Skip to main content

Robert F. Smith on AI, Innovation, and Enterprise Software

Vista Equity Partners Interview​

Data Sovereignty and the Personal Dimension​

That's all important and an important construct around utilizing AI in the enterprise. If we think about that concept—sovereignty and dominion over your own data—let's take that in the personal context.

Neither of us grew up with AI. Younger people, or investors, or executives, or founders are now growing up with AI as a tool that they can use.

Skills Needed in the Age of AI​

Question: What do you think are the skills that they need to have in the age of AI? Are they different from or similar to the skills that we had to have when growing up and thinking about how we wanted to build our careers?

Robert F. Smith: We had a lot of emphasis on STEM education—being able to solve, come up with algorithms to solve complex problems. That isn't going to be as useful because these systems will be more effective at it later.

So there is one argument that says: let me be effective at managing agents. That is a skill you and I didn't have to have. That is going to be, for certain people in their jobs, critically important.

For other people, it will be more important to understand how to communicate as a human being to another human being. To spend time—rather than the sales agent worrying about "how do I pull together this information"—thinking about how do I connect with that person as a human being so that I can actually make this sale, or whatever it might be.

So it depends on your job, your role, and what your plan and ambition is as to how you should utilize an agentic world or a generative AI world that you're now living in.

I've got a couple of my kids who are creatives. "No, dad, that's not—it's okay." Now I can see how they use it to advance part of the work that they are doing, which is still creative. But now things that would take them weeks to do, they can do in minutes, which gives them a chance to think about the next order of opportunity.

Defining Gen AI and Agentic Systems​

Question: I want to make sure we unpack some definitions here. Tell us a bit about Gen AI.

Robert F. Smith: We've been utilizing what I'll call artificial intelligence for over a decade in our companies. Machine learning, artificial intelligence—where you're actually able to create unique algorithms that create highly efficient tools to deliver very specific outcomes using those algorithms.

When you start thinking about Gen AI, it's a generative dynamic where you're generating not just the next letter, word, or phrase, but ultimately an outcome based on what the context is. This creates a whole different utility of opportunity.

And of course, you've got some great people out there—in Microsoft and Google and OpenAI and all those who are developing very unique models that can do this at size and scale, and with a fidelity that actually creates new products, new opportunities, and fascinates the imagination.

Agentic AI takes this further. It can actually start to not just take this generative information, but actually reason with it and do something. And actually act—in some cases, act based on an instruction from a person, or an agent orchestrator, or a human orchestrator orchestrating multiple agents to do series of specific tasks to do work.

The context of that actually creates a whole other set of dimensions of opportunity in the world of enterprise software. That is what we've been focused on with our partners, Microsoft and Anthropic—how do we take this agentic idea, make it real in the enterprise with the precision and capability that no one else can do.

From Users to Agent Resources​

If you think about it, I've got 750 million users of our software. Each user is a knowledge worker. You can actually take that knowledge worker's job and break it into a series of tasks that agents can effect on. Some will be induced or managed by the orchestrator—a person or another agent—which can do it. And now deliver that into our customer's environment to create value for the customer.

So I tell my team: we're going out of the user business into the agent resources business. 750 million users of our software basically equates to about 8 to 10 billion agents that we are going to actually be delivering into the marketplaces of our companies and our customers—agents that actually do work.

Parallels to the Internet Wave​

Question: When do you think we'll see the impact of all of this on companies? You saw the first wave of the internet. People got excited. It then obviously got over-exuberant. And then the impacts of the internet started to permeate and reverberate. Twenty years later, Amazon in 2015 was a much different company than Amazon in 2000. How do you think about what you saw in the first wave of the internet as it relates to AI?

Robert F. Smith: A similar paradigm. The first wave was of course with the hardware vendors, then infrastructure, and then software vendors really started to capture the opportunity set.

I actually struck a deal with this young man by the name of Andy Jassy back then, because he had been building a really interesting technology called AWS to support this other business called Amazon. He had some compute, and we were able to come up with some unique opportunities leveraging their capabilities and our infrastructure to now create a factory—to go from on-prem to cloud.

As far as I know, we've converted more business from on-prem to cloud than anyone else on the planet. That was a process that we saw still play out. I think we're still only like 50% or so of enterprise software companies that are now cloud. There's still on-prem, believe it. So it's 20 years later—16 years—that's the kind of dynamic we're dealing with.

I think something similar is happening now. Of course, the hardware vendors today—Nvidia, probably the most prominent—are capturing a vast majority of the economic rent and attention. The hyperscalers are saying, "Hey, we're going to build that infrastructure and capability because we've got a lot of free cash, we were infrastructure providers, and now we're going to provide a different type of infrastructure." Great, capturing economic rent.

Then you have Nvidia deciding, "Well, I'm going to get in that game too. I'm not just going to provide chips; I'm going to provide infrastructure as well." And you've got a bunch of sovereigns out there saying, "I'm going to do it as well."

So there's a whole infrastructure dynamic where you're now going to start to see the swelling of, over time, GPU capacity. But I still believe it's going to be the enterprise software vendors—those with sovereignty and dominion—that are capturing the vast majority of the economic rent, because they actually have the workflows and data sets that no one else has.

We're seeing it now because we're doing it now. But it is not trivial. This is not something you just plop an agent in and say, "Hey, go update these databases, go figure it out yourself." You've got to construct it in a way that actually creates the high-precision outcomes in the workflow environments.

So it may take a while to manifest broadly, but as far as I know, we are on the forefront of making these enterprise-class solutions effective today.

Democratizing Access to Enterprise Software​

Question: What made you decide to build a wealth business that would help the broader wealth community access enterprise software?

Robert F. Smith: Like all things, you sit back and look. Our institutional investors have been fantastic and great supporters of our business, and we will continue to support their businesses and their ambitions to expand in this space and, in many cases, to diversify what their opportunity set is.

We have found, however, that there are a lot of private investors who don't have access to this. 97% of these companies are private. We have the ability to actually democratize the opportunity set, and so we should do that. I think it's the right thing, frankly, for our country—to be able to deliver a set of investment opportunities that can actually change outcomes.

AI for Wealth Managers​

Question: How do you think about applying Vista's knowledge and network when it comes to AI in the context of helping the wealth partners you're working with run their own businesses?

Robert F. Smith: We get asked that question a lot. We are working with many of them to help them understand how we're using it and what the utility is. I think we'll see over time the emergence of very specific applications and use cases of agentic solutions that enable them to be more effective—not only in their services, but ultimately in the sales of their product to their customer base.

One thing that's interesting about wealth managers is they have sovereignty and dominion over their data, because their clients represent unique data to them. They can leverage that. That was something illuminated throughout this entire conversation, and you've done it as well at Vista with the business you've built.

From the Alt Goes Mainstream Podcast