that can really be applied today to help you create highly resilient AI guided operations. Jesse, thanks so much for taking the time to be here. Really appreciate it.

>> Thank you, Scott. Really glad to have this opportunity.

>> I'm so glad we met up. I think it's been a couple of months now, hasn't it? The Power of LinkedIn, we've been spending a lot of time together and you got a really impressive group of people there. Impressive set of solutions and you've always impressed me with how customer-centric you are. It's always about the customer and what they're trying to do. This is going to be a great learning and experience for everyone at the summit and looking forward to having you there.

>> Thank you. I have to say that really enjoyed following the research, you and George working on the Agentic AI space and also on the in general, I know George also get into more like this architectural software engineering innovation needed to really be able to take Agentic AI into operations. We frequently highlight your research during our company meetings.

>> That's awesome. Thank you. George and I are two sides of the same coin. We come have different points of let's say expertise and knowledge, but together we work all the time together and I think we're learning a lot from each other. I appreciate the feedback on that. Jesse, tell us a little bit more about your journey from creating AgilePoint to where you are today. Just give us a little bit of a overview for people that may not be familiar with AgilePoint.

>> As a matter of fact, I know that like you say, I think you and George probably just learn about AgilePoint four or five months ago, but however, we have actually been around since 2003. This has been an amazing journey. The vision of Agentic AI, especially when it comes to how agentic can drive the business operations end to operations and automations. Actually, that's actually what we created AgilePoint back in 203. That was part of what we architected was that vision in mind. Part of that was influenced by my graduate study was related to hard real-time systems that's my advisor's main focus. That means that you're running some mission-critical hard real-time systems for industrial, for telecom, for military, for space, all these mission-critical systems. You can run into unexpected events. How does a system can adapt to that? Almost like you have to talk about self-healing to some extent. Part of my research was about, "Okay, can you use the AI to decide what to change on the system?" However, it's so mission-critical, almost like you are changing the software systems. One of the key criteria was cannot involve code changes. That means that no new code can be generated. No existing code can be changed. In order to ensure the trust of the assurance that was actually how we got started with the mindset. We built a company, a startup company and was being adapted by what I would call Global 100. Those are the largest companies across the globe and they build those mission-critical industrial control systems and everything, emergency command control systems. All year long take knowledge as part of their AS system. We saw the company in 1999, that's when I had the opportunity to come to Silicon Valley in 2000. Six months later I saw the dotcom crash. Then I went to a lot of gatherings afterward and then I heard a lot of CXOs were talking about, one thing really hit me was that they say, "We invest in a lot of technology, ERP, supply chain management, CRM content management systems, tons of them." They say, "We thought we invest in super technology, we should get agility and resiliency." Of course at that time because of dotcom, the key was about supply chain agility and resiliency. That really knock me awake I say, "Wow, people don't understand if you create application by code, you actually started chipping away agility and you're never going to get resiliency." That's really when we actually did the research when we work is no, we realized that you must apply abstraction. Programming is an abstraction, but that's not enough. We need to take abstraction all the way into the business domain. It's a tier abstraction and we call it holistic abstraction. Then using that as a way to bring in the model-driven architecture. That really simplify the design, simplify the execution, but in making that when you go to execution, there's actually no code generated, no compilation needed. That's why when we founded AgilePoint after I saw the dotcom crash, I say, "Wow. I'll know how there's an opportunity in the business computing domain." That way we found the AgilePoint. We knew we were going to be so ahead of the monkey because this holistic abstraction will simplify so much complexity.

>> Well, I think you're hitting on one of the challenges that I've seen in my 30 plus years in the industry, almost every technology transformation is people spend so much time creating a solution, whether it be a data warehouse, a data lake, whether it been old client server, even some of the cloud new as a service apps. You code it and by the time you get it out there, you're like when things have changed around you. What's great, and that actually is happening with generative AI and LLMs is they encode what we know about the past to help you predict the future, but they don't adapt with changing conditions. That's the power I think of these AI agents and agentic systems which you're helping to automate and abstract the complexity out of so people can, mere mortals can go off and do this and they can adapt it. I think it's a powerful combination and I think you're onto the right value prop, which is it's all about the resiliency and the adaptability. Because one thing you know about business and about the world is it's always changing.

>> Exactly. It was funny, after we created, we started building AgilePoint and we were eyeing that we say this is a bill for one day if AI is here. AI going to need to be able to adapt the running applications but without touching code, without changing code. Because if you change the code, you start impacting the trust. We say, "Okay, how can you get to that level?" This architecture, what we call is almost like the architecturally ubiquitous no code in a way to make this possible. The other thing is, and I think even today, so what we are seeing is in the whole AI community right now. I think a lot of this AI development just started five to seven years ago. It really needs a new technology foundation to really enable that AI being, I mean the opportunity with AI is it can discover things that we cannot by human, but we discover that to really realize the opportunity just that AOM can discover things. Now it's probably it's bridging to actions. The bridging to actions in the enterprise for example is our actual business operations. That is actually, and the business operations is now one system, the most valuable value creation business operations that end to end orchestration and automation that come across multiple different functional areas. That is really where that seems to become even more complex. We are seeing right now is I think what you just mentioned in the data warehouse in almost every technology until today we have been able to just using the traditional deterministic coding paradigm to do something and deliver something. The AI really right, the AI, especially when it comes to genetic AI now and arguably I tell people that I say we are going to touch on the ROI challenges and in the AI space as well. If you really want to overcome the ROI challenges for AI initiative. I will say the number one, the top prime candidate is be able to operationalize agentic AI for your end-to-end business orchestration. I always tell people that those may be 20% in terms of the number of the application, maybe only 20% of your business in your company, but they drive 80% of the business outcomes. You are able to operationalize on those end to end orchestrations. That's almost like you spend $1 now you can get $10 return versus now we're spending $10, we are now getting $1 return right now. That's why a lot of ROI challenges today.

>> One of the things that George and I love about what you guys are doing is you've moved up the value stack, meaning that you can wire together into agentic workflows and you can orchestrate AI agents that were either built in-house, they could be acquired in the open marketplace. You're very open in terms of what agents, co-workers that are being wired into these agentic workflows. You don't have to build it and deploy it in the same agentic framework you can take in from developing is a big marketplace of pre-built agents that are specialized. People are going to buy them, customize them and plug them into things like what you're delivering. I just think that's an incredibly strong value prop because like you just said, you're moving up the value stack, right? You say spend 1 get 10 where if I'm just building agents, maybe I'm spending three, but I still have to get value of those agents by putting them into a workflow. You've elevated that and I think that's really pretty cool.

>> Absolutely. I think about two months ago, Microsoft CEO Nadella had the podcast you mentioned about eventually all the business logic is going to go to AI agents. AI agent not going to be able to differentiate, "I'm dealing with this system right now. I need to work in this way. I'm dealing with another system now. I need to work another way." I think that's where I think in George's podcast why he talk about the importance of harmonization, acquire the tech stacks. As a matter of fact, when I talk about holistic abstraction, that's almost like a given because I think today our business operations has suffered by a lot of the siloed. The platforms are operating in silos. You go to a company, the Salesforce, the Salesforce platform, it's got a sales force team and they know how to create application Salesforce, SAP, the same thing. You maybe go to HR, you go to your ITSM, but the things they say, "Okay, now we need to create that end to end." Then they're just passing data between the platforms. Now back to what we were talking about Nadella say all the business logic need to go to AI agents. Business logic does not data because when you want to get to optimization is about agendas about goal-driven, right? The goal-driven means decision makings as well. The decision making is all about business logic. Many times of course data may drive that, but however is actually being able to make that decision change, the business logic changes. We just talked about the most important, I will say that anybody want to make your CEO and board really happy about your AI initiative right now is show that you are able to really make one of the end-to-end business orchestration that drive key business KPIs and make it agentic capable. Even before agentic, I would tell people that I think in the market, the automation market has been a term called hyperautomation that's only about automating as much as many as possible. They are different automation use cases. You can use different automation tool like RPA or maybe user workflow. The thing you actually build, you just going to build of unrelated automation applications.

>> Another layer of, another layer of.

>> I always tell people that I say you can build as many of that as many of those automations possible, but CEO may not notice that. If you build one end-to-end orchestrations, they will very quickly notice that because each of that drive key business KPIs. That's also one of the biggest challenge that we have been hearing from a lot of digital transformation leaders always about, "We spend a lot in digital transformation but we are not meeting our business objectives." Many times they found out they're building a lot of siloed automations but they're lacking the end-to-end orchestration and just because the natural of the platform silos. When it comes to AI agents, as you mentioned, same thing. We cannot have the agent specific only for platform. How do we be able to bring them to work together? It was actually very interesting just about three days ago, we had a session last week and we had a session with one of the and this is a project leader and they focus on, for example, IBM systems. He mentioned one examples, because we were talking about what this can do. I think this is a perfect, really, really intriguing. He said, "If I have a SharePoint workflow, let me clarify. If I have a SharePoint workflow and I want to add agent to it, but I want to add IBM agent, first thing I have to do is I have to migrate the SharePoint workflow to be IBM based before I can do that." Are you saying that was AgilePoint? You don't have to do that. I say no, you can keep your SharePoint workflow, I can bring IBM agent, I can bring another agent. Because the layer holistic abstraction. Holistic abstraction that enables instead of bow down the agent into the application directly into automation directly, you have a separate layer and that's where the value. I mean, the separate layer is where you have a clean separation and then that's where you can govern. That's also where you can democratize the business can go through this layer now, they don't have to learn about AI in order for the company to feel comfortable to start adapting AI. How many times I heard saying that, "We cannot afford to train out the business users on AI, so the adoption of that into real operation will get delayed." We say, "Why that has to be the case? When you actually insert AI directly into your operations, maybe that your business process and things, you put that right in front of the face of the business, well, then you need to do that." That's why we talk about the beauty of abstraction. Not only do holistic, you have a layer, so that's a separate layer of AI. We call it AI Control Tower layer, the framework. We actually developed that back in 2017 when we started seeing the AI, the open AI started doing something, I would say, "We've been waiting." Look at the market started moving the direction now. We actually created AI Control Tower framework layer back in 2017.

>> We're going to get into that. You hit on some really key points here. I, in the end, the business is being driven by people. People operate off of metrics and how they're going to get paid and whatever they do rarely is so siloed from a business perspective. They always have to work with other parts of the organization. Rarely do you see these processes if you will, they work within a process to achieve their goals. The process is multifunctional and you need to wire this all together to work across all that. To do that, we've been using the word harmonization layer, but it's really about semantics context meaning of how the data relates to each other so that it's not just data, like you said. It's semantical understanding and knowledge which over time is going to evolve into reasoning and is really going to get super powerful. Let me tell you just another aspect of what I like, what you're doing is the closed-loop stuff. As a long time marketing guy, I love the value prop that you have on your website It's so cleanly stated with the clear differentiation, I think it's really hard hitting. Let's bring up that graphic here with the statement on it. Can you just give us a little insight into what this really means? I think I get it, but I want to hear your explanation of it.

>> Well, closed-loop optimization is really the key critical capabilities to really realize the vision of autonomous business. What that means is say you have a process, you have an end-to-end orchestration and we are talking about not only each of them could actually be able to continuously self-improve it, right? Of course when we say that that takes a lot what we talk about to be, I mean what we believe it is not just what we are creating agent and with LLMs can give us some answers, but it's about bridging to actions. If you are not able to enable. Enable that, you actually realize the action in a way that can be trusted, that can improve, enhance the trust. You'll never get too close to optimization because you will not be able to adapt that. You will not be able to put that into your operations, right? That's what we are saying is in AgilePoint because this holistic abstraction from your infrastructure across your existing text stacks. We harmonize that and when we harmonize a different system more than today, we have done that for more than 120 systems already, the popular system being adopted in enterprise today. When we say we harmonize that, we actually abstract the data and also the business logic. The capabilities into that into a unified language will basically with this abstraction, we enable your IT stacks or your tech stack that can speak one language, but the one language that coherence semantics describes a metadata. That really enable one thing very important I think for agentic. I actually talked to our team, I say yeah, we always talk about this integration page on the website, talk about integrations. I say what we do is really more than just integration because for agentic integration tend to be just data integration, exchanging data. What we talk about is interoperability. It's interoperability now across systems. That's when AI, that's when agent can change the business logic in the way, but how do you enable? The thing is how can you change that? In a very interesting conversation with another analyst just about a couple of weeks ago and when we were talking about what we want to do and she said, "Yeah, I probably know what you are doing." "You have an agent, you have a code behind it. Then you have some kind of containers and then you just based on that, you then let me generating code." The first question, the first reaction I had was who can trust that? Who can trust that dynamic generated code and go to execution? Which enterprise can trust that? I think that's also why up to 90% of the AI PoC stays never going into production.

>> It's a trust issue.

>> Never going into production. The other thing is, okay, you run out of AOM to generate the code, it's the AOM going to generate the same code each time. What we are saying is what we do with this abstraction, so with this holistic abstraction, the age and when they want to change, they actually change the metadata. The model-driven engine that AgilePoint provides, be able to read the metadata change can dynamically invoke your IT systems and realize that. That's what we talk about again, right back to the research. I say that's why we have the benefit to really think about this so early 20 years ago. That's why we architected from day one with this mindset we say, "One day, one day will be here." However, before AI was here we say this architecture really going to democratize. Democratize because it just simplifies. It simplifies. We use a complexity, you end up with the business to work at same thing, the business can model, can change. They all work at the metadata layer. They never have to deal with the complexity of underlying systems. This also enables that. We also got to really sometimes we got a question rather say, "Well, people think about the data harmonization. They think about you have a data warehouse or data lake kind of platform. We say, "That's only the data for your analytics. We are talking about AgilePoint is abstracting the data for that context, the business context. That's really where the real AI data needed to train agents so that you enhance the trust of the agent can make decisions for you. If they are not trained by your business context, the business context, the most important thing is that end-to-end. The end-to-end business context, but the end-to-end business context is supported by multiple different systems. If you're not enabling every system to speak the same language, you're never going to capture that end-to-end business content in one language. I think that's really what the foundation to enable this coastal optimization because with that now then I can trust my agent making decisions and then whatever the decision they make can go back and retrain the AI agent again.

>> The industry has been talking about this notion of composable enterprises and composable business models and micro processes and things that you can always assemble and quickly adapt. I think with agentic AI and the evolution from gen AI to actually agents that can help you plan, can help you make decisions that can help you problem solve. Know how to autonomously work with others including humans, you start to get to the point where this can be a reality. I do like to me you have actually three layers of abstraction. You're going to help abstract the technology. You're helping to abstract the business semantics so that they can work together in understanding all that. This goes to my research that I've been doing, which shows about a 50/50 mix of organizations that either want to build their own AI agents or acquire and customize pre-built agents. I think you're very well positioned to support that level of abstraction too. You are bringing to life the value of this notion of an HR department of digital coworker agents that whether you build that talent in-house or whether you hire those agents externally, you can wire all that stuff together to meet your business semantics. There's a lot to unpack that makes it a very powerful concept. Again, that's agentic AI.

>> What we believe in is technology should become programmable because it changes so fast, especially in the age of AI now. We already seen how fast things are changing. What we talk about this holistic abstraction and then we talk about model-driven and really to bring them all together is the composable architecture. One thing we do just because we start from day one, our vision is about how do we help organization to leverage technology they adopted, but how can bring them together to create application and solution that deliver their business objective. Along the way does not, it simplified it in a simplified way and also reduce creating technical debts. Then at the same time it will actually naturally the application created in this abstracted, it will become future-proof and that also means it's more agile, it's more resilient. I know that sometimes when I talk to some ITD they say, "How can this be possible? You're talking about it will be easier and faster and you actually reduce all these burdens and everything." I say it's so hard to describe that in this current deterministic coding paradigm. You have to think it's a different paradigm of building applications. Like I say, humbly speak speaking I will say that for AI to take its maximum potential, we need this completely abstracted dynamic new foundation. When I put a new foundation, we actually started building it 20 years ago.

>> Let's briefly talk a little bit about your AgilePoint AI Control Tower framework. We're going to get into this in a lot more detail and show a demo most likely right at the actual summit. Why don't we bring up that chart and just why don't you briefly just touch on, I think there's four key attributes to the Control Tower. Again, we will go into this much deeper at the summit, but tell us about this.

>> Great. On the top, we will go from the top to the goal. The goal is co-suboptimization, The first one on the top, so we touch a number of different things. That means that we have this, so this abstraction harmonization across the test stacks. This abstraction harmonization is not specific to a specific platform. You can go to some major platform's vendors, they say, "We have abstraction, we have composability." It's just limited it to the platform. We want to make sure that you build it in a way that really is about, as a business, I don't care what technology I have and I can actually replace them. If I create application now, I'm leveraging those tech stack right now, but if I do replace them, it will not disrupt my whole applications. It will just actually plug something out and plug something, I mean, a new system in. The first one, that enable that holistic abstraction and harmonization and it's cross-platform so it's problem-agnostic, right? The composable architecture enables to create that what I just mentioned at the beginning, the most valuable business operations business application in the company today is that end-to-end orchestrations that drive business KPIs, that could be your customers onboarding, that could be your customer issue resolution. That could be your new product development that will always cut across multiple different departments so that's the first one. If you are able to build application this way, I'm still using under underlying, I'm still using my Salesforce, using my SAP, using my other systems. This application now can be built in a way that what we call resilient automation. Almost I'm preparing every automation or orchestration you build based on this architecture is agentic ready basically. Means that before agentic AI is here, is our customer found out that it's exception resistant. It means that when you run to exception, you can respond to exceptions and dynamically because application can still run, but now it was that exception resistant is really the same, it's the same capability that you make it that. Every automation you build is resilient. When I say resilient means that agent can adapt those operations, the automation orchestrations. What we were talking about earlier is it can adapt that at runtime in real time, but there's no one single night of coping generated. There's no one single night the code being changed. We start now, our customer, even before our customer they started, our first five customers at 2006 was Global Enterprises. They are the visionary company. They recognize what we are building and they adopted us when we only have three people in the company. All this customer over the year, even as every application built in 2006, they already have that platform-agnostic of composability. Also that resilient automation as a foundation. What that means is they have the foundation to go to agentic now directly.

>> This is a fascinating conversation. There's just so much potential. Again, as I've been preaching on my podcast, from what I've been learning to dozens and dozens of companies I've been talking to including you, is that these innovations are moving at warp speed. The game played on the field as Irvin Wladawsky-Berger said from IBM and that and the dugout. You got to keep up with this stuff and what you're delivering is just another leap in innovation that's coming. This is awesome. Look, Jesse, it's time for us to wrap up here, but thank you so much for being here. We are definitely going to dive deeper into the Control Tower and actually bring it to life and show what can actually do at the summit. Make sure you all get there. I really appreciate again, you taking the time to do this video and then of course to be a part of the summit for all of you, please make sure you visit AgilePoint.com to learn more. In the meantime, there's some really great content out there you'll learn from and get a little bit up to date before our session. Also you can reach out to Jesse on LinkedIn or on the AI Builder Summit portal, which is on thecube.net website. Finally, don't forget the summit on April 16th. As we've been pointing out here, we'll be exploring with the help of AgilePoint and many others, the key success factors in building AI agents and agentic systems. That includes how to deploy domain-specific knowledge, the semantics that Jesse was talking about, infuse explainability, AI decision intelligence and reasoning into these agents and building customized and integrated how you take those customized and pre-integrated agents and turn them into workflows. That's really about wiring up all these agentic workflows. Most importantly, and you touched on this, Jesse, is the theme of the summit is anchored in achieving trust in AI outcomes. Trust is now the currency of innovation. No trust, no ROI. Visit theCUBE.net to add the summit to your calendar and we'll see you at the summit. Bye for now.