Exploring the Advancements and Challenges in AI Agent Deployment
John Nay, founder and Chief Executive Officer of Norm Ai, joins theCUBE's special presentation with NYSE Wired, focusing on the upcoming Artificial Intelligence Agent Conference 2025. Hosted by John Furrier, co-founder and co-Chief Executive Officer of SiliconANGLE Media, this insightful discussion covers the pivotal developments in AI infrastructure and the regulatory complexities faced by enterprises.
In this episode, Nay shares their expertise in regulatory AI infrastructure, particularly as it pertains to AI agent deployment in highly regulated sectors. The conversation, hosted by Furrier, delves into the evolving landscape of AI technology, compliance challenges, and the strategic initiatives underway at Norm Ai to address the pressing issues surrounding AI deployment. The discussion provides valuable insights for both technology and policy influencers.
Key takeaways from the discussion include the emphasis on the need for dynamic, real-time compliance frameworks that align with regulatory standards, as emphasized by Nay. Furthermore, the episode highlights how enterprises can leverage existing compliance structures to integrate AI technologies more effectively, offering a glimpse into the future of AI agent scalability and regulation. The conversation underscores the importance of bridging the gap between engineering, policy, and technology for sustainable AI innovation.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Agent Conference. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: AI Agent Conference.
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 theCUBE + NYSE Wired: AI Agent Conference
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 theCUBE + NYSE Wired: AI Agent Conference.
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
theCUBE + NYSE Wired: AI Agent Conference. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: AI Agent Conference.
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 theCUBE + NYSE Wired: AI Agent Conference
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: AI Agent Conference. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Joe Moura, CrewAI
Exploring the Advancements and Challenges in AI Agent Deployment
John Nay, founder and Chief Executive Officer of Norm Ai, joins theCUBE's special presentation with NYSE Wired, focusing on the upcoming Artificial Intelligence Agent Conference 2025. Hosted by John Furrier, co-founder and co-Chief Executive Officer of SiliconANGLE Media, this insightful discussion covers the pivotal developments in AI infrastructure and the regulatory complexities faced by enterprises.
In this episode, Nay shares their expertise in regulatory AI infrastructure, particularly as it pertains to AI agent deployment in highly regulated sectors. The conversation, hosted by Furrier, delves into the evolving landscape of AI technology, compliance challenges, and the strategic initiatives underway at Norm Ai to address the pressing issues surrounding AI deployment. The discussion provides valuable insights for both technology and policy influencers.
Key takeaways from the discussion include the emphasis on the need for dynamic, real-time compliance frameworks that align with regulatory standards, as emphasized by Nay. Furthermore, the episode highlights how enterprises can leverage existing compliance structures to integrate AI technologies more effectively, offering a glimpse into the future of AI agent scalability and regulation. The conversation underscores the importance of bridging the gap between engineering, policy, and technology for sustainable AI innovation.
>> Welcome back. I'm John Furrier, host of theCUBE. We are here at theCUBE's New York Stock Exchange studio. Of course, we have our Palo Alto studio connecting Silicon Valley to Wall Street. This is our mixture of experts series also featuring, as a preview, the upcoming AI Agent Conference, agentconference.com. It's a growing conference that started as a community meetup. Now it's a full-blown conference. All the experts, these are the players making it happen. We got a great guest here, Joe Moura who's the CEO and founder of CrewAI who built this company when it wasn't fashionable to have agents. Now everyone wants agents. They want to understand it. Joe, great to have you on this program with us and the NYC Wired community and the NYC Wired program at CUBE Original. agentconference.com, he's got a great URL, Simon does, so certainly that's a great draw. Agents are hot. Welcome to theCUBE.
Joe Moura
>> Thank you so much, John. I really appreciate it. A blast to be here and chatting with you today and agents are hot. There's so much that we can talk about that that's insane. I mean, we're seeing this being adopted from all sorts of customers, all different sides, all different industries. I think when people think about it, it truly can be as transformation as the internet was or even more, right? And I think the main differences is we're seeing a lot of the ROI happening way faster as well. So very exciting times.
John Furrier
>> One of the things we've seen with GenAI, we've seen the movie before with cloud on the previous generation. The AI native world now is kind of waking up to the reality that one, it's a total paradigm shift. The threshold has been crossed, the excitement and the strategy formulations not so much a risk factor as much as it's an execution risk. We're seeing that with generative AI. But now you've got the agents, which is a very frothy market because there's new strategy discussions around how agents are working, but still execution is involved because they're end-to-end workflows. They have autonomous capabilities, there's data involved. So you start to see the rise of the agentic infrastructure and the agentic applications. Claude bought Moltbook recently, OpenClaude now part of OpenAI. You're starting to see that agents are going to start talking to agents as part of work, but it's got to be powered by some infrastructure. So it almost feels like a DevOps moment for me because, in the cloud world, you saw that DevOps movement, the clouderati, the OGs, the ones that were pure, pure DevOps and then built a cloud. But that's now the foundation. Now you have an agent ops, agent infrastructure and the applications and it's very clear that the interfaces are changing. So where's the value? Where's the value capture? Where do I program? What do I invest in? These are all questions. What's your reaction?
Joe Moura
>> I got to say, you got in even the head there, right? And I mean, you put yourself in the shoes of the CIOs trying to make decisions right now, and it is insane. It's so busy because imagine you have every vendor on earth calling you and saying like, "Hey, you're using Salesforce. Why don't you use Agentforce? You're using ServiceNow. Why not use the ServiceNow agents?"
So you have the hyperscalers on one side, you have those incumbents on the other, you have the model providers now trying. I think everyone's realizing how much of a big shift this can be and everyone's trying to get the bigger piece of the pie that they can get. But there's one thing to be said where the market's moving so fast and these companies, they want to have progress in ROI, but they cannot pick winners. It's moving too fast for you to pick a winner. So what we're seeing out there is companies really favoring not only open source software but also favoring things that creates optionality. I think that's a big thing into all the C-level leaders right now is how I make progress by keeping the door open.
John Furrier
>> Yeah. And like I was mentioning with the clouds, the cloud migration, cloud growth, that was driven by open source. Now you have the Linux Foundation picking up MCP A2A, and now you got OpenClaw in there. You're starting to see the formation of another level. What's your view on this? What's your vision of how open source is going to change or accelerate the AI native movement?
Joe Moura
>> Yeah, I think you've got the word right. The thing there is operations, right? I think if you look at the last couple of years, there has been almost like a runway to the bottom in terms of making it extremely easy for people to build these agents. Like building, it's almost becoming a commodity. There are so many options that you can do in there. But what a lot of people don't realize, or people are starting to realize now, is that there is zero ROI. There is zero impact on the bottom line if you just build that. The value actually gets from running these things into production under a certain scale. And for you to get to that stage, then a bunch of other requirements starts to pop up from your regular suspects on enterprise, where these companies now need to start thinking about governance, about data privacy, around PII, or road-based level of access. And to your point, all that infrastructure that was built in the cloud and how that translates now into this enterprise environment, and there's even other things that in there that needs to be considered. But what we're seeing is there's definitely a new stack that is coming up, building on top of that data layer foundation, but now focus on that operational layer for all these agents that work around the MCP, work around the HWA, but allows you not only building but deploying, getting those things to value and actually observing not only their executions but the business outcome.
John Furrier
>> I always complimented Jensen Huang because he's the only CEO that used the word computer science in keynotes more than any other CEO. We talk about operating systems all the time on theCUBE. We love that tech. We're seeing an operating system kind of vibe with agents. It's not a chatbot anymore. You're starting to see the movement to multi-agent, agent-to-agent delegations involved. It's more than an API. So explain what you guys are doing, because I know you're doing some work in the area. How are you taking the open source movement, bringing that into a rapid production workflow or track when you have now this operating environment for multiple agents. And they could be agents from different applications and companies, so it feels like an API moment going on, but a whole nother level and what's the difference between the two?
Joe Moura
>> A thousand percent. I think we were talking about this as an agent management platform. So if you look at what we have been doing commercially, it's a lot about rooting that platform that helps these enterprises get the progress that they need, leverages much of their existing infrastructure and giving them all the optionality. So what that means is you look at many of our customers from AB InBev, one of the largest breweries in the world, actually the largest, to the PepsiCo, the Johnson & Johnsons and everyone that we have been working with. And you can see they all have similar needs where they want to deploy the systems, one, on-premises, so they're benefiting from their currently cloud infrastructure, so building on top of that. Integrating on their data providers, integrating with their Databricks, integrating with their Snowflake, and we basically create a platform that runs on top of that infrastructure that allows them to build, allows them to deploy, either with code or no code, but making sure that you get those reusable building blocks. So as you adopt this agent on a global scale, you're not rebuilding the same agents twice. You're not rebuilding the same two strikes, but instead accelerating towards that. That's the common theme that we keep hearing. And a big thing that you mentioned yourself is this idea of agents will talk over the wire. I think, you go back a year ago, a lot of people would not believe on that, but it's actually happening. We have some of this Fortune 500 companies using CrewAI as a platform to basically orchestrate ServiceNow agents or Salesforce agents. A2A is actually start to getting a lot of traction in there. So in the end of the day, I think there's four main things that these leaders care about right now. One is central visibility and control, two is reusability of pieces, so they allow technical and non-technical teams to adopt this. And then three is basically making sure they're leveraging all that flexibility, all the existing systems, existing models and everything that comes with that. And the fourth and final one would be this idea of talking with agents over the wire that seems to actually be accelerating extremely fast.
John Furrier
>> Yeah. And it's all those things under the covers, like identity you mentioned earlier. These are things that have got to be buttoned up, they're not for the faint of heart. You got to make sure all that security's in place. I want to get your thoughts. I just wrote a post yesterday on siliconangle.com around the Anthropic valuation, the title is, The $380 Billion Orchestration Bet: Understanding the Coding Wedge as AI Labs and Developers Move Beyond the Model Layer. And here's the comment, "The coding wedge has become a strategic beachhead where winning developers in the coding workflows represents the point of leverage to win AI enterprise adoption and broader orchestration control. Coding emerged as the first domain where autonomous agents were reliable enough for real production use, becoming the practical proving ground for capability and costs. Whoever controls how developers build with AI gains control of this orchestration layer, which then extends from developers to all knowledge workers and enterprise workflows." What's your take on that? Do you see it the same? How would you react to that? Do you see a similar dynamic where people are trying to figure out where's the leverage? What's your thoughts?
Joe Moura
>> I mean, I see the same way because in the end of the day, if you look at engineering, it's funny because no one thought that data would be the first target, but it turns out to be an amazing target. And what we're seeing is many people adopting. I mean, engineering use cases, like development life cycle use cases are all over the place. And the ROI is there. It's real. I mean, you look from companies on our size and how much we're leveraging AI to help us write and ship software nowadays, and you look at bigger companies as well, that is happening. The one thing, though, that I would add to there is as you lower the thresholds to build these things and you bring all ... You just make it easier for people to get these things out there. There's a few other things that becomes way top of mind and that is security, governance, all those things now need to come all the way to the left. You cannot have your entire team or your CEO now vibe coding apps and figuring out how to get these things out there without bringing all those concerns that, as you mentioned, is not for the fainted heart, all the way to the left. So that make sure that out of the gate you're taking care of all those things. But I agree with you. I feel like on engineering, especially for the folks that are on the cutting edge, they're the ones that experienced this first, this idea where the job is changing and you're going from doing a lot of the work yourself to actually delegating, kicking off all these different groups and clusters of agents to do the work and you actually managing them. But I think we're probably less than one or two years away from many other different horizontals within these companies actually start to feeling the same thing.
John Furrier
>> I had Sam Partee on from Arcade last week on an area with Harris and Chase, and we're talking about how fast this is moving. Can you share from your perspective, just in the past eight months, the velocity of change and the different approaches that people are taking? OpenAI saying, "We're the horizontal plane. Use us." Anthropic's saying, "We'll, vertically integrate." Isn't that either/or, and how should we be ... People should be thinking about how to leverage agents. If you've got your workflow and you've got your data mode, why wouldn't you want to go end-to-end? Or what do I want to put that into as a service if I'm AI native? Is it a different approach or horses for courses? What's your opinion and the speed?
Joe Moura
>> Yeah, no, that's the great thing. I actually spent some time talking with Sam Altman about this himself and we're talking about what is the last mile? Because the models are doing amazing work, and as these models get better and better, there's definitely a movement where people are removing the scaffolding. Where you would have, in the past, people putting all these sorts of scaffolding to make sure that these agents would behave in a certain way, as the models are getting better, people were finding themselves removing the scaffoldings and allocating more to the model. But when you actually look into these enterprises and you try to understand how you get them from that model being ready to actually impact on the bottom line to actual business outcome, the last mile is more like a hundred miles. What's happening in there? And we talking about a few things. We're talking about the security and the governance, but there's the integrations, the data prep, there's so much that is happening in there. So you got to have some appreciation for the fact that these companies want to move so fast and you want to make sure that we get them to accelerate and get these things out in production. But the reality on the day to day is you look at some of our customers and some of them have over 30 instances of SAP, and like, all right, how you get a model to understand and work through that. So I think that's going to take a while. But I see the acceleration. I mean, I look at our own numbers. Our average selling price has basically more than 11x. Like this year, you look at some of the expansions and we're seeing customers expand 12x in a year. So for every dollar that we close, we got another 11, 12. So we are seeing the adoption in there and the results are coming. It's all about who is being the most savvy about it.
John Furrier
>> One of the things last year at the Asian Conference that Simon put on, I didn't make it, but I did review the coverage. I was out of town. This year, I'll be there. The conversations were very, I would say, inner circle, really tight community and now it's grown. So I have to ask you, what's the bets being made right now? Because there's a lot of discussion. It is moving fast, the open source side's growing. You're in the know. What bets are you making? What are people talking about? Because clearly the modes are changing. We see LLMs and the foundation models changing access, interface. You got Claude Cowork is damn good. OpenAI is amazing. I mean the AI is really phenomenal. So where's the value? What's the bets? What do you see the bets for the agents, the companies either doing them, suppliers they work with? How would you share your vision on the key bets?
Joe Moura
>> Yeah, there's a few trends in the industry that I think will have major impacts across different businesses. One is this idea of self-improving agents is something that is trending a lot right now. It's something that people are experimenting on R&D, like how I get my agents not only to work and not only to work reliably, but how I get them to self-evolve, to get better over time. I think the other idea is the idea of long-running agents as well, enabling very specific use cases where it's less about go out and do work for 10 minutes or half an hour, but more about go out and do work for days and what that will look like. And we're seeing some benchmarks actually coming around that. I think Anthropic just posted about how they use their new model to build a C compiler through weeks of work completely autonomous. So I think there's some of those things that are going. And then there's more longer term bets that people are making. I think it seems that I wouldn't call necessarily consensus, but definitely seems there's people that are now betting on conversational being the end and final experience for agents where you might have as many agents as you think. But in the end of the day it's just a conversational UI, your experience, and you don't care what is happening in the infrastructure behind it, similar to what OpenClaw or Moobot or whatever we're calling it these days. But the idea of you have this one single conversational UI that just feels like AGI. You're like just sending these things into the void and it gets worked on.
John Furrier
>> Yeah, I liked that whole viral thing that went on. I think it did go off the rails, but it started as basically, it's kind of a meme in my mind, but it does show the power. So if you look at it, you say it was a deep-seek moment in the GPU side. You say, wow, that's possible with very little, not very little, but some really ingenuity. So I have to ask you, what are you guys doing now? How do people use what you guys are doing? What agents do you see as easy things to knock down and what's the tools available? Take us through CrewAI.
Joe Moura
>> Yeah, sure. I mean, honestly for us it's interesting. What we're seeing a lot of success for us is that what we call like S&P 493. So basically if you remove the Mag 7 and you have all these remainder companies that basically have been promised to get all this value from AI. But a lot of their stocks might be sideways and they want to make sure that they're benefiting from this where they might have tougher margins and every 1% they get on the vision scans has a major impact for them. It's overseeing a lot of the opportunity to really drive value on these companies. So you go into some of these companies and some of the use cases are going to vary, but there's definitely a few clusters. I would say back office automation in general is a major one. Go-to-market automation is something that is a no-brainer. You can do it. This idea of enterprise GPT where you can chat with that and a mix of exploring data and taking actions. And then some of the obvious, marketing support and Code Jam. But I would say the combination of the six things like back office, go-to-market, enterprise GPT, marketing support and Code Jam are probably the low-hanging fruits right now in the organization that you get.
John Furrier
>> Is there any best practice on data wrangling or data hygiene or can I just throw it in the cloud and just say, "Agents, figure it out?"
Joe Moura
>> No, there's definitely ... Because it's not only data hygiene in terms of the agents being able to access this data, but there's a lot of data labeling as well. Sometimes you don't want to necessarily your agentic system to start exposing data that it was not supposed to. And right now the problem is not even the agent themselves, sometimes the underlying data is not a proper label, but that was obscure because there was no way to tap into that data. So we have seen some use cases where deploying an agent system actually has such an impact that now it's surfacing a lot of data and a lot of information that was being untapped before, but now you can leverage. And then the question goes back to like, wow, is that data ready to go? So in terms of what we're seeing is, again, it's not only about the tech, it's about the playbook. So when we partner with some of these companies, a lot of what we do is actually train. For example, we go into India, we go into South Korea, we just did a major training with Korea Telecom in South Korea, enabling their team. But the idea is going there and educating these people not only on how to build these agents, because it is different unless you're sitting on the cutting edge and the front lines of this, but also understand how to use the search, how to optimize it, how to think about use cases and ROI and all that. I think there's an educational component in here that actually is massive for these companies that goes not only to the data, like untangling, but also to understand how prioritize the use cases themselves.
John Furrier
>> Well, Joe, it's great to have you on. Looking forward to seeing at the Agent Conference in New York in May at the Hilton. Final minute we have, I'll give you a plug. What's going on with CrewAI? Number of employees, funding status, growth plans, your focus, what are you optimizing for? Put a plug in for the company and the group.
Joe Moura
>> Well, first of all, thank you so much for having me again. I think it's very exciting times to be running an industry in an industry like this. And we're seeing massive success. I'll be working with some of the biggest companies out there, heavily focused on enterprise. It is a lot of our team's background and it's kind of like what we are pursuing, but we have a new product features coming out. We are doubling down on conversational as the experience and meeting these customers where they are. And I think a big idea that we are building that these customers are getting super excited is retrofitting the systems around their company, understanding their company, creating a memory about their organization, their teams, their different business units and people. And that is unlocking massive use cases. So very excited about this. The company has 5x in people last year. It's going to at least like 2, 3x this year and having a lot of fun in the process. So very excited.
John Furrier
>> Yeah, it should be. It's a great time. You've got a great market. And again, the innovation's happening. People are doing more with less, the productivity's off the charts. You nail the agents, it's just they work around the clock, they don't sleep. So it's a good productivity. Joe, thank you so much. Good to see you. We'll see you when you get back in New York from the Bay Area next time you're in town.
Joe Moura
>> Thank you. I appreciate, John, and have an amazing day.
John Furrier
>> All right, take care. I'm John Furrier, host of theCUBE. It's a mixture of expertise, but a special coverage of the Agent Conference. Simon Chan and the elite community of developers really start as a meetup now growing into a full-blown conference. Because they're not hyping it up, all the players are there. They're grinding away. It's moving super fast. And again, the stakes are high and the productivity's there as ROI. The proof is in the pudding. You get the agents right, they will work. So thanks for watching. We're doing our part. We'll be there with the NYC Wired community. Thanks for watching.
>> Welcome back. I'm John Furrier, host of theCUBE. We are here at theCUBE's New York Stock Exchange studio. Of course, we have our Palo Alto studio connecting Silicon Valley to Wall Street. This is our mixture of experts series also featuring, as a preview, the upcoming AI Agent Conference, agentconference.com. It's a growing conference that started as a community meetup. Now it's a full-blown conference. All the experts, these are the players making it happen. We got a great guest here, Joe Moura who's the CEO and founder of CrewAI who built this company when it wasn't fashionable to have agents. Now everyone wants agents. They want to understand it. Joe, great to have you on this program with us and the NYC Wired community and the NYC Wired program at CUBE Original. agentconference.com, he's got a great URL, Simon does, so certainly that's a great draw. Agents are hot. Welcome to theCUBE.
Joe Moura
>> Thank you so much, John. I really appreciate it. A blast to be here and chatting with you today and agents are hot. There's so much that we can talk about that that's insane. I mean, we're seeing this being adopted from all sorts of customers, all different sides, all different industries. I think when people think about it, it truly can be as transformation as the internet was or even more, right? And I think the main differences is we're seeing a lot of the ROI happening way faster as well. So very exciting times.
John Furrier
>> One of the things we've seen with GenAI, we've seen the movie before with cloud on the previous generation. The AI native world now is kind of waking up to the reality that one, it's a total paradigm shift. The threshold has been crossed, the excitement and the strategy formulations not so much a risk factor as much as it's an execution risk. We're seeing that with generative AI. But now you've got the agents, which is a very frothy market because there's new strategy discussions around how agents are working, but still execution is involved because they're end-to-end workflows. They have autonomous capabilities, there's data involved. So you start to see the rise of the agentic infrastructure and the agentic applications. Claude bought Moltbook recently, OpenClaude now part of OpenAI. You're starting to see that agents are going to start talking to agents as part of work, but it's got to be powered by some infrastructure. So it almost feels like a DevOps moment for me because, in the cloud world, you saw that DevOps movement, the clouderati, the OGs, the ones that were pure, pure DevOps and then built a cloud. But that's now the foundation. Now you have an agent ops, agent infrastructure and the applications and it's very clear that the interfaces are changing. So where's the value? Where's the value capture? Where do I program? What do I invest in? These are all questions. What's your reaction?
Joe Moura
>> I got to say, you got in even the head there, right? And I mean, you put yourself in the shoes of the CIOs trying to make decisions right now, and it is insane. It's so busy because imagine you have every vendor on earth calling you and saying like, "Hey, you're using Salesforce. Why don't you use Agentforce? You're using ServiceNow. Why not use the ServiceNow agents?"
So you have the hyperscalers on one side, you have those incumbents on the other, you have the model providers now trying. I think everyone's realizing how much of a big shift this can be and everyone's trying to get the bigger piece of the pie that they can get. But there's one thing to be said where the market's moving so fast and these companies, they want to have progress in ROI, but they cannot pick winners. It's moving too fast for you to pick a winner. So what we're seeing out there is companies really favoring not only open source software but also favoring things that creates optionality. I think that's a big thing into all the C-level leaders right now is how I make progress by keeping the door open.
John Furrier
>> Yeah. And like I was mentioning with the clouds, the cloud migration, cloud growth, that was driven by open source. Now you have the Linux Foundation picking up MCP A2A, and now you got OpenClaw in there. You're starting to see the formation of another level. What's your view on this? What's your vision of how open source is going to change or accelerate the AI native movement?
Joe Moura
>> Yeah, I think you've got the word right. The thing there is operations, right? I think if you look at the last couple of years, there has been almost like a runway to the bottom in terms of making it extremely easy for people to build these agents. Like building, it's almost becoming a commodity. There are so many options that you can do in there. But what a lot of people don't realize, or people are starting to realize now, is that there is zero ROI. There is zero impact on the bottom line if you just build that. The value actually gets from running these things into production under a certain scale. And for you to get to that stage, then a bunch of other requirements starts to pop up from your regular suspects on enterprise, where these companies now need to start thinking about governance, about data privacy, around PII, or road-based level of access. And to your point, all that infrastructure that was built in the cloud and how that translates now into this enterprise environment, and there's even other things that in there that needs to be considered. But what we're seeing is there's definitely a new stack that is coming up, building on top of that data layer foundation, but now focus on that operational layer for all these agents that work around the MCP, work around the HWA, but allows you not only building but deploying, getting those things to value and actually observing not only their executions but the business outcome.
John Furrier
>> I always complimented Jensen Huang because he's the only CEO that used the word computer science in keynotes more than any other CEO. We talk about operating systems all the time on theCUBE. We love that tech. We're seeing an operating system kind of vibe with agents. It's not a chatbot anymore. You're starting to see the movement to multi-agent, agent-to-agent delegations involved. It's more than an API. So explain what you guys are doing, because I know you're doing some work in the area. How are you taking the open source movement, bringing that into a rapid production workflow or track when you have now this operating environment for multiple agents. And they could be agents from different applications and companies, so it feels like an API moment going on, but a whole nother level and what's the difference between the two?
Joe Moura
>> A thousand percent. I think we were talking about this as an agent management platform. So if you look at what we have been doing commercially, it's a lot about rooting that platform that helps these enterprises get the progress that they need, leverages much of their existing infrastructure and giving them all the optionality. So what that means is you look at many of our customers from AB InBev, one of the largest breweries in the world, actually the largest, to the PepsiCo, the Johnson & Johnsons and everyone that we have been working with. And you can see they all have similar needs where they want to deploy the systems, one, on-premises, so they're benefiting from their currently cloud infrastructure, so building on top of that. Integrating on their data providers, integrating with their Databricks, integrating with their Snowflake, and we basically create a platform that runs on top of that infrastructure that allows them to build, allows them to deploy, either with code or no code, but making sure that you get those reusable building blocks. So as you adopt this agent on a global scale, you're not rebuilding the same agents twice. You're not rebuilding the same two strikes, but instead accelerating towards that. That's the common theme that we keep hearing. And a big thing that you mentioned yourself is this idea of agents will talk over the wire. I think, you go back a year ago, a lot of people would not believe on that, but it's actually happening. We have some of this Fortune 500 companies using CrewAI as a platform to basically orchestrate ServiceNow agents or Salesforce agents. A2A is actually start to getting a lot of traction in there. So in the end of the day, I think there's four main things that these leaders care about right now. One is central visibility and control, two is reusability of pieces, so they allow technical and non-technical teams to adopt this. And then three is basically making sure they're leveraging all that flexibility, all the existing systems, existing models and everything that comes with that. And the fourth and final one would be this idea of talking with agents over the wire that seems to actually be accelerating extremely fast.
John Furrier
>> Yeah. And it's all those things under the covers, like identity you mentioned earlier. These are things that have got to be buttoned up, they're not for the faint of heart. You got to make sure all that security's in place. I want to get your thoughts. I just wrote a post yesterday on siliconangle.com around the Anthropic valuation, the title is, The $380 Billion Orchestration Bet: Understanding the Coding Wedge as AI Labs and Developers Move Beyond the Model Layer. And here's the comment, "The coding wedge has become a strategic beachhead where winning developers in the coding workflows represents the point of leverage to win AI enterprise adoption and broader orchestration control. Coding emerged as the first domain where autonomous agents were reliable enough for real production use, becoming the practical proving ground for capability and costs. Whoever controls how developers build with AI gains control of this orchestration layer, which then extends from developers to all knowledge workers and enterprise workflows." What's your take on that? Do you see it the same? How would you react to that? Do you see a similar dynamic where people are trying to figure out where's the leverage? What's your thoughts?
Joe Moura
>> I mean, I see the same way because in the end of the day, if you look at engineering, it's funny because no one thought that data would be the first target, but it turns out to be an amazing target. And what we're seeing is many people adopting. I mean, engineering use cases, like development life cycle use cases are all over the place. And the ROI is there. It's real. I mean, you look from companies on our size and how much we're leveraging AI to help us write and ship software nowadays, and you look at bigger companies as well, that is happening. The one thing, though, that I would add to there is as you lower the thresholds to build these things and you bring all ... You just make it easier for people to get these things out there. There's a few other things that becomes way top of mind and that is security, governance, all those things now need to come all the way to the left. You cannot have your entire team or your CEO now vibe coding apps and figuring out how to get these things out there without bringing all those concerns that, as you mentioned, is not for the fainted heart, all the way to the left. So that make sure that out of the gate you're taking care of all those things. But I agree with you. I feel like on engineering, especially for the folks that are on the cutting edge, they're the ones that experienced this first, this idea where the job is changing and you're going from doing a lot of the work yourself to actually delegating, kicking off all these different groups and clusters of agents to do the work and you actually managing them. But I think we're probably less than one or two years away from many other different horizontals within these companies actually start to feeling the same thing.
John Furrier
>> I had Sam Partee on from Arcade last week on an area with Harris and Chase, and we're talking about how fast this is moving. Can you share from your perspective, just in the past eight months, the velocity of change and the different approaches that people are taking? OpenAI saying, "We're the horizontal plane. Use us." Anthropic's saying, "We'll, vertically integrate." Isn't that either/or, and how should we be ... People should be thinking about how to leverage agents. If you've got your workflow and you've got your data mode, why wouldn't you want to go end-to-end? Or what do I want to put that into as a service if I'm AI native? Is it a different approach or horses for courses? What's your opinion and the speed?
Joe Moura
>> Yeah, no, that's the great thing. I actually spent some time talking with Sam Altman about this himself and we're talking about what is the last mile? Because the models are doing amazing work, and as these models get better and better, there's definitely a movement where people are removing the scaffolding. Where you would have, in the past, people putting all these sorts of scaffolding to make sure that these agents would behave in a certain way, as the models are getting better, people were finding themselves removing the scaffoldings and allocating more to the model. But when you actually look into these enterprises and you try to understand how you get them from that model being ready to actually impact on the bottom line to actual business outcome, the last mile is more like a hundred miles. What's happening in there? And we talking about a few things. We're talking about the security and the governance, but there's the integrations, the data prep, there's so much that is happening in there. So you got to have some appreciation for the fact that these companies want to move so fast and you want to make sure that we get them to accelerate and get these things out in production. But the reality on the day to day is you look at some of our customers and some of them have over 30 instances of SAP, and like, all right, how you get a model to understand and work through that. So I think that's going to take a while. But I see the acceleration. I mean, I look at our own numbers. Our average selling price has basically more than 11x. Like this year, you look at some of the expansions and we're seeing customers expand 12x in a year. So for every dollar that we close, we got another 11, 12. So we are seeing the adoption in there and the results are coming. It's all about who is being the most savvy about it.
John Furrier
>> One of the things last year at the Asian Conference that Simon put on, I didn't make it, but I did review the coverage. I was out of town. This year, I'll be there. The conversations were very, I would say, inner circle, really tight community and now it's grown. So I have to ask you, what's the bets being made right now? Because there's a lot of discussion. It is moving fast, the open source side's growing. You're in the know. What bets are you making? What are people talking about? Because clearly the modes are changing. We see LLMs and the foundation models changing access, interface. You got Claude Cowork is damn good. OpenAI is amazing. I mean the AI is really phenomenal. So where's the value? What's the bets? What do you see the bets for the agents, the companies either doing them, suppliers they work with? How would you share your vision on the key bets?
Joe Moura
>> Yeah, there's a few trends in the industry that I think will have major impacts across different businesses. One is this idea of self-improving agents is something that is trending a lot right now. It's something that people are experimenting on R&D, like how I get my agents not only to work and not only to work reliably, but how I get them to self-evolve, to get better over time. I think the other idea is the idea of long-running agents as well, enabling very specific use cases where it's less about go out and do work for 10 minutes or half an hour, but more about go out and do work for days and what that will look like. And we're seeing some benchmarks actually coming around that. I think Anthropic just posted about how they use their new model to build a C compiler through weeks of work completely autonomous. So I think there's some of those things that are going. And then there's more longer term bets that people are making. I think it seems that I wouldn't call necessarily consensus, but definitely seems there's people that are now betting on conversational being the end and final experience for agents where you might have as many agents as you think. But in the end of the day it's just a conversational UI, your experience, and you don't care what is happening in the infrastructure behind it, similar to what OpenClaw or Moobot or whatever we're calling it these days. But the idea of you have this one single conversational UI that just feels like AGI. You're like just sending these things into the void and it gets worked on.
John Furrier
>> Yeah, I liked that whole viral thing that went on. I think it did go off the rails, but it started as basically, it's kind of a meme in my mind, but it does show the power. So if you look at it, you say it was a deep-seek moment in the GPU side. You say, wow, that's possible with very little, not very little, but some really ingenuity. So I have to ask you, what are you guys doing now? How do people use what you guys are doing? What agents do you see as easy things to knock down and what's the tools available? Take us through CrewAI.
Joe Moura
>> Yeah, sure. I mean, honestly for us it's interesting. What we're seeing a lot of success for us is that what we call like S&P 493. So basically if you remove the Mag 7 and you have all these remainder companies that basically have been promised to get all this value from AI. But a lot of their stocks might be sideways and they want to make sure that they're benefiting from this where they might have tougher margins and every 1% they get on the vision scans has a major impact for them. It's overseeing a lot of the opportunity to really drive value on these companies. So you go into some of these companies and some of the use cases are going to vary, but there's definitely a few clusters. I would say back office automation in general is a major one. Go-to-market automation is something that is a no-brainer. You can do it. This idea of enterprise GPT where you can chat with that and a mix of exploring data and taking actions. And then some of the obvious, marketing support and Code Jam. But I would say the combination of the six things like back office, go-to-market, enterprise GPT, marketing support and Code Jam are probably the low-hanging fruits right now in the organization that you get.
John Furrier
>> Is there any best practice on data wrangling or data hygiene or can I just throw it in the cloud and just say, "Agents, figure it out?"
Joe Moura
>> No, there's definitely ... Because it's not only data hygiene in terms of the agents being able to access this data, but there's a lot of data labeling as well. Sometimes you don't want to necessarily your agentic system to start exposing data that it was not supposed to. And right now the problem is not even the agent themselves, sometimes the underlying data is not a proper label, but that was obscure because there was no way to tap into that data. So we have seen some use cases where deploying an agent system actually has such an impact that now it's surfacing a lot of data and a lot of information that was being untapped before, but now you can leverage. And then the question goes back to like, wow, is that data ready to go? So in terms of what we're seeing is, again, it's not only about the tech, it's about the playbook. So when we partner with some of these companies, a lot of what we do is actually train. For example, we go into India, we go into South Korea, we just did a major training with Korea Telecom in South Korea, enabling their team. But the idea is going there and educating these people not only on how to build these agents, because it is different unless you're sitting on the cutting edge and the front lines of this, but also understand how to use the search, how to optimize it, how to think about use cases and ROI and all that. I think there's an educational component in here that actually is massive for these companies that goes not only to the data, like untangling, but also to understand how prioritize the use cases themselves.
John Furrier
>> Well, Joe, it's great to have you on. Looking forward to seeing at the Agent Conference in New York in May at the Hilton. Final minute we have, I'll give you a plug. What's going on with CrewAI? Number of employees, funding status, growth plans, your focus, what are you optimizing for? Put a plug in for the company and the group.
Joe Moura
>> Well, first of all, thank you so much for having me again. I think it's very exciting times to be running an industry in an industry like this. And we're seeing massive success. I'll be working with some of the biggest companies out there, heavily focused on enterprise. It is a lot of our team's background and it's kind of like what we are pursuing, but we have a new product features coming out. We are doubling down on conversational as the experience and meeting these customers where they are. And I think a big idea that we are building that these customers are getting super excited is retrofitting the systems around their company, understanding their company, creating a memory about their organization, their teams, their different business units and people. And that is unlocking massive use cases. So very excited about this. The company has 5x in people last year. It's going to at least like 2, 3x this year and having a lot of fun in the process. So very excited.
John Furrier
>> Yeah, it should be. It's a great time. You've got a great market. And again, the innovation's happening. People are doing more with less, the productivity's off the charts. You nail the agents, it's just they work around the clock, they don't sleep. So it's a good productivity. Joe, thank you so much. Good to see you. We'll see you when you get back in New York from the Bay Area next time you're in town.
Joe Moura
>> Thank you. I appreciate, John, and have an amazing day.
John Furrier
>> All right, take care. I'm John Furrier, host of theCUBE. It's a mixture of expertise, but a special coverage of the Agent Conference. Simon Chan and the elite community of developers really start as a meetup now growing into a full-blown conference. Because they're not hyping it up, all the players are there. They're grinding away. It's moving super fast. And again, the stakes are high and the productivity's there as ROI. The proof is in the pudding. You get the agents right, they will work. So thanks for watching. We're doing our part. We'll be there with the NYC Wired community. Thanks for watching.
John Furrier