Join Rob Bearden, CEO of Sema4, and Paul Ferguson, Vice President of Global Business Services at Emerson, as they explore the transformative potential of agentic artificial intelligence at the AI Agent Builder Summit. Gain insights into the collaboration between Sema4 and Emerson to develop AI agents that optimize business processes in accounting and beyond significantly.
In this session, hosted by SiliconANGLE Media and theCUBE Research, Bearden and Ferguson discuss how their collaboration has resulted in real-world outcomes in enterprise agentic AI. By leveraging Sema4's platform, Emerson has implemented AI agents to streamline customer payment processes, achieving an 80% automation rate and allowing knowledge workers to focus on more critical tasks. Bearden and Ferguson share their visions for AI agents in the enterprise and explore the journey of adopting and integrating agentic technology.
The conversation provides key takeaways on the adoption of AI agents within enterprises, according to Ferguson and industry analysts. Bearden highlights the importance of secure efficient AI platforms in fostering innovation and value creation. Ferguson emphasizes adopting a strategic approach to AI implementation, focusing on end-to-end automation and the integration of data-driven decisions within business processes.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
Sema4.ai Summit Portal. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For Sema4.ai Summit Portal
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for Sema4.ai Summit Portal.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
Sema4.ai Summit Portal. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to Sema4.ai Summit Portal
Please sign in with LinkedIn to continue to Sema4.ai Summit Portal. Signing in with LinkedIn ensures a professional environment.
>> Hello, welcome back to the AI Agent Builder Summit, featuring the proven best practices and solutions of agentic AI industry leaders. I am Scott Hebner, the principal analyst for AI here at SiliconANGLE Media and theCUBE Research. Thank you for tuning in today. In this session, we'll discuss real-world outcomes achieved by intelligently automating how work gets done with the help of AI agents. The discussion will feature Sema4, an industry pioneer shaping the future of enterprise agentic AI and Emerson, a global leader in automation technology and engineering services that helps the industrial sector create more efficient, sustainable, and innovative manufacturing processes. Today you'll learn why and how Emerson collaborated with Sema4 to develop AI agents that streamlined re-admittance matching for thousands of customer payments within their accounting processes. These agents now autonomously perform over 80% of what used to be done by knowledge workers freeing them up to focus on more business-critical tasks. To tell us more, I'm thrilled to introduce Rob Bearden, the CEO of Sema4 and Paul Ferguson, the Vice President of Global Business Services at Emerson. Paul, Rob, welcome and thank you so much for joining us today.
Rob Bearden
>> Well, Scott, it is great to be part of the AI Agent Builder Summit. This is a great format for us to talk a lot about the value creation. Paul, it's great to be here with you on this format. As part of the summit, we're actually announcing our new release of the Enterprise Edition that we've just made available, and we're also very excited to discuss how our safe services are going to enable our agentic platform to empower the enterprise to build, run and manage safe enterprise agents that execute knowledge work. So this is a great exciting time for us as a company. And Paul, again, the journey we've been on together, seeing what your vision around the transformation of value capture and creation that you have chartered for agents to do in your role as the VP of Global Business Services, you've focused on some very high value creation use cases and we as a company have gotten tremendous exposure to what the highest level of standards that our safe services need to deliver to be able to execute on the knowledge work that you've defined an agent has to go deliver. And your vision and how we build that into the platform has just been a tremendous journey and I'm excited for the viewers to learn more about how agents can transform value, the role they can play, and really the lessons learned that you've had so far in the journey of enabling AI agents to create value for your knowledge workers. So with that, why don't we jump in, Paul, maybe just start with the quick overview of as VP of Global Business Services, what that role is and does at Emerson. But what I'd really then like to do is talk about in your vision, what's the role of an agent across the enterprise as you see it.
Paul Ferguson
>> Yeah. Hey, thanks for having me on. I'm thrilled to be here. And so yeah, when you think of global business services, you can think of it as shared services as well, but instead of being individual pockets within individual regions, we focus on providing a global delivery network. It's usually around G&A type activities, so things like finance, accounting, procurement, HR, those types of activities. And a lot of those are going to be somewhat structured routine, but there's also a mix of knowledge work in there as well. There's a lot of value that the teams at the sites bring beyond just pushing buttons and doing transactional work. And so I think of agents within that space, within GBS, I think it definitely has a potential to be transformational. So not only you think about what's happened over the last at least 10 years, certainly you started with automation that was available within the RPs, then RPA came into place that helped you pick little pieces of the process to be able to automate. And now having agents available in the toolkit, we're really able to look at end-to-end automation and transformation. And that's where I think the real value is both from a service cost perspective as well as knowledge work in moving up the value chain. So lots of opportunity and I think we're very much in the early innings.
Rob Bearden
>> Yeah, that's great. Well, it's been truly impressive watching your vision go actually through the process of being implemented and actually moving into production and agents driving knowledge work in a very automated fashion and being able to measure the value that you're returning back into Emerson. So congratulations on your hard work becoming very much high value transformational value. Let me just transition a little bit. Scott, Paul has to do all the work that he does from a strategy through execution within use cases that he identifies across the enterprise. You talked to a number of different companies that have a variety of different strategies with agents. How do you see the enterprise looking at the role of an enterprise agent and what's your perspective on where the adoption carbon model is across the enterprise with agents?
Scott Hebner
>> Yeah, about 70 plus companies that I've talked to over the last six or so months. So I've learned a lot. And I can tell you the way Paul articulated it I think is very consistent with what I've heard. There is a view that it's transformative in nature, both in how work gets done through automation, intelligent automation between an agent and a human being as well as how organizations are going to make decisions. And I think what I've learned from all this is the generative AI was a step forward. It's good for AI assistants that are task-oriented. Maybe if it's a predictive model you can ask it to make a prediction for you, generate content, things of that nature, help you do information analysis and pull things together. It's very task-oriented and the ROI has been marginal, but it's been there. So I think it's a good step forward. I think agents are taking it to a whole nother level where they're not just task-oriented around a prompt. You can ask them to plan. They're goal oriented. What's the plan? How do I make the right decisions to achieve that plan? How do I problem solve? And often go off and autonomously get things done when human involvement's not feasible or warranted. And then I think agentic AI then takes that to the next level where it's not about an individual, it's about an organizational set of automation and decision-making where you need all these different agents that are collaborating within a workflow. That's how I've seen it kind of articulated if I bring it all together. And I think we're in the infancy of all this, especially when it comes to AI decision intelligence and explainability and getting these agents to not just help an individual but working in an organizational construct. And I think Sema4 and what you're doing, Paul at Emerson, you guys are ahead of the curve here. I think there's only 30 or so percent that have actually really gotten started with this so far.
Rob Bearden
>> Well, I think the common view certainly, and Scott that's very helpful and I think very consistent with the conversations Paul and I have, conversations that we have with our other customers and organizations that we're involved with. It's our view as a company is that the enterprise agents are really going to be the next generation of applications, and that's because they can perform very complex work. Levels of accuracy and predictability that drive returns of efficiency in a way, the traditional world of SaaS applications and analytics really couldn't without tremendous human involvement, human interpretation. And now enterprise AI agents can actually drive those predictive analytic models in a very repeatable, consistent way and do so with great accuracy and that creates a lot of value for knowledge workers. So Paul, when you look at the world, we've all, as you've said, the last 10 years, what we've had to operate in terms of our traditional business process and models and we look at the traditional SaaS and analytics applications and you've clearly demonstrate what the art of the possible is and delivered on that art of the possible with enterprise AI agents in very specific use cases across Emerson. Can you give us an insight and how you see and think about the relationship between agents, applications and analytics, and how do these things come together?
Paul Ferguson
>> Yeah, interesting. Look, I'll give you my perspective. Doesn't mean it's right by any means, right? So first of all, I think the foundation is data. You need access to data, preferably a centralized spot, that'd be great if you have a data lake or something. And so if you think of financial forecasting, like forecasting orders or forecasting sales, having that rich history that agents excel at, being able to read large amounts of data, understand the correlations and come up with predictions, phenomenal at that. It's tough though, and a lot of times in an enterprise world, you don't always have access to centralized data that's clean and perfect. And so there's a bunch of work that has to be done there, but there's definitely a relationship there between data agents and then I would throw on there a workflow or orchestration, and that kind of gets more into the process side of things. And so having a good, again, foundation around data and then orchestration or across the process, so what's step A, and what do you need to do in step A, maybe it's pull data from the data lake, maybe it's enter something in the system. And then step B, right? Again, could be a different action. And that's where the whole world and multi-agents coming to play. And so again, this is my perspective, I don't know how much it has to do with SaaS. I think right now, especially if you are in a modern environment where you have APIs or I guess now MCPs access to all that stuff, it can be very powerful, right? In the enterprise world though, especially a company like Emerson that has a rich history of acquisitions, that has a lot of different systems that go back decades, it's sometimes not as easy as just being able to pull APIs and then therefore being able to tap into RPA as part of the orchestration can be really beneficial. But I'm not sure. I look at SaaS as an applications and think if you were just to blow everything up and start with an AI agent environment, maybe you go directly into the database layer through MCPs and APIs and whatnot. I wouldn't be surprised and my kids, I don't know is this helpful or not, want to ask me what my first job was? And I say, "It was a paperboy." And they're like, "What's a paperboy?" And I'm like, "Yeah, nevermind." They may be saying the same thing about, you used to go into a system and punch in numbers and then hit submit? And so in the agent world, I think there's probably opportunity to go directly from that data layer orchestration decision-making directly to updating the source. And then of course, on top of that, once you have all that, being able to run analytics, it becomes much, much easier and probably more powerful and insightful. So that's my perspective at least.
Rob Bearden
>> Yeah. Well, I will tell you, I think that's very consistent with ours and you've clearly demonstrated the ability to define an outcome, build an agent that executes on that knowledge work pattern, and you've done it with great accuracy and efficiency. And you've done this now a couple of times across different parts of the business. Can you give us some insight into how you pick the first use case and or advice you would have as some of our viewers think about the criteria they should have in their first use case when they're thinking about putting together an enterprise AI agent strategy?
Paul Ferguson
>> So our first use case, and I'll say give credit to Rob and your team, one, really came to us with the vision and capabilities early on, and it was hard for me at first to really understand what the is between an AI agent, maybe some of the historical capabilities we had. And you and your team really helped educate me and enlighten us on what those opportunities are. And so once we understood what the framework was and the benefits. The use case we selected is one that's structural, very routine, but it requires a lot of matching of unstructured data. And so what we want to be able to do is say, okay, can an agent outperform a person in terms of doing that matching of unstructured data? And if so, we have many use cases in which that applies to. Maybe not within the same process itself, but many use cases across the organization in which you are essentially looking at data from source A and trying to match it against in source B. And then if it matches, do something. If it doesn't match, do something else. And so we picked the cash remittance process where we get payments in from customers, we got to match it to an open invoice on the AR subledger or invoice subledger. And so as a person, when we were doing it, we would get a hit rate about 20, 30%, or I'm sorry, we had an application that would get a hit rate of 20 to 30% and then the remaining 70%, 80% a person would've to go and resolve themselves. And so it was timely, it's kind of annoying because it's not something that we want to be able to spend our time on. And so as we ramped up the agent, we ended up getting around 70, 80% auto hit rate, which is phenomenal. And so what that did was able to allow the people that'd be not spend their time trying to do that remedial matching and free up their time to add more value add. And so it proved out that the technology is there. And so now really our opportunity is how do we take the lessons from that and apply it to broader end-to-end opportunities that have a higher impact?
Rob Bearden
>> That's great. I think that's very good insight and the process you went through to identify this and then be able to pressure test the value creation the use case can deliver was a really, really good process. So it's a really hats off to you and your team on that. Let me change gears. Paul, there's a lot of different kinds of agents that are available for enterprises to go work with. We really think about it in for primary categories in terms of the vendor landscape or types of agents that enterprises can go embrace, in no particular order of course, but there's purpose-built agents that do very specific sort of singular lane things and they do them extraordinarily well. And examples of that would be things like customer service or customer support or employee onboarding or password reset, but those are very purpose driven agents that within that specific use case, that agent does that role purpose built for that function in a world-class way and the enterprise will need to transform its process to fit that agent path. The other side of that trade is the do-it-yourself frameworks, the DIY frameworks that are traditionally largely open source driven, very prompt driven and developer centric to go build your own outcome and agent in a very definitional procedural way that you go build up and it's very, very good developer tooling to enable agent construction. The third category is sort of the ecosystem centric, and this is where we think about it as the very large SaaS application leaders have agent strategies that will complete the last mile of automation of a business process that their application traditionally manages in a very procedural way. Our category that we have focused on and we believe largely have created is the enterprise agent factory from a platform standpoint. It's a horizontal platform that allows the line of business owner and the developer to define an outcome. And our AI capability builds the agent, knows how to connect to the data, knows how to build the reasoning paths, knows how to properly secure that and how to deploy that in the environment that runs most efficiently secure with the proper governance. And so those are very different models. Purpose-built, do-it-yourself frameworks, ecosystem centric, which is extensions to SaaS apps, and then the enterprise agent factory, which is our horizontal platform. Can you give your perspectives on any of these approaches, but more particular where you see the benefits of having a horizontal platform capability in particular?
Paul Ferguson
>> Yeah, there's a lot there to try to unpack. I guess what I would look at, what's really been helpful from the Sema4 platform is the concept of the run book and how that applies to the line of business. Being able to describe kind of like a process narrative of the steps that the agent needs to take from an end-to-end process, combining it with actions is really powerful. And to be able to be horizontal and cut across multiple applications in that one run book is really powerful and been really helpful. And I can see as this progresses, it seems like every month is kind of a Dog Year in the AI agent space. And so as you start seeing more and more, I can see where you have a kind of agent to agent type relationship and being able to have a place to be able to monitor that across the enterprise has really been really helpful as well.
Rob Bearden
>> Perfect. Yep, that's great insight. Scott, from an analyst standpoint, do you have any perspective on the market landscape and what you're seeing emerge and how you think this plays through?
Scott Hebner
>> Yeah, I think actually the way you articulated it is very consistent, maybe some different terminology, but I think the way you segmented where these agents fall is right on. I think in the bigger scheme of things, I think businesses have gotten a taste of what they can do with generative AI. They're now looking for more profound value out of AI, new ways to work, new ways to build talent, new ways to make decisions, new ways to understand how the business actually operates. And the notion of an agent and agentic AI is that you're starting to mirror your HR organization. So you're going to be hiring "agents" and they're going to have different skills and different purposes within an organization. You have a planner and you have someone who's going to orchestrate things. You're going to have someone that helps make decisions, others make tasks. And that's what I love about the horizontal approach that you're taking is that it operates as a factory, which in many ways could mirror what you're trying to do to hire the right agents and pair them with your actual human people. And I think that's very, very powerful concept where obviously there'll be agents you can hire or buy that, like you said, are special purpose built, that you can customize. There's agents that you can build yourself. I think a lot of businesses are going to do that at a necessity because they have unique proprietary ways of doing things. Clearly the SaaS apps will be decomposed over time into agents, which will change the pricing models and everything. But I think you're onto a real notion here with the horizontal platform for sure. Very consistent.
Rob Bearden
>> Well, great. Well, so Paul, you have clearly executed and delivered real value, measurable economic value through enterprise agents. It's been great to be on that journey with you, but more importantly for our viewers today, what are some of the best lessons that you've learned just in your agentic journey journey, I don't want to call it the good, bad and the ugly, but what have you learned and any advice you'd pass along for people to put in their thought pattern as they're building their agent strategy?
Paul Ferguson
>> Yeah, I guess a couple of things. One is think big, to start, think end-to-end, not just a individual task, think across the whole end-to-end process, but then start with something that's within that end-to-end process that you continue to build on over time. I think it's tough to jump into something to say, we're going to just automate this whole thing end-to-end because it becomes a slog and you run into a lot of challenges with data and systems and all that stuff. But if you start small and continue to build on it, that's where I think we're going to see more success, being able to prove the value of it. Some of the challenges we had, I don't know, again, the challenges I don't think are the AI technology. It's getting access to data, being able to get the right access to systems and systems that may be somewhat old that don't have all the capabilities of a modern SaaS, for example. And then it's also level setting expectations. There's certainly some really high expectations out there in today's world about what agents can do, and all you got to do is flip the switch and it automatically works. Setting expectations and understanding that it's a journey and that you just got to start somewhere. We talked about how we selected our use case or the use case we selected to begin with. I don't know if you have to go searching for a unicorn use case that's going to be the one that drives the most value and you're just going to put an agent on it and call it done. I think you got to recognize it's a journey, but that the potential, it's greater today with AI agents than it was with any other technology that we had over the last decade.
Rob Bearden
>> Yeah. Well, Paul, congratulations on all the success you and your team have had inside of Emerson on, I would call it this first phase of enabling great outcomes and value creation through agentic approach and strategy. I think it's been very well done, very well executed. So impressive watching how you've approached it. As I think about now our path forward and what we as a company are doing and what we're very focused on, we're very focused on making sure that our platform really empowers the enterprise to build, run, and manage safe enterprise AI agents that very simply execute on knowledge work and do that reliably, predictably. And our approach and model is clearly by providing that horizontal platform that enables the developer and their line of business center to define an outcome. And then our platform leverages the Sema4 AI or Sai AI engine to build those agents to know how to connect to the data, to know how then to build the reasoning paths and deploy that agent in the best execution environment, whether it's behind your VPC or in a marketplace like Snowflake. And then to do that at highest standard. And the way we've approached this is through our set of safe services. And what's really table stakes for an agent is that they have to be secure because if an agent isn't secure, the rest doesn't matter. And we built from the ground up with the highest enterprise standard, the Sema4 control room, and that's what provides the enterprise grade security, this seamless scalability and really the complete lifecycle management that lets the AI agent operate securely and efficiently inside the enterprise IT infrastructure. But what really drives agents' value is to know how to do that work very accurately and how to take the actions that execute the work. And we connect our AI agents to the enterprise data to really help ensure that we create very accurate decision making through a zero copy access model, whether that's structured data, unstructured data documents, and what this does, it enables our agents to execute work with the highest points of accuracy, predictability, and reliability. And that's what drives the outcomes and delivers on the value. But what's also very important is our agents have to be able to be built fast. And I think that's what Paul and his team have been able to do so well, is they identify an opportunity, they have an idea, they see either a business model point of evolution that can create value, and they can turn that into an AI agent leveraging Sai, which is through the Sema4 studio, leveraging in natural language run books. And those are then truly AI driven agents and allow you to create those agents without writing a single line of code so you can build agents once and deploy them anywhere architecturally that's the most accurate and efficient way to do it. But through the lifecycle, agent execution has to be able to be explainable, and the agent reasoning should be very easy to see and understand and to really have that lineage to understand how the transparent reasoning works. And that's what our reasoning feature capabilities does through our workroom. You can really see how your agents made the decision, how they executed the work, what were the reasoning paths they went down. And that way you see that the agent worked reliably and predictably, and it just validates through our workroom, the process and procedure in which the agent executes how the work is done, and that its proper lifecycle has been properly captured, so the proper auditing and governance can be applied into those models. We worked very hard on that with Paul's team, Emerson, as well as another half dozen or so co-DEB partners and seeing great results leveraging our horizontal platform and our safe services. So it's been a great journey. We're excited about what the future of enterprise AI agents will do and be and become, and we're excited to be on our journey with Paul and our other enterprise customers. So Paul, thank you for being here with us.
Scott Hebner
>> Rob, you really just hit on the overriding theme of our summit, which is trust, right? Trust we believe is the currency of innovation going forward. No trust, no ROI, and it looks like you're doing a ton to address that for your customers. So awesome. And thank you, Paul. Thank you, Rob for being here. It was a great conversation. I'm sure it's going to be very valuable to our audience. I suspect there's a lot just getting started or looking to get started on their agentic AI journey. So I think this was an awesome conversation to help them all out. And for all of you, thank you for tuning in into the AI Agent Builder Summit. Please visit the Sema4 AI portal on the Summit website to access this video session so you can share it with your colleagues and on social media, and you can learn more about the Sema4 AI portfolio of offerings right out in that portal. And also visit Emerson.com to learn more about how they are advancing the world's most essential industries with automation technologies. And make sure you stay logged on for our next session and a CUBE special event later the spring to learn how Sema4.AI will be delivering agentic platforms that are even more accurate and predictive and thus trustworthy. So we'll see you again real soon. We are the leader in enterprise tech news and analysis. Bye for now.
>> Hello, welcome back to the AI Agent Builder Summit, featuring the proven best practices and solutions of agentic AI industry leaders. I am Scott Hebner, the principal analyst for AI here at SiliconANGLE Media and theCUBE Research. Thank you for tuning in today. In this session, we'll discuss real-world outcomes achieved by intelligently automating how work gets done with the help of AI agents. The discussion will feature Sema4, an industry pioneer shaping the future of enterprise agentic AI and Emerson, a global leader in automation technology and engineering services that helps the industrial sector create more efficient, sustainable, and innovative manufacturing processes. Today you'll learn why and how Emerson collaborated with Sema4 to develop AI agents that streamlined re-admittance matching for thousands of customer payments within their accounting processes. These agents now autonomously perform over 80% of what used to be done by knowledge workers freeing them up to focus on more business-critical tasks. To tell us more, I'm thrilled to introduce Rob Bearden, the CEO of Sema4 and Paul Ferguson, the Vice President of Global Business Services at Emerson. Paul, Rob, welcome and thank you so much for joining us today.
Rob Bearden
>> Well, Scott, it is great to be part of the AI Agent Builder Summit. This is a great format for us to talk a lot about the value creation. Paul, it's great to be here with you on this format. As part of the summit, we're actually announcing our new release of the Enterprise Edition that we've just made available, and we're also very excited to discuss how our safe services are going to enable our agentic platform to empower the enterprise to build, run and manage safe enterprise agents that execute knowledge work. So this is a great exciting time for us as a company. And Paul, again, the journey we've been on together, seeing what your vision around the transformation of value capture and creation that you have chartered for agents to do in your role as the VP of Global Business Services, you've focused on some very high value creation use cases and we as a company have gotten tremendous exposure to what the highest level of standards that our safe services need to deliver to be able to execute on the knowledge work that you've defined an agent has to go deliver. And your vision and how we build that into the platform has just been a tremendous journey and I'm excited for the viewers to learn more about how agents can transform value, the role they can play, and really the lessons learned that you've had so far in the journey of enabling AI agents to create value for your knowledge workers. So with that, why don't we jump in, Paul, maybe just start with the quick overview of as VP of Global Business Services, what that role is and does at Emerson. But what I'd really then like to do is talk about in your vision, what's the role of an agent across the enterprise as you see it.
Paul Ferguson
>> Yeah. Hey, thanks for having me on. I'm thrilled to be here. And so yeah, when you think of global business services, you can think of it as shared services as well, but instead of being individual pockets within individual regions, we focus on providing a global delivery network. It's usually around G&A type activities, so things like finance, accounting, procurement, HR, those types of activities. And a lot of those are going to be somewhat structured routine, but there's also a mix of knowledge work in there as well. There's a lot of value that the teams at the sites bring beyond just pushing buttons and doing transactional work. And so I think of agents within that space, within GBS, I think it definitely has a potential to be transformational. So not only you think about what's happened over the last at least 10 years, certainly you started with automation that was available within the RPs, then RPA came into place that helped you pick little pieces of the process to be able to automate. And now having agents available in the toolkit, we're really able to look at end-to-end automation and transformation. And that's where I think the real value is both from a service cost perspective as well as knowledge work in moving up the value chain. So lots of opportunity and I think we're very much in the early innings.
Rob Bearden
>> Yeah, that's great. Well, it's been truly impressive watching your vision go actually through the process of being implemented and actually moving into production and agents driving knowledge work in a very automated fashion and being able to measure the value that you're returning back into Emerson. So congratulations on your hard work becoming very much high value transformational value. Let me just transition a little bit. Scott, Paul has to do all the work that he does from a strategy through execution within use cases that he identifies across the enterprise. You talked to a number of different companies that have a variety of different strategies with agents. How do you see the enterprise looking at the role of an enterprise agent and what's your perspective on where the adoption carbon model is across the enterprise with agents?
Scott Hebner
>> Yeah, about 70 plus companies that I've talked to over the last six or so months. So I've learned a lot. And I can tell you the way Paul articulated it I think is very consistent with what I've heard. There is a view that it's transformative in nature, both in how work gets done through automation, intelligent automation between an agent and a human being as well as how organizations are going to make decisions. And I think what I've learned from all this is the generative AI was a step forward. It's good for AI assistants that are task-oriented. Maybe if it's a predictive model you can ask it to make a prediction for you, generate content, things of that nature, help you do information analysis and pull things together. It's very task-oriented and the ROI has been marginal, but it's been there. So I think it's a good step forward. I think agents are taking it to a whole nother level where they're not just task-oriented around a prompt. You can ask them to plan. They're goal oriented. What's the plan? How do I make the right decisions to achieve that plan? How do I problem solve? And often go off and autonomously get things done when human involvement's not feasible or warranted. And then I think agentic AI then takes that to the next level where it's not about an individual, it's about an organizational set of automation and decision-making where you need all these different agents that are collaborating within a workflow. That's how I've seen it kind of articulated if I bring it all together. And I think we're in the infancy of all this, especially when it comes to AI decision intelligence and explainability and getting these agents to not just help an individual but working in an organizational construct. And I think Sema4 and what you're doing, Paul at Emerson, you guys are ahead of the curve here. I think there's only 30 or so percent that have actually really gotten started with this so far.
Rob Bearden
>> Well, I think the common view certainly, and Scott that's very helpful and I think very consistent with the conversations Paul and I have, conversations that we have with our other customers and organizations that we're involved with. It's our view as a company is that the enterprise agents are really going to be the next generation of applications, and that's because they can perform very complex work. Levels of accuracy and predictability that drive returns of efficiency in a way, the traditional world of SaaS applications and analytics really couldn't without tremendous human involvement, human interpretation. And now enterprise AI agents can actually drive those predictive analytic models in a very repeatable, consistent way and do so with great accuracy and that creates a lot of value for knowledge workers. So Paul, when you look at the world, we've all, as you've said, the last 10 years, what we've had to operate in terms of our traditional business process and models and we look at the traditional SaaS and analytics applications and you've clearly demonstrate what the art of the possible is and delivered on that art of the possible with enterprise AI agents in very specific use cases across Emerson. Can you give us an insight and how you see and think about the relationship between agents, applications and analytics, and how do these things come together?
Paul Ferguson
>> Yeah, interesting. Look, I'll give you my perspective. Doesn't mean it's right by any means, right? So first of all, I think the foundation is data. You need access to data, preferably a centralized spot, that'd be great if you have a data lake or something. And so if you think of financial forecasting, like forecasting orders or forecasting sales, having that rich history that agents excel at, being able to read large amounts of data, understand the correlations and come up with predictions, phenomenal at that. It's tough though, and a lot of times in an enterprise world, you don't always have access to centralized data that's clean and perfect. And so there's a bunch of work that has to be done there, but there's definitely a relationship there between data agents and then I would throw on there a workflow or orchestration, and that kind of gets more into the process side of things. And so having a good, again, foundation around data and then orchestration or across the process, so what's step A, and what do you need to do in step A, maybe it's pull data from the data lake, maybe it's enter something in the system. And then step B, right? Again, could be a different action. And that's where the whole world and multi-agents coming to play. And so again, this is my perspective, I don't know how much it has to do with SaaS. I think right now, especially if you are in a modern environment where you have APIs or I guess now MCPs access to all that stuff, it can be very powerful, right? In the enterprise world though, especially a company like Emerson that has a rich history of acquisitions, that has a lot of different systems that go back decades, it's sometimes not as easy as just being able to pull APIs and then therefore being able to tap into RPA as part of the orchestration can be really beneficial. But I'm not sure. I look at SaaS as an applications and think if you were just to blow everything up and start with an AI agent environment, maybe you go directly into the database layer through MCPs and APIs and whatnot. I wouldn't be surprised and my kids, I don't know is this helpful or not, want to ask me what my first job was? And I say, "It was a paperboy." And they're like, "What's a paperboy?" And I'm like, "Yeah, nevermind." They may be saying the same thing about, you used to go into a system and punch in numbers and then hit submit? And so in the agent world, I think there's probably opportunity to go directly from that data layer orchestration decision-making directly to updating the source. And then of course, on top of that, once you have all that, being able to run analytics, it becomes much, much easier and probably more powerful and insightful. So that's my perspective at least.
Rob Bearden
>> Yeah. Well, I will tell you, I think that's very consistent with ours and you've clearly demonstrated the ability to define an outcome, build an agent that executes on that knowledge work pattern, and you've done it with great accuracy and efficiency. And you've done this now a couple of times across different parts of the business. Can you give us some insight into how you pick the first use case and or advice you would have as some of our viewers think about the criteria they should have in their first use case when they're thinking about putting together an enterprise AI agent strategy?
Paul Ferguson
>> So our first use case, and I'll say give credit to Rob and your team, one, really came to us with the vision and capabilities early on, and it was hard for me at first to really understand what the is between an AI agent, maybe some of the historical capabilities we had. And you and your team really helped educate me and enlighten us on what those opportunities are. And so once we understood what the framework was and the benefits. The use case we selected is one that's structural, very routine, but it requires a lot of matching of unstructured data. And so what we want to be able to do is say, okay, can an agent outperform a person in terms of doing that matching of unstructured data? And if so, we have many use cases in which that applies to. Maybe not within the same process itself, but many use cases across the organization in which you are essentially looking at data from source A and trying to match it against in source B. And then if it matches, do something. If it doesn't match, do something else. And so we picked the cash remittance process where we get payments in from customers, we got to match it to an open invoice on the AR subledger or invoice subledger. And so as a person, when we were doing it, we would get a hit rate about 20, 30%, or I'm sorry, we had an application that would get a hit rate of 20 to 30% and then the remaining 70%, 80% a person would've to go and resolve themselves. And so it was timely, it's kind of annoying because it's not something that we want to be able to spend our time on. And so as we ramped up the agent, we ended up getting around 70, 80% auto hit rate, which is phenomenal. And so what that did was able to allow the people that'd be not spend their time trying to do that remedial matching and free up their time to add more value add. And so it proved out that the technology is there. And so now really our opportunity is how do we take the lessons from that and apply it to broader end-to-end opportunities that have a higher impact?
Rob Bearden
>> That's great. I think that's very good insight and the process you went through to identify this and then be able to pressure test the value creation the use case can deliver was a really, really good process. So it's a really hats off to you and your team on that. Let me change gears. Paul, there's a lot of different kinds of agents that are available for enterprises to go work with. We really think about it in for primary categories in terms of the vendor landscape or types of agents that enterprises can go embrace, in no particular order of course, but there's purpose-built agents that do very specific sort of singular lane things and they do them extraordinarily well. And examples of that would be things like customer service or customer support or employee onboarding or password reset, but those are very purpose driven agents that within that specific use case, that agent does that role purpose built for that function in a world-class way and the enterprise will need to transform its process to fit that agent path. The other side of that trade is the do-it-yourself frameworks, the DIY frameworks that are traditionally largely open source driven, very prompt driven and developer centric to go build your own outcome and agent in a very definitional procedural way that you go build up and it's very, very good developer tooling to enable agent construction. The third category is sort of the ecosystem centric, and this is where we think about it as the very large SaaS application leaders have agent strategies that will complete the last mile of automation of a business process that their application traditionally manages in a very procedural way. Our category that we have focused on and we believe largely have created is the enterprise agent factory from a platform standpoint. It's a horizontal platform that allows the line of business owner and the developer to define an outcome. And our AI capability builds the agent, knows how to connect to the data, knows how to build the reasoning paths, knows how to properly secure that and how to deploy that in the environment that runs most efficiently secure with the proper governance. And so those are very different models. Purpose-built, do-it-yourself frameworks, ecosystem centric, which is extensions to SaaS apps, and then the enterprise agent factory, which is our horizontal platform. Can you give your perspectives on any of these approaches, but more particular where you see the benefits of having a horizontal platform capability in particular?
Paul Ferguson
>> Yeah, there's a lot there to try to unpack. I guess what I would look at, what's really been helpful from the Sema4 platform is the concept of the run book and how that applies to the line of business. Being able to describe kind of like a process narrative of the steps that the agent needs to take from an end-to-end process, combining it with actions is really powerful. And to be able to be horizontal and cut across multiple applications in that one run book is really powerful and been really helpful. And I can see as this progresses, it seems like every month is kind of a Dog Year in the AI agent space. And so as you start seeing more and more, I can see where you have a kind of agent to agent type relationship and being able to have a place to be able to monitor that across the enterprise has really been really helpful as well.
Rob Bearden
>> Perfect. Yep, that's great insight. Scott, from an analyst standpoint, do you have any perspective on the market landscape and what you're seeing emerge and how you think this plays through?
Scott Hebner
>> Yeah, I think actually the way you articulated it is very consistent, maybe some different terminology, but I think the way you segmented where these agents fall is right on. I think in the bigger scheme of things, I think businesses have gotten a taste of what they can do with generative AI. They're now looking for more profound value out of AI, new ways to work, new ways to build talent, new ways to make decisions, new ways to understand how the business actually operates. And the notion of an agent and agentic AI is that you're starting to mirror your HR organization. So you're going to be hiring "agents" and they're going to have different skills and different purposes within an organization. You have a planner and you have someone who's going to orchestrate things. You're going to have someone that helps make decisions, others make tasks. And that's what I love about the horizontal approach that you're taking is that it operates as a factory, which in many ways could mirror what you're trying to do to hire the right agents and pair them with your actual human people. And I think that's very, very powerful concept where obviously there'll be agents you can hire or buy that, like you said, are special purpose built, that you can customize. There's agents that you can build yourself. I think a lot of businesses are going to do that at a necessity because they have unique proprietary ways of doing things. Clearly the SaaS apps will be decomposed over time into agents, which will change the pricing models and everything. But I think you're onto a real notion here with the horizontal platform for sure. Very consistent.
Rob Bearden
>> Well, great. Well, so Paul, you have clearly executed and delivered real value, measurable economic value through enterprise agents. It's been great to be on that journey with you, but more importantly for our viewers today, what are some of the best lessons that you've learned just in your agentic journey journey, I don't want to call it the good, bad and the ugly, but what have you learned and any advice you'd pass along for people to put in their thought pattern as they're building their agent strategy?
Paul Ferguson
>> Yeah, I guess a couple of things. One is think big, to start, think end-to-end, not just a individual task, think across the whole end-to-end process, but then start with something that's within that end-to-end process that you continue to build on over time. I think it's tough to jump into something to say, we're going to just automate this whole thing end-to-end because it becomes a slog and you run into a lot of challenges with data and systems and all that stuff. But if you start small and continue to build on it, that's where I think we're going to see more success, being able to prove the value of it. Some of the challenges we had, I don't know, again, the challenges I don't think are the AI technology. It's getting access to data, being able to get the right access to systems and systems that may be somewhat old that don't have all the capabilities of a modern SaaS, for example. And then it's also level setting expectations. There's certainly some really high expectations out there in today's world about what agents can do, and all you got to do is flip the switch and it automatically works. Setting expectations and understanding that it's a journey and that you just got to start somewhere. We talked about how we selected our use case or the use case we selected to begin with. I don't know if you have to go searching for a unicorn use case that's going to be the one that drives the most value and you're just going to put an agent on it and call it done. I think you got to recognize it's a journey, but that the potential, it's greater today with AI agents than it was with any other technology that we had over the last decade.
Rob Bearden
>> Yeah. Well, Paul, congratulations on all the success you and your team have had inside of Emerson on, I would call it this first phase of enabling great outcomes and value creation through agentic approach and strategy. I think it's been very well done, very well executed. So impressive watching how you've approached it. As I think about now our path forward and what we as a company are doing and what we're very focused on, we're very focused on making sure that our platform really empowers the enterprise to build, run, and manage safe enterprise AI agents that very simply execute on knowledge work and do that reliably, predictably. And our approach and model is clearly by providing that horizontal platform that enables the developer and their line of business center to define an outcome. And then our platform leverages the Sema4 AI or Sai AI engine to build those agents to know how to connect to the data, to know how then to build the reasoning paths and deploy that agent in the best execution environment, whether it's behind your VPC or in a marketplace like Snowflake. And then to do that at highest standard. And the way we've approached this is through our set of safe services. And what's really table stakes for an agent is that they have to be secure because if an agent isn't secure, the rest doesn't matter. And we built from the ground up with the highest enterprise standard, the Sema4 control room, and that's what provides the enterprise grade security, this seamless scalability and really the complete lifecycle management that lets the AI agent operate securely and efficiently inside the enterprise IT infrastructure. But what really drives agents' value is to know how to do that work very accurately and how to take the actions that execute the work. And we connect our AI agents to the enterprise data to really help ensure that we create very accurate decision making through a zero copy access model, whether that's structured data, unstructured data documents, and what this does, it enables our agents to execute work with the highest points of accuracy, predictability, and reliability. And that's what drives the outcomes and delivers on the value. But what's also very important is our agents have to be able to be built fast. And I think that's what Paul and his team have been able to do so well, is they identify an opportunity, they have an idea, they see either a business model point of evolution that can create value, and they can turn that into an AI agent leveraging Sai, which is through the Sema4 studio, leveraging in natural language run books. And those are then truly AI driven agents and allow you to create those agents without writing a single line of code so you can build agents once and deploy them anywhere architecturally that's the most accurate and efficient way to do it. But through the lifecycle, agent execution has to be able to be explainable, and the agent reasoning should be very easy to see and understand and to really have that lineage to understand how the transparent reasoning works. And that's what our reasoning feature capabilities does through our workroom. You can really see how your agents made the decision, how they executed the work, what were the reasoning paths they went down. And that way you see that the agent worked reliably and predictably, and it just validates through our workroom, the process and procedure in which the agent executes how the work is done, and that its proper lifecycle has been properly captured, so the proper auditing and governance can be applied into those models. We worked very hard on that with Paul's team, Emerson, as well as another half dozen or so co-DEB partners and seeing great results leveraging our horizontal platform and our safe services. So it's been a great journey. We're excited about what the future of enterprise AI agents will do and be and become, and we're excited to be on our journey with Paul and our other enterprise customers. So Paul, thank you for being here with us.
Scott Hebner
>> Rob, you really just hit on the overriding theme of our summit, which is trust, right? Trust we believe is the currency of innovation going forward. No trust, no ROI, and it looks like you're doing a ton to address that for your customers. So awesome. And thank you, Paul. Thank you, Rob for being here. It was a great conversation. I'm sure it's going to be very valuable to our audience. I suspect there's a lot just getting started or looking to get started on their agentic AI journey. So I think this was an awesome conversation to help them all out. And for all of you, thank you for tuning in into the AI Agent Builder Summit. Please visit the Sema4 AI portal on the Summit website to access this video session so you can share it with your colleagues and on social media, and you can learn more about the Sema4 AI portfolio of offerings right out in that portal. And also visit Emerson.com to learn more about how they are advancing the world's most essential industries with automation technologies. And make sure you stay logged on for our next session and a CUBE special event later the spring to learn how Sema4.AI will be delivering agentic platforms that are even more accurate and predictive and thus trustworthy. So we'll see you again real soon. We are the leader in enterprise tech news and analysis. Bye for now.
Scott Hebner