This discussion examines mixture of experts in enterprise artificial intelligence, no-code platforms and application modernization. Bob Picciano of IBM is senior vice president and master technical client advisor. Gary Hoberman of Unqork is founder and chief executive officer and a former global chief information officer. Picciano brings decades of experience in analytics, artificial intelligence, high-performance computing and cloud and they emphasize institutional responsibility and safety in AI adoption. Hoberman focuses on enterprise application decay and they explain Unqork's no-code, AI-first platform and reuse-driven architectures. The panel explores how deterministic governed runtimes intersect with generative tooling and practical approaches for portfolio rationalization.
Key takeaways include Unqork's model for turning probabilistic generative outputs into deterministic reusable components, which Hoberman describes as "making intent deterministic in implementation." Picciano highlights portfolio consolidation, security gains from a governed runtime and strategies to reduce technical debt while scaling application delivery. The conversation provides actionable guidance for enterprise IT leaders on adopting no-code platforms, governed runtimes and generative AI to accelerate modernization and improve security and developer productivity.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
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: Mixture of Experts Series
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: Mixture of Experts Series.
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: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
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: Mixture of Experts Series
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: Mixture of Experts Series. 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
Bob Picciano, IBM & Gary Hoberman, Unqork
This discussion examines mixture of experts in enterprise artificial intelligence, no-code platforms and application modernization. Bob Picciano of IBM is senior vice president and master technical client advisor. Gary Hoberman of Unqork is founder and chief executive officer and a former global chief information officer. Picciano brings decades of experience in analytics, artificial intelligence, high-performance computing and cloud and they emphasize institutional responsibility and safety in AI adoption. Hoberman focuses on enterprise application decay and they explain Unqork's no-code, AI-first platform and reuse-driven architectures. The panel explores how deterministic governed runtimes intersect with generative tooling and practical approaches for portfolio rationalization.
Key takeaways include Unqork's model for turning probabilistic generative outputs into deterministic reusable components, which Hoberman describes as "making intent deterministic in implementation." Picciano highlights portfolio consolidation, security gains from a governed runtime and strategies to reduce technical debt while scaling application delivery. The conversation provides actionable guidance for enterprise IT leaders on adopting no-code platforms, governed runtimes and generative AI to accelerate modernization and improve security and developer productivity.
play_circle_outlineAvoiding Technical Debt: Reuse, SDLC, Componentization vs LLM-Only Risks
replyShare Clip
play_circle_outlineUnqork AI: Combining Probabilistic Generative Outputs with Deterministic Enterprise Systems as Business Analyst and Enterprise Architect
replyShare Clip
play_circle_outlineGoverned runtime delivering application-as-a-service with high security and compliance
replyShare Clip
play_circle_outlineMassive application portfolio rationalization: shrinking hundreds of apps to few
replyShare Clip
play_circle_outlineOpen-weight models and enterprise model sovereignty: bring-your-own models into Unqork
>> Palo Alto Studio Connection, Silicon Valley and Wall Street.
Dave Vellante
>> I'm John Furrier, the host of theCUBE, here with Dave Vellante, my co-host. Welcome back to theCUBE's NYSE Wired Studio. We're here at the Buttonwood Podium. I'm Dave Vellante, and one of our next guests has spent more than 3 decades inside IBM. He was helping shape a lot of the technologies that helped define what we know as enterprise computing today. Bob Picciano. He led major businesses at IBM— software, hardware, analytics, AI, cloud. He ran global sales for a $20 billion software business. He led IBM's analytics through a major transformation. I remember watching that transformation from the front row seat. Watson Data Platform. He later ran IBM's cognitive systems portfolio, which included, as you know, Power Systems. the team delivered Summit and Sierra to Oak Ridge and Livermore respectively. Those are two of the world's at the time most powerful supercomputers. Today he's an independent board director at Rocket Software and Unqork, a company we're going to talk about today. And he's a chair of the advisory board at Senzing, Jeff Jonas's company. And of course, he's a private equity investor. Bob, great to have you back on theCUBE. It's great to be back here, my friend. Yeah, I was down at Oak Ridge this May. Oh, is that right? Just checking out what they're doing with quantum. So super excited there.
Bob Picciano
>> Yeah.
Dave Vellante
>> Gary Hoberman has held technology leadership roles at some of the world's biggest financial institutions like Citi, MetLife. He was global CIO, led a big team across many dozens of countries, probably 40 plus. Along the way, he saw the same issue over and over again that enterprises, they had a lot of ambition and a lot of talent. and they even had plenty of money, but they were sort of stuck with decaying applications and technical debt and legacy development models. So that led him to found Unqork back in 2017. And the premise that we're going to talk about today is that enterprise software, you don't want to build it the old way. Unqork was one of the pioneers of a codeless architecture and AI-first foundation for building complex, secure, and mission-critical applications much more quickly with better quality and lower cost than traditional code-based methods. Gary, welcome. Good to have you guys here. Thank you. So Bob, you have a wide observation space, as our friend Jeff Jonas likes to say.
Bob Picciano
>> Yes.
Dave Vellante
>> What do you see in the landscape out there? We've seen nothing like this before.
Bob Picciano
>> No, we haven't. First off, I want to congratulate you and John and thank you and the teams here that have continued to do a great job. I think your viewers watching you stay very well informed. You have a great— appreciate that, you know, also broad observation space. But you've done a great job curating the right people to come in and talk. Yeah, this is a very exciting time. Obviously the zeitgeist around AI is ever-changing, ever-evolving, but you're also covering it on a spectrum that's more than just the software and the model space. It's the responsibility of how this capability needs to get used in the enterprise by governments, by individuals, by consumers, and also the physical systems that need to continue to evolve, whether that's what's now in your phone or what's in next generation data center. And then adjacent to that and different from that is things like world models. You know, the AI space will evolve to be more deterministic. Not going to happen with frontier models. I don't think that's going to be breaking news. I think your viewershipknows that already. But I think the air in the balloon is going to start shifting around in a very interesting way, starting probably in 2027, 2028, maybe even more so. And also with things like quantum on the horizon. And we're now seeing the pioneers of quantum like IBM start to do some really interesting work. And we're seeing that technology accelerate. And, you know, very, very heartened to watch people like DarÃo Gil, who's now at the Department of Energy, really doing the right job to curate the right discussions and the right safety layers and the right applications of quantum technology in a responsible way for our administration, our current administration, our government. And I think we're leading on the world stage with that as well. So it is an exciting time. And, you know, I've been fortunate to touch a lot of those pieces contextually over the years from a leadership perspective and now seeing a lot of it mature. But I'll also say there's a lot of immaturity around this space as well. Right. You can't get away from the histrionics. I think people need and companies need to take a lot more personal and institutional responsibility for the things they're building. It's not sufficient to say, oh my God, there's a threat, there's doom and gloom. One of the core competencies and capabilities of these organizations that are overseeing these powerful technology evolutions has to be to apply it in a way that can do good, that does not do harm, and that can be managed in a safe and secure way. So I don't have much tolerance for people waving flags and asking for government intervention. So the government will step in and pour concrete around everybody's feet. We all know what they're up to, but it's time for enterprises to kind of hold them accountable for that as well.
Dave Vellante
>> It's interesting just to hear you talk. The enterprise used to be about, okay, I either have to steal market share from somebody to grow or disrupt. We saw that with SaaS or I've got to acquire companies to expand my portfolio.
Bob Picciano
>> Yeah.
Dave Vellante
>> But now when you talk about things like quantum, we had a guest on yesterday talking about fusion energy. Yeah, certainly AI itself and frontier models.
Bob Picciano
>> Yep.
Dave Vellante
>> Physical AI. The TAM is as an investor, it's just, it's, it's beyond calculation.
Bob Picciano
>> Absolutely.
Dave Vellante
>> How big the opportunities are.
Bob Picciano
>> And, you know, even on top of that, we have extraplanetary pursuits and building a moon base. So if there's not enough market here. Let's land on another planet and establish a new market.
Dave Vellante
>> So, Gary, let's get into your founding premise. What— why did you start Unqork?
Gary Hoberman
>> So I grew up coding. I still think in code in many ways. I always like to say I'm a hacker.
Dave Vellante
>> Do you dream in code?
Gary Hoberman
>> I do. I'm sure I do. And, you know, when you think of the ability to create something and the computer acts and executes it, it's magical. And that's been how I've always viewed technology. I started on Wall Street building trading systems back in 1994 and climbed that corporate ladder. And my dream was to be the C-suite CIO, to control and own and set the strategy. And as I climbed the ladder and I became an MD at Citi and then EVP, Global CIO at MetLife, two Fortune 50 companies, I suddenly realized that every day I was there, I was making the company worse. And it's a gut punch. Can you imagine?
Dave Vellante
>> I know what you mean by that.
Gary Hoberman
>> But could you imagine, your dream job, you've achieved it, and suddenly you feel like Every day I'm doing this and I kept picturing the CIO that comes after me opening up the envelope to see what I chose to do and going, what was I thinking? And the reason why I prepared 3 envelopes— joke, if you don't know, it's not the coding. It's not everything we hear about the coding. It was 80% of the budget was keeping the lights on everything before I joined. And I set the strategy to say, how do we move that 80% and build new systems? And it was impossible to replace the legacy. It was impossible to move the business forward. So we built brand new systems that added to the legacy. The second they went live, we'd get a technical debt report of everything we just poured into the company that will be someone else's problem in the future. And that vision was what set me to say, time to jump out of the C-suite, create Unqork. The vision I had when I created the company was how do you enable software such that 10,000 apps that are in an enterprise only look like one? How do you make it so everyone doesn't have their own software stack running on their own virtual machines on their own hardware? How do you enable it so it's a single application that's powering all applications across all industries just so you could get the scale that you saw in cloud computing? it sounds like cloud I'm describing.
Dave Vellante
>> Back in the day, and as you guys remember, a company called META Group, they got acquired by Gartner. Dale Kutnick was one of the co-CEOs. He had a very simple sort of rubric: run the business, grow the business, transform the business. What percent of your allocation is on each? And to your point, Bob, 80% was run and manage. Gary, it was run the business. Do you feel like organizations, your former peers, can actually move that needle? Does AI, bring that opportunity? Does Unqork bring that opportunity? Where can it actually go?
Gary Hoberman
>> Yes. So I'll talk about it from an Unqork perspective and my customers. So our customers are the largest enterprises in financial services, insurance, healthcare, and government, including anyone married in New York City. You've been using Unqork for the last 5 or 6 years. so when you think about highly regulated, secure compliance systems, the data we could share is we're running about 2 billion lines of code for our customers, supporting it with 40 engineers. So just from a scale point of view, you could look at the largest bank in the world, the biggest bank. They have 50,000+ engineers, an $18 billion IT spend, and they're running about a third of our code. So we're about 3 times bigger than the largest bank. And so we're empowering a world where the businesses and our customers could for the first time take not just their legacy systems that they didn't know what to do with— the mainframes, the Java, the .NET— they can actually take the new applications and set it on a path where it'll be 50% of the cost as opposed to draining your budget and keeping them up.
Dave Vellante
>> Are the big institutions leaning into that or is there a— no, I built my own code. I'm going to keep that.
Bob Picciano
>> There's always a—
Dave Vellante
>> keep, keep, keep that technical debt going. Is that inertia, right?
Bob Picciano
>> There's always a playful friction between business innovation and what the business tries to implement. And what technology wants implemented and how they want to implement it. And in some ways, the current understanding of how generative AI serves development is actually making that worse. Correct.
Dave Vellante
>> What do you mean by that?
Bob Picciano
>> Because it's creating more code, it's creating more legacy, and it's doing it at such a rate and pace that the ability to keep up with what it's doing is becoming very poor. And then, if people are reinforcing the wrong behaviors in that, not knowing what the code does, it has bad effects. And we've seen people complaining that their code has hacked other systems and has violated the trust of some benefit-oriented systems. So that is not what you want. What you want to be able to do is stay beholden to the software development lifecycle. One of the things that I've been very impressed about with Unqork is they're maniacally focused, in a positive way, on the software development lifecycle, meaning To what extent can we optimize the architecture and reuse assets that we already have to build what we want to in the next system? And most organizations don't understand their code legacy enough to be able to do that work. But Unqork AI does. And Unqork AI has taken a very interesting way of utilizing things like classes to maximize the reuse and this is the legacy of SDLC going back to people like, Barry Boehm and Tom DeMarco and Rational. Grady Booch had the vision of being able to create a rejuvenating enterprise by doing more reuse. And I think we're actually able now to do that with Unqork AI.
Dave Vellante
>> Let's get into the sort of heart of the matter here where when GenAI first came out, everybody thought, all right, this is going to run my software.
Gary Hoberman
>> Yes.
Dave Vellante
>> The we saw the SaaSpocalypse, which I'd love your thoughts on that. a lot of it is probably correct that there's major disruption going on. But what we've learned is that you got to bring the probabilistic, you got to bring the stochastic and the probabilistic and the deterministic together.
Bob Picciano
>> Yes.
Dave Vellante
>> Right. You can't just throw LLMs at the problem.
Bob Picciano
>> Businesses are deterministic. Right. And so these are deterministic.
Dave Vellante
>> But in so many ways, determinism is kind of a myth too, because you've got determinism in the sales department, determinism in the finance department, determinism in your ERP. And so there's that tribal knowledge that has to come together. Are you saying that you can sort of unpack that hairball of enterprise software?
Gary Hoberman
>> Dave, this week in New York City, big asset management conference on stage. In 19 minutes, I asked the audience, what do you want to build? And what Unqork AI answered back to them was, as soon as they said institutional client onboarding, which is a problem no one solved, our Unqork AI played the role of the business analyst. To Bob's point, it said, let me rewrite your business requirements into user journeys, stories. Let me rewrite it into modules and data and test plans. I'll create your test plans. And the second the business goes, that's what I want, it's a deterministic build to create the product from that point forward. So we're using a probabilistic method to solve the requirements process. But then the Enterprise Architect agent kicks in and goes, hey, I'm going to look first to see if I have anything to reuse. So before I build anything, let me see if there's something on the shelf that will save me time. And then when it says, I found a few things, I'm going to use those, but there are some gaps. I'm going to build those gaps as components for you. So the next time they're available for anyone in your company to reuse, anyone in your company, it's kind of like having our entire R&D shop working directly for our customer, forward deployed, if you will, as an agent, not as a person, which is And so with that said, let's talk about that. Yeah. Yeah.
Dave Vellante
>> Because you got deploycos now.
Bob Picciano
>> Yeah.
Dave Vellante
>> That are basically sort of mirroring what Palantir has done and they're doing some good work, but it's very narrow.
Bob Picciano
>> Yeah.
Dave Vellante
>> It's like, okay, I got a problem. Here you go. I have an AI engineer go fix it and then you're going to pay me. Okay. And it's those businesses probably aren't going to scale that well, although you see a lot of them getting, big value acquired.
Bob Picciano
>> But that's what, that's what Ajentic is all about, right? Being able to actually catalog that as a service, an MCP service, right? And then being able to call upon that service to actually do that forward deployment. That's the way it works in Unqork AI, correct? These are capabilities that are implemented virtually, but the philosophy of optimizing the architecture and maintaining the architecture is core to the system. So we talk about determinism in this context. It's making your intent deterministic in implementation. So the things that you verified in your requirements document, your design documents, the personas you agreed to serve, the data that you want to share and how that data should be used and the rules around that are implemented in a deterministic way. And we don't take liberties in thinking that we know how to rewrite it the best, the next best way when you want to change something else unrelated to the changes that get made by generative AI.
Gary Hoberman
>> And then taking it one step further, the second you say that's what I want and you test it and it works, when you say promote the build, DevOps automation. It's going into our governed runtime, which for 9 years has been powering the biggest institutions and governments. And the governed runtime concept is we replace the entire stack of enterprise software with our own code that we've been testing, securing, and we've been trusted by the largest banks and governments to run. That governed runtime means you can't break out of it once you're there. The governed runtime we could demonstrate is 100 times more secure than any code being written by any agent today. The last results we saw were 6,000 times more secure.
Dave Vellante
>> Why, Gary? Why were you able to achieve those results.
Gary Hoberman
>> And this goes back to my days at MetLife. So we had 8,000 enterprise apps and I would go present to the regulators and say, we're secure. We pen test these apps twice a year, these once a year, these every 2 years. When we get back the report in 3 weeks, we go back and we have 90 days to fix the critical issues, the highs, the mediums. If anyone from technology goes, I know, that's exactly what we do. We're testing one app. In the 3 weeks it took to get the report back, there were 1,800 vulnerabilities created in the world that they did not know about. They didn't test for. We're not secure. The answer was I was telling regulators we're doing all we can possibly do. Now, when they ran that pen test against the app, that app was just tested a day ago by someone else and the day before that by someone else. The largest bank ran a pen test. The biggest investment bank just pen tested it. The government benefited. New York City benefits. So in a world where, similar to cloud computing, where you abstract away the idea, you're able to operate on a fleet, you're able to give compute as a resource. We're giving application code as a resource and scaling it across 2 billion lines of code. And if we double that to 4 billion, I'm not hiring a single resource.
Dave Vellante
>> So we talk about AI factories a lot on theCUBE, which is largely a hardware instance of accelerated computing. It sounds like you— I can build a software factory essentially.
Bob Picciano
>> 100%. And the AI hardware is really all about training for the most part. Yeah. And so that takes a lot of resource. We'll continue. John's paper just the other night on how the data center is evolving and how compute is evolving is wonderful. And I've seen that for many years and IBM thinking about how to optimize what we're going to put on that die and how to build hybrid computing capabilities, high bandwidth capabilities. I compliment the IBM team on the ARM next generation. Yes.
Dave Vellante
>> Right.
Bob Picciano
>> And it's also similar to what Mark Papermaster and Lisa Su have been doing at AMD and that certainly Jensen has been doing for a long period of time with the evolution of GPUs.
Dave Vellante
>> If I think about this, the software stack, when we went from on-prem to SaaS, everything changed. The pricing model, the technology model, the whole operating model changed, moved to cloud. That's happening again in maybe a bigger way. But we still have the operational systems, the transaction systems. You've got the data platforms that are all trying to grab more territory. The BI, $50 billion BI business is obviously under attack from all these client services.
Bob Picciano
>> Yes.
Dave Vellante
>> You've got this kind of new system of intelligence, we call it this context ontology. And then you've got this cognitive layer on top and an agentic control framework. Where does Unqork fit in that stack?
Gary Hoberman
>> We provide out of the box in our governed runtime, the entire data stack, the entire file system stack. So you want to store WORM-compliant files for regulators. It's built in. You don't need a separate tool. You don't need a separate database. We've replaced the largest databases in the world. Using Unqork. So above that, we built an integration stack from the ground up. You don't need to bring in a third-party integration gateway, API gateway. Above that, you have built-in service catalogs to enable you to centrally support and secure your keys, your encryption, your certificates. So I ran enterprise technology and I saw the entire stack that was needed to build for the future company, a company that uses Unqork. The customers, they benefit because they don't need DBAs, SAs, DevOps. There's no DR testing or COB testing because we're hot, hot across 3 clouds. So we could actually be cloud agnostic today. There's no more security and pen testing where you're spending years trying to rectify. No more technical debt, no more end-of-life issues. We take that on for our customers as a service. We're doing that maintenance for you. But we're doing it across a fleet instead of that one app.
Dave Vellante
>> So what happens to the application portfolio? Is this abstraction layer that we have it lives in?
Gary Hoberman
>> Yeah, we have a Fortune 100 company about to hopefully do this press release saying they're going down from 600 apps down to 20 in their portfolio.
Dave Vellante
>> We are. Okay. So they're able to rationalize their portfolio.
Gary Hoberman
>> I recall, Dave, what we're seeing is the great consolidation, including the SaaSpocalypse, where you're seeing customers that say, how do I bring my portfolio of apps down? We would say shrink your surface area. Your code is your surface area. The more code you have, the larger you have to maintain and secure it. So in a world where you could shrink that, your costs go down.
Dave Vellante
>> Well, you know this when you went through Y2K, your portfolio exploded and then you as the CIO had to foot the bill for maintaining all these applications. You couldn't rip them out because the business would start screaming and then you were stuck You know, with this mess. And if you tried to rationalize it, it was always this sort of knife fight. Yeah. So explain how that dynamic has changed in your world.
Gary Hoberman
>> So there's two main entry points, I would say, for a customer today. One would be a technology group that wants to empower their business to build software that's secure, compliant, and trusted by IT, governed by IT. No shadow IT, no shadow AI. Using AI.
Dave Vellante
>> Okay.
Gary Hoberman
>> And that's a world that is untapped. That's a world where imagine the businesses being able to build software just by describing it and knowing that software is secure and architected well. The other side to this, the other extreme is what are those legacy systems, the $2.5 trillion of enterprise software each year spent that I described, I was mentioning that customers don't know what to do with, but they know are a risk. The second they run a Fable or Mythos model against their portfolio, they've identified ethical hack issues they have to go resolve for the regulator. They can't because these systems are 25 years old, 30 years old.
Bob Picciano
>> They don't have the time to maintain.
Gary Hoberman
>> So in that space, we come in and help them migrate that all to Unqork and shut down the entire legacy systems. And those two extremes are places we are welcomed in.
Dave Vellante
>> Interesting. All right. So, Bob, your intro narrative was laced with some great nuggets. Yeah, you were kind of alluding to the frontier model vendors sort of saying let's tap the brakes. I'm inferring from your comments, like Jensen's saying, they're in control. If they want to tap the brakes, they should tap the brakes, tap their own brakes. So play this forward. They've got, let's assume, trillion-plus dollar valuations. They've raised gobs of money. Everybody's sort of focused on them. They've, in my view anyway, they've got to grab a big chunk of this new AI stack. They don't have the deterministic software. They're doing partnerships with guys like Salesforce. I call it letting the fox in the henhouse.
Bob Picciano
>> Yes.
Dave Vellante
>> And they're going to learn from that. Yeah, because we've heard Alex Karp, every enterprise is sitting on their alpha. Yeah, that's right. Of course. Of course. You know, I'm not sure Palantir is the answer either. that's another layer that you have to worry about your sovereignty. But nonetheless, play this forward, the frontier model vendors have to grab more of that stack. They somehow have to do that. They've got to either partner or acquire or invent. They're struggling to go from research into product right now. We're seeing that.
Bob Picciano
>> Yeah.
Dave Vellante
>> How do you see this playing out going forward? Do you feel like the frontier model is a bad business model or they're going to get aggressive and really disrupt? the whole SaaS market.
Bob Picciano
>> I think it'll be all of that. I think that there are people in the SaaS market whose moats are either not as great as they think they are and they could very well be disrupted. And I think there's others that are in interesting layers where they have very strong value propositions. I think about the enterprise, I think about the enterprise customer.
Dave Vellante
>> Right.
Bob Picciano
>> And in some cases, that's a different type of customer. One of the companies that I've helped get funded and start to build is ARYA Labs. It's a world model AI company. That's a deterministic model that's bound by physics. And it applies itself to sustainable energy and aerospace and defense and life sciences and discovery. And they're very fixated on how to do this safely and in a deterministic, guarded way. And it's a very complete world model system, meaning it has rendering, simulation, and planning. So you can do very sophisticated, high-variable capabilities. So that's a different industry, different enterprise.
Dave Vellante
>> You can.
Bob Picciano
>> Yeah, and a very, very large TAM, very different than what Gary and I focus on. But when I think about the enterprise customer, they have to experiment with all these potential advantaging technologies, right? They need to be able to take advantage of what is being offered to them to the best of their ability. But they're constantly experimenting with multiple things on that horizon. And thank goodness that they are. And as things start to evolve to be more deterministically aligned to enterprise needs and as things evolve to have more safety and better economics. So I think about things like open weight models, and I think that's very exciting. And I see a lot of aggregation now with smart companies really understanding the benefits of that and the benefits for someone in financial services. are massive. GenAI's capability in Unqork utilizes generative AI to do some of its work, but you get to choose your model. You can bring your tokens for Opus into this model and use those tokens without any extra fees on top of those tokens to build what you need to build. But in the future, if you're a JPMC and you're thinking about open weight models, you can bring those models into this environment as well. And as those weightings develop, as your expertise informs those models, that stays sovereign in your enterprise. And I think that's a big way of how this plays
Dave Vellante
>> out.Yeah, open model is obviously a big part of Jensen's going forward. I'm not sure he's quite on the frontier yet, but knowing Jensen, he'll get there. Guys, we're going to go. Bob, Gary, thanks so
Bob Picciano
>> much.
Dave Vellante
>> you.Great to see you guys. Pleasure. And best of luck. And please come back and share with us your progress. All right. Keep it right there. This is Dave Vellante for theCUBE's NYSE Wired program. We'll be right back right after this short break.
>> Palo Alto Studio Connection, Silicon Valley and Wall Street.
Dave Vellante
>> I'm John Furrier, the host of theCUBE, here with Dave Vellante, my co-host. Welcome back to theCUBE's NYSE Wired Studio. We're here at the Buttonwood Podium. I'm Dave Vellante, and one of our next guests has spent more than 3 decades inside IBM. He was helping shape a lot of the technologies that helped define what we know as enterprise computing today. Bob Picciano. He led major businesses at IBM— software, hardware, analytics, AI, cloud. He ran global sales for a $20 billion software business. He led IBM's analytics through a major transformation. I remember watching that transformation from the front row seat. Watson Data Platform. He later ran IBM's cognitive systems portfolio, which included, as you know, Power Systems. the team delivered Summit and Sierra to Oak Ridge and Livermore respectively. Those are two of the world's at the time most powerful supercomputers. Today he's an independent board director at Rocket Software and Unqork, a company we're going to talk about today. And he's a chair of the advisory board at Senzing, Jeff Jonas's company. And of course, he's a private equity investor. Bob, great to have you back on theCUBE. It's great to be back here, my friend. Yeah, I was down at Oak Ridge this May. Oh, is that right? Just checking out what they're doing with quantum. So super excited there.
Bob Picciano
>> Yeah.
Dave Vellante
>> Gary Hoberman has held technology leadership roles at some of the world's biggest financial institutions like Citi, MetLife. He was global CIO, led a big team across many dozens of countries, probably 40 plus. Along the way, he saw the same issue over and over again that enterprises, they had a lot of ambition and a lot of talent. and they even had plenty of money, but they were sort of stuck with decaying applications and technical debt and legacy development models. So that led him to found Unqork back in 2017. And the premise that we're going to talk about today is that enterprise software, you don't want to build it the old way. Unqork was one of the pioneers of a codeless architecture and AI-first foundation for building complex, secure, and mission-critical applications much more quickly with better quality and lower cost than traditional code-based methods. Gary, welcome. Good to have you guys here. Thank you. So Bob, you have a wide observation space, as our friend Jeff Jonas likes to say.
Bob Picciano
>> Yes.
Dave Vellante
>> What do you see in the landscape out there? We've seen nothing like this before.
Bob Picciano
>> No, we haven't. First off, I want to congratulate you and John and thank you and the teams here that have continued to do a great job. I think your viewers watching you stay very well informed. You have a great— appreciate that, you know, also broad observation space. But you've done a great job curating the right people to come in and talk. Yeah, this is a very exciting time. Obviously the zeitgeist around AI is ever-changing, ever-evolving, but you're also covering it on a spectrum that's more than just the software and the model space. It's the responsibility of how this capability needs to get used in the enterprise by governments, by individuals, by consumers, and also the physical systems that need to continue to evolve, whether that's what's now in your phone or what's in next generation data center. And then adjacent to that and different from that is things like world models. You know, the AI space will evolve to be more deterministic. Not going to happen with frontier models. I don't think that's going to be breaking news. I think your viewershipknows that already. But I think the air in the balloon is going to start shifting around in a very interesting way, starting probably in 2027, 2028, maybe even more so. And also with things like quantum on the horizon. And we're now seeing the pioneers of quantum like IBM start to do some really interesting work. And we're seeing that technology accelerate. And, you know, very, very heartened to watch people like DarÃo Gil, who's now at the Department of Energy, really doing the right job to curate the right discussions and the right safety layers and the right applications of quantum technology in a responsible way for our administration, our current administration, our government. And I think we're leading on the world stage with that as well. So it is an exciting time. And, you know, I've been fortunate to touch a lot of those pieces contextually over the years from a leadership perspective and now seeing a lot of it mature. But I'll also say there's a lot of immaturity around this space as well. Right. You can't get away from the histrionics. I think people need and companies need to take a lot more personal and institutional responsibility for the things they're building. It's not sufficient to say, oh my God, there's a threat, there's doom and gloom. One of the core competencies and capabilities of these organizations that are overseeing these powerful technology evolutions has to be to apply it in a way that can do good, that does not do harm, and that can be managed in a safe and secure way. So I don't have much tolerance for people waving flags and asking for government intervention. So the government will step in and pour concrete around everybody's feet. We all know what they're up to, but it's time for enterprises to kind of hold them accountable for that as well.
Dave Vellante
>> It's interesting just to hear you talk. The enterprise used to be about, okay, I either have to steal market share from somebody to grow or disrupt. We saw that with SaaS or I've got to acquire companies to expand my portfolio.
Bob Picciano
>> Yeah.
Dave Vellante
>> But now when you talk about things like quantum, we had a guest on yesterday talking about fusion energy. Yeah, certainly AI itself and frontier models.
Bob Picciano
>> Yep.
Dave Vellante
>> Physical AI. The TAM is as an investor, it's just, it's, it's beyond calculation.
Bob Picciano
>> Absolutely.
Dave Vellante
>> How big the opportunities are.
Bob Picciano
>> And, you know, even on top of that, we have extraplanetary pursuits and building a moon base. So if there's not enough market here. Let's land on another planet and establish a new market.
Dave Vellante
>> So, Gary, let's get into your founding premise. What— why did you start Unqork?
Gary Hoberman
>> So I grew up coding. I still think in code in many ways. I always like to say I'm a hacker.
Dave Vellante
>> Do you dream in code?
Gary Hoberman
>> I do. I'm sure I do. And, you know, when you think of the ability to create something and the computer acts and executes it, it's magical. And that's been how I've always viewed technology. I started on Wall Street building trading systems back in 1994 and climbed that corporate ladder. And my dream was to be the C-suite CIO, to control and own and set the strategy. And as I climbed the ladder and I became an MD at Citi and then EVP, Global CIO at MetLife, two Fortune 50 companies, I suddenly realized that every day I was there, I was making the company worse. And it's a gut punch. Can you imagine?
Dave Vellante
>> I know what you mean by that.
Gary Hoberman
>> But could you imagine, your dream job, you've achieved it, and suddenly you feel like Every day I'm doing this and I kept picturing the CIO that comes after me opening up the envelope to see what I chose to do and going, what was I thinking? And the reason why I prepared 3 envelopes— joke, if you don't know, it's not the coding. It's not everything we hear about the coding. It was 80% of the budget was keeping the lights on everything before I joined. And I set the strategy to say, how do we move that 80% and build new systems? And it was impossible to replace the legacy. It was impossible to move the business forward. So we built brand new systems that added to the legacy. The second they went live, we'd get a technical debt report of everything we just poured into the company that will be someone else's problem in the future. And that vision was what set me to say, time to jump out of the C-suite, create Unqork. The vision I had when I created the company was how do you enable software such that 10,000 apps that are in an enterprise only look like one? How do you make it so everyone doesn't have their own software stack running on their own virtual machines on their own hardware? How do you enable it so it's a single application that's powering all applications across all industries just so you could get the scale that you saw in cloud computing? it sounds like cloud I'm describing.
Dave Vellante
>> Back in the day, and as you guys remember, a company called META Group, they got acquired by Gartner. Dale Kutnick was one of the co-CEOs. He had a very simple sort of rubric: run the business, grow the business, transform the business. What percent of your allocation is on each? And to your point, Bob, 80% was run and manage. Gary, it was run the business. Do you feel like organizations, your former peers, can actually move that needle? Does AI, bring that opportunity? Does Unqork bring that opportunity? Where can it actually go?
Gary Hoberman
>> Yes. So I'll talk about it from an Unqork perspective and my customers. So our customers are the largest enterprises in financial services, insurance, healthcare, and government, including anyone married in New York City. You've been using Unqork for the last 5 or 6 years. so when you think about highly regulated, secure compliance systems, the data we could share is we're running about 2 billion lines of code for our customers, supporting it with 40 engineers. So just from a scale point of view, you could look at the largest bank in the world, the biggest bank. They have 50,000+ engineers, an $18 billion IT spend, and they're running about a third of our code. So we're about 3 times bigger than the largest bank. And so we're empowering a world where the businesses and our customers could for the first time take not just their legacy systems that they didn't know what to do with— the mainframes, the Java, the .NET— they can actually take the new applications and set it on a path where it'll be 50% of the cost as opposed to draining your budget and keeping them up.
Dave Vellante
>> Are the big institutions leaning into that or is there a— no, I built my own code. I'm going to keep that.
Bob Picciano
>> There's always a—
Dave Vellante
>> keep, keep, keep that technical debt going. Is that inertia, right?
Bob Picciano
>> There's always a playful friction between business innovation and what the business tries to implement. And what technology wants implemented and how they want to implement it. And in some ways, the current understanding of how generative AI serves development is actually making that worse. Correct.
Dave Vellante
>> What do you mean by that?
Bob Picciano
>> Because it's creating more code, it's creating more legacy, and it's doing it at such a rate and pace that the ability to keep up with what it's doing is becoming very poor. And then, if people are reinforcing the wrong behaviors in that, not knowing what the code does, it has bad effects. And we've seen people complaining that their code has hacked other systems and has violated the trust of some benefit-oriented systems. So that is not what you want. What you want to be able to do is stay beholden to the software development lifecycle. One of the things that I've been very impressed about with Unqork is they're maniacally focused, in a positive way, on the software development lifecycle, meaning To what extent can we optimize the architecture and reuse assets that we already have to build what we want to in the next system? And most organizations don't understand their code legacy enough to be able to do that work. But Unqork AI does. And Unqork AI has taken a very interesting way of utilizing things like classes to maximize the reuse and this is the legacy of SDLC going back to people like, Barry Boehm and Tom DeMarco and Rational. Grady Booch had the vision of being able to create a rejuvenating enterprise by doing more reuse. And I think we're actually able now to do that with Unqork AI.
Dave Vellante
>> Let's get into the sort of heart of the matter here where when GenAI first came out, everybody thought, all right, this is going to run my software.
Gary Hoberman
>> Yes.
Dave Vellante
>> The we saw the SaaSpocalypse, which I'd love your thoughts on that. a lot of it is probably correct that there's major disruption going on. But what we've learned is that you got to bring the probabilistic, you got to bring the stochastic and the probabilistic and the deterministic together.
Bob Picciano
>> Yes.
Dave Vellante
>> Right. You can't just throw LLMs at the problem.
Bob Picciano
>> Businesses are deterministic. Right. And so these are deterministic.
Dave Vellante
>> But in so many ways, determinism is kind of a myth too, because you've got determinism in the sales department, determinism in the finance department, determinism in your ERP. And so there's that tribal knowledge that has to come together. Are you saying that you can sort of unpack that hairball of enterprise software?
Gary Hoberman
>> Dave, this week in New York City, big asset management conference on stage. In 19 minutes, I asked the audience, what do you want to build? And what Unqork AI answered back to them was, as soon as they said institutional client onboarding, which is a problem no one solved, our Unqork AI played the role of the business analyst. To Bob's point, it said, let me rewrite your business requirements into user journeys, stories. Let me rewrite it into modules and data and test plans. I'll create your test plans. And the second the business goes, that's what I want, it's a deterministic build to create the product from that point forward. So we're using a probabilistic method to solve the requirements process. But then the Enterprise Architect agent kicks in and goes, hey, I'm going to look first to see if I have anything to reuse. So before I build anything, let me see if there's something on the shelf that will save me time. And then when it says, I found a few things, I'm going to use those, but there are some gaps. I'm going to build those gaps as components for you. So the next time they're available for anyone in your company to reuse, anyone in your company, it's kind of like having our entire R&D shop working directly for our customer, forward deployed, if you will, as an agent, not as a person, which is And so with that said, let's talk about that. Yeah. Yeah.
Dave Vellante
>> Because you got deploycos now.
Bob Picciano
>> Yeah.
Dave Vellante
>> That are basically sort of mirroring what Palantir has done and they're doing some good work, but it's very narrow.
Bob Picciano
>> Yeah.
Dave Vellante
>> It's like, okay, I got a problem. Here you go. I have an AI engineer go fix it and then you're going to pay me. Okay. And it's those businesses probably aren't going to scale that well, although you see a lot of them getting, big value acquired.
Bob Picciano
>> But that's what, that's what Ajentic is all about, right? Being able to actually catalog that as a service, an MCP service, right? And then being able to call upon that service to actually do that forward deployment. That's the way it works in Unqork AI, correct? These are capabilities that are implemented virtually, but the philosophy of optimizing the architecture and maintaining the architecture is core to the system. So we talk about determinism in this context. It's making your intent deterministic in implementation. So the things that you verified in your requirements document, your design documents, the personas you agreed to serve, the data that you want to share and how that data should be used and the rules around that are implemented in a deterministic way. And we don't take liberties in thinking that we know how to rewrite it the best, the next best way when you want to change something else unrelated to the changes that get made by generative AI.
Gary Hoberman
>> And then taking it one step further, the second you say that's what I want and you test it and it works, when you say promote the build, DevOps automation. It's going into our governed runtime, which for 9 years has been powering the biggest institutions and governments. And the governed runtime concept is we replace the entire stack of enterprise software with our own code that we've been testing, securing, and we've been trusted by the largest banks and governments to run. That governed runtime means you can't break out of it once you're there. The governed runtime we could demonstrate is 100 times more secure than any code being written by any agent today. The last results we saw were 6,000 times more secure.
Dave Vellante
>> Why, Gary? Why were you able to achieve those results.
Gary Hoberman
>> And this goes back to my days at MetLife. So we had 8,000 enterprise apps and I would go present to the regulators and say, we're secure. We pen test these apps twice a year, these once a year, these every 2 years. When we get back the report in 3 weeks, we go back and we have 90 days to fix the critical issues, the highs, the mediums. If anyone from technology goes, I know, that's exactly what we do. We're testing one app. In the 3 weeks it took to get the report back, there were 1,800 vulnerabilities created in the world that they did not know about. They didn't test for. We're not secure. The answer was I was telling regulators we're doing all we can possibly do. Now, when they ran that pen test against the app, that app was just tested a day ago by someone else and the day before that by someone else. The largest bank ran a pen test. The biggest investment bank just pen tested it. The government benefited. New York City benefits. So in a world where, similar to cloud computing, where you abstract away the idea, you're able to operate on a fleet, you're able to give compute as a resource. We're giving application code as a resource and scaling it across 2 billion lines of code. And if we double that to 4 billion, I'm not hiring a single resource.
Dave Vellante
>> So we talk about AI factories a lot on theCUBE, which is largely a hardware instance of accelerated computing. It sounds like you— I can build a software factory essentially.
Bob Picciano
>> 100%. And the AI hardware is really all about training for the most part. Yeah. And so that takes a lot of resource. We'll continue. John's paper just the other night on how the data center is evolving and how compute is evolving is wonderful. And I've seen that for many years and IBM thinking about how to optimize what we're going to put on that die and how to build hybrid computing capabilities, high bandwidth capabilities. I compliment the IBM team on the ARM next generation. Yes.
Dave Vellante
>> Right.
Bob Picciano
>> And it's also similar to what Mark Papermaster and Lisa Su have been doing at AMD and that certainly Jensen has been doing for a long period of time with the evolution of GPUs.
Dave Vellante
>> If I think about this, the software stack, when we went from on-prem to SaaS, everything changed. The pricing model, the technology model, the whole operating model changed, moved to cloud. That's happening again in maybe a bigger way. But we still have the operational systems, the transaction systems. You've got the data platforms that are all trying to grab more territory. The BI, $50 billion BI business is obviously under attack from all these client services.
Bob Picciano
>> Yes.
Dave Vellante
>> You've got this kind of new system of intelligence, we call it this context ontology. And then you've got this cognitive layer on top and an agentic control framework. Where does Unqork fit in that stack?
Gary Hoberman
>> We provide out of the box in our governed runtime, the entire data stack, the entire file system stack. So you want to store WORM-compliant files for regulators. It's built in. You don't need a separate tool. You don't need a separate database. We've replaced the largest databases in the world. Using Unqork. So above that, we built an integration stack from the ground up. You don't need to bring in a third-party integration gateway, API gateway. Above that, you have built-in service catalogs to enable you to centrally support and secure your keys, your encryption, your certificates. So I ran enterprise technology and I saw the entire stack that was needed to build for the future company, a company that uses Unqork. The customers, they benefit because they don't need DBAs, SAs, DevOps. There's no DR testing or COB testing because we're hot, hot across 3 clouds. So we could actually be cloud agnostic today. There's no more security and pen testing where you're spending years trying to rectify. No more technical debt, no more end-of-life issues. We take that on for our customers as a service. We're doing that maintenance for you. But we're doing it across a fleet instead of that one app.
Dave Vellante
>> So what happens to the application portfolio? Is this abstraction layer that we have it lives in?
Gary Hoberman
>> Yeah, we have a Fortune 100 company about to hopefully do this press release saying they're going down from 600 apps down to 20 in their portfolio.
Dave Vellante
>> We are. Okay. So they're able to rationalize their portfolio.
Gary Hoberman
>> I recall, Dave, what we're seeing is the great consolidation, including the SaaSpocalypse, where you're seeing customers that say, how do I bring my portfolio of apps down? We would say shrink your surface area. Your code is your surface area. The more code you have, the larger you have to maintain and secure it. So in a world where you could shrink that, your costs go down.
Dave Vellante
>> Well, you know this when you went through Y2K, your portfolio exploded and then you as the CIO had to foot the bill for maintaining all these applications. You couldn't rip them out because the business would start screaming and then you were stuck You know, with this mess. And if you tried to rationalize it, it was always this sort of knife fight. Yeah. So explain how that dynamic has changed in your world.
Gary Hoberman
>> So there's two main entry points, I would say, for a customer today. One would be a technology group that wants to empower their business to build software that's secure, compliant, and trusted by IT, governed by IT. No shadow IT, no shadow AI. Using AI.
Dave Vellante
>> Okay.
Gary Hoberman
>> And that's a world that is untapped. That's a world where imagine the businesses being able to build software just by describing it and knowing that software is secure and architected well. The other side to this, the other extreme is what are those legacy systems, the $2.5 trillion of enterprise software each year spent that I described, I was mentioning that customers don't know what to do with, but they know are a risk. The second they run a Fable or Mythos model against their portfolio, they've identified ethical hack issues they have to go resolve for the regulator. They can't because these systems are 25 years old, 30 years old.
Bob Picciano
>> They don't have the time to maintain.
Gary Hoberman
>> So in that space, we come in and help them migrate that all to Unqork and shut down the entire legacy systems. And those two extremes are places we are welcomed in.
Dave Vellante
>> Interesting. All right. So, Bob, your intro narrative was laced with some great nuggets. Yeah, you were kind of alluding to the frontier model vendors sort of saying let's tap the brakes. I'm inferring from your comments, like Jensen's saying, they're in control. If they want to tap the brakes, they should tap the brakes, tap their own brakes. So play this forward. They've got, let's assume, trillion-plus dollar valuations. They've raised gobs of money. Everybody's sort of focused on them. They've, in my view anyway, they've got to grab a big chunk of this new AI stack. They don't have the deterministic software. They're doing partnerships with guys like Salesforce. I call it letting the fox in the henhouse.
Bob Picciano
>> Yes.
Dave Vellante
>> And they're going to learn from that. Yeah, because we've heard Alex Karp, every enterprise is sitting on their alpha. Yeah, that's right. Of course. Of course. You know, I'm not sure Palantir is the answer either. that's another layer that you have to worry about your sovereignty. But nonetheless, play this forward, the frontier model vendors have to grab more of that stack. They somehow have to do that. They've got to either partner or acquire or invent. They're struggling to go from research into product right now. We're seeing that.
Bob Picciano
>> Yeah.
Dave Vellante
>> How do you see this playing out going forward? Do you feel like the frontier model is a bad business model or they're going to get aggressive and really disrupt? the whole SaaS market.
Bob Picciano
>> I think it'll be all of that. I think that there are people in the SaaS market whose moats are either not as great as they think they are and they could very well be disrupted. And I think there's others that are in interesting layers where they have very strong value propositions. I think about the enterprise, I think about the enterprise customer.
Dave Vellante
>> Right.
Bob Picciano
>> And in some cases, that's a different type of customer. One of the companies that I've helped get funded and start to build is ARYA Labs. It's a world model AI company. That's a deterministic model that's bound by physics. And it applies itself to sustainable energy and aerospace and defense and life sciences and discovery. And they're very fixated on how to do this safely and in a deterministic, guarded way. And it's a very complete world model system, meaning it has rendering, simulation, and planning. So you can do very sophisticated, high-variable capabilities. So that's a different industry, different enterprise.
Dave Vellante
>> You can.
Bob Picciano
>> Yeah, and a very, very large TAM, very different than what Gary and I focus on. But when I think about the enterprise customer, they have to experiment with all these potential advantaging technologies, right? They need to be able to take advantage of what is being offered to them to the best of their ability. But they're constantly experimenting with multiple things on that horizon. And thank goodness that they are. And as things start to evolve to be more deterministically aligned to enterprise needs and as things evolve to have more safety and better economics. So I think about things like open weight models, and I think that's very exciting. And I see a lot of aggregation now with smart companies really understanding the benefits of that and the benefits for someone in financial services. are massive. GenAI's capability in Unqork utilizes generative AI to do some of its work, but you get to choose your model. You can bring your tokens for Opus into this model and use those tokens without any extra fees on top of those tokens to build what you need to build. But in the future, if you're a JPMC and you're thinking about open weight models, you can bring those models into this environment as well. And as those weightings develop, as your expertise informs those models, that stays sovereign in your enterprise. And I think that's a big way of how this plays
Dave Vellante
>> out.Yeah, open model is obviously a big part of Jensen's going forward. I'm not sure he's quite on the frontier yet, but knowing Jensen, he'll get there. Guys, we're going to go. Bob, Gary, thanks so
Bob Picciano
>> much.
Dave Vellante
>> you.Great to see you guys. Pleasure. And best of luck. And please come back and share with us your progress. All right. Keep it right there. This is Dave Vellante for theCUBE's NYSE Wired program. We'll be right back right after this short break.
Dave Vellante