Adopting AI remains a challenge for most, and the fact that the world of AI is advancing so incredibly rapidly doesn’t help. Nathaniel Whittemore aims to make both adoption and keeping up with change a lot easier. He is the founder and CEO of Superintelligent, the AI enablement platform offering interactive tutorials that provide practical AI education and clear paths to business solutions. He is also the host of the podcast, AI Daily Brief, which seeks to keep its listeners up to date with AI as it evolves. In this episode of The Committed Innovator, McKinsey innovation leader and senior partner Erik Roth speaks with Whittemore about the intersection between Whittemore’s two companies, the challenges of adopting and scaling AI for enterprises, and what he sees in store for AI in 2026. This is an edited transcript of their conversation. You can follow the series on your preferred podcast platform.
Erik Roth, McKinsey: We are joined today by Nathaniel Whittemore, the founder and CEO of Superintelligent, and also the host of the AI Daily Brief podcast. Nathaniel, welcome.
Nathaniel Whittemore, Superintelligent: Thanks for having me.
Erik Roth: Do you see yourself as a founder, a communicator, or something in between? And how does that fit into Superintelligent as the company you’re building?
Nathaniel Whittemore: I’ve always viewed myself primarily as a builder–operator. It’s why I’ve never had the journalist tag. For me, the podcast is very much part and parcel of a larger way of interacting with the world.
Erik Roth: How would you describe Superintelligent, and what problem are you trying to solve in the world by creating it?
Nathaniel Whittemore: Superintelligent is at core an AI planning platform for enterprises, which are going through a big AI transformation that is dramatically and structurally different from previous technology changes. This shift will involve a total redesign of systems and a hybridization of workforces to include digital and human employees. There’s a lot of work to do to help people figure out how to do that well. Superintelligent plays in that space. It has evolved a number of times, but has always been concerned with the enablement gap between the value enterprises want to get out of AI and the value they are currently getting out of it. The way we’ve attacked that gap has changed as AI, and our adoption of it, changes.
Erik Roth: What’s been driving the evolution of your proposition? Is it just the recognition and assessment of that gap? Are there other factors?
Nathaniel Whittemore: It’s a combination. One part of it is our assessment of where the challenge in opportunity lies, but also how the technology changes the nature of that opportunity. There have been two or three iterations of Superintelligent. The first was an education platform. The thesis at the time was that the challenge was mainly employees not knowing how to use these new tools well enough. That was October 2023, and it has evolved to include a team sharing proposition, because even if people knew how to use these tools, what to use them for was a whole different question. In the fall of 2024, we had this platform where we had a number of large teams sharing how they were using and getting value out of AI, and it was paired with the educational side of the platform. But pretty soon I knew that what leaders were going to care about was what agentification can do. So we do a lot of agent planning now, and it incorporates AI more broadly. But agent planning wasn’t a thing people were doing up until the introduction of reasoning models because it just wasn’t viable. Even now, the capabilities of those tools are still evolving.
Erik Roth: Can you give us an example of how your platform is changing right now?
Nathaniel Whittemore: We’re in the midst of another small evolution. When we started, we introduced this agent readiness audit product. Basically, we deploy voice agents. The product interviews people across a company and then analyzes the results to provide a set of recommendations that are contextual to the firm. One of the unique things enabled by agents is, instead of having to take a small sample of people because that’s all you can realistically do from an interview perspective, you get to ask everyone. You get to talk to all of them about what they’re working on, and you get to process all of the responses in much faster succession. When we started, almost everyone we were interacting with was in the very early phases of figuring out what to do with such agents. So it was: Where are there opportunities to deploy agents? What should we prioritize? What’s our ranked-stacked list of where we should deploy?
That has evolved a little bit. We’re calling the new version “plateau breaker,” because we are now seeing lots of organizations that are 12 to 18 months deep or even 24 months deep in their AI strategies, and maybe six to 12 months deep in their agent strategy and are stuck on some particular plateau. I think this is even more common at this point than situations where people are still at the phase of saying, “Oh my God, we’ve got to start somewhere.”
Erik Roth: What do you think that plateau is? What’s going on for people stuck on them?
Nathaniel Whittemore: There is a set of very common underlying factors in the plateau. Data readiness is a big one. People find themselves running up against the limits of what they can do with the data they have. A second factor has to do with coordination, governance, and leadership. Their strategy is fragmented across different areas where there’s real disparity in terms of how far one group is versus another. A third is workflow or process mapping. Companies want to agentify and automate things, but they are realizing that the knowledge of how those things work currently is locked in people’s heads. So automating that process requires mapping it first. It’s actually good, though, that these factors are relatively consistent across organizations, because it makes figuring out how to help them much easier.
Erik Roth: How do you think about navigating this evolving technology space? Do you shift to a new product, or stick with what you’re doing? These are the questions everyone is asking.
Nathaniel Whittemore: I’ve done start-ups for my entire life, and you begin to have a sense over time about what the different phases of a start-up’s life need and which parts you are or are not natively suited to. We are still very much in a phase where we have a lot of what would be considered product-market fit around these agent readiness assessments and this AI planning suite. It turns out there’s a lot of AI planning to be done. Because the industry is moving so quickly, however, this is not a type of product-market fit where you can just instantly flip a switch and say, “OK, now it’s time to just scale.” Right now I am absolutely optimized for customers in the early stages of start-up development. I do not have some cost fallacy. I think my strength is being willing to totally upend the apple cart, even if we’ve been working incredibly hard on it, because the winds may shift and the landscape may change around us. It’s very easy to over-develop and over-change.
Erik Roth: But how do you know when you’ve done enough? And does that prevent you, whether explicitly or implicitly, from getting to scale as fast as you could?
Nathaniel Whittemore: The way I have dealt with this is to do a lot of listening to my folks who are actually on the front lines doing these audits. If it were up to me, I would have created an evolved product that is way past where our customers are. I’m thinking about the trajectory of AI and agents over the next few years, but many of our customers are still in the early phase. I create barriers where my mental restlessness is not going to have the opportunity to drag us to new places that aren’t where our customers are.
Erik Roth: How do you set those boundary conditions for yourself?
Nathaniel Whittemore: A lot of it is art rather than science. One piece that is foundational is having trust in the people interfacing the most with customers. My default proposition is that my team is going to have better insight into what our customers want and need right now. Another piece is that there are ways to field-test ideas without fully changing everything. We will often insert new product ideas into presentations without turning fully to some new thing. That gives us a chance before a product is developed to see whether people pick up on it and show interest.
Erik Roth: How do you balance customer input versus building out the platform? The enterprise business market is a huge opportunity that nobody’s quite cracked. How do you balance the evolving technology with customers giving you signals of what’s valuable or not?
Nathaniel Whittemore: I think one of the great tensions for foundation-model companies is, if this really is a race to AGI [artificial general intelligence] or ASI [artificial superintelligence], there is a very strong argument that it is a waste of time to deploy new engineering resources to use the AI we have now.
Erik Roth: That is the debate many companies are having.
Nathaniel Whittemore: It creates a great opportunity for third-party partners, because the foundation-model companies can’t be as responsive as enterprises need. There is still a question within that around how much you are willing to do to accommodate technologies that you know are transitional. I’ll give you an example: One of the tensions for us is that a very natural extension of our work would be to do workflow process mapping. Right now we ask a bunch of questions, we find out about current work processes, and we make a set of recommendations. Process mapping is a big piece of the plateau people get stuck on, so on the one hand it makes sense for us to do it. But I do not want to do it.
Erik Roth: Why?
Nathaniel Whittemore: One, there are tons of companies already doing it. Two, I think it is extraordinarily short-term. Obviously companies have to do it, but for us to build a business around it seems shortsighted. That’s because the current process mapping being done assumes that robots are going to imitate humans in how they do the work we use them to automate. I just don’t see it happening that way. As AI advances, the robots are going to do things their own way and find better ways to do things. But there is going to be this in-between period where there is billions of dollars’ worth of process mapping to be done. There is no right or wrong here; it’s just a question of what sort of business we want to be in and where we are well positioned to compete or not.
Erik Roth: That’s a good segue to your podcast, AI Daily Brief. It publishes six days a week and is a news analysis show that looks at everything going on in AI. How does it fit into the business model?
Nathaniel Whittemore: The podcast and Superintelligent are two totally independent companies with a very symbiotic relationship, in the sense that the podcast is the funnel for Superintelligent. But even more than just a funnel from a distribution standpoint, it is a credibility-building force. I view the show as a conduit and centralizing force. It’s me interpreting the news, curating it, and figuring out what I think is important. That is an immense amount of macro knowledge. There isn’t really a thing that happens in AI that I’m not reviewing, even if I’m not talking about it on the show. It gives me a very close, ground-level view informed by thousands of interviews I do all the time with executives and individual contributors—that gives me a more applied perspective. And so my sense of AI and, in particular, AI for enterprises and their workplaces, is informed by both the high-level macro views but also a ground-level view of what’s being reported now and what enterprises are actually doing now.
Erik Roth: How do you balance your time?
Nathaniel Whittemore: This is not my first podcast rodeo, so I had a pretty good process going into it. It would be easy to get lost in the sauce with a daily podcast. I already knew how to constrain it to a discrete process. It’s not short, but it’s discrete and predictable. And the rest of my time goes to Superintelligent.
Erik Roth: Has anything about doing the podcast surprised you?
Nathaniel Whittemore: The speed with which it has grown. I had a strong sense that the format would be resonant. I think it just speaks to how significant AI is going to be to everyone.
Erik Roth: Do people pitch you to get on the show?
Nathaniel Whittemore: They do. But I don’t think a single pitch has ever ended up on the show. That’s probably because if it’s interesting to me, I have probably already learned about it before they pitch me.
Erik Roth: As you learn from the podcast, how does that change the way you actually build Superintelligent?
Nathaniel Whittemore: If you think of the podcast as being at the top of the funnel for Superintelligent, do you move up or down that funnel? Do you move down the funnel toward more implementation? Do you move up the funnel toward more providing of information and live in that space? My assumption has been more down the funnel—not necessarily toward implementation but more software-focused agent interactive products. But I’m increasingly interested in the information space that exists pretty firmly between the podcast and Superintelligent. The simple way I’ve been thinking about it recently is the podcast is wide intelligence, and Superintelligent is applied intelligence.
Erik Roth: Every recent year has seen huge advances in AI. What’s next in 2026?
Nathaniel Whittemore: In no particular order, one is that we are going to see a lot of emphasis on context, engineering, and data. Organizations have known that they need to spend a little bit more time on data readiness. But it’s so hard to allocate time and resources toward these intermediate steps, as opposed to going after a flashy agent demo. Already I think 2025 showed that systems are the difference between the leaders and the laggards when it comes to enterprise AI. So I think there’s going to be more “narrative cloud cover” for the hard data work enterprises need to do. That said, there will be more of an emphasis on ROI in 2026. But I don’t think enterprises will stop their strategies because the ROI isn’t clear yet. I don’t think it’s like we’re going to have some super-coherent system where everyone agrees on the standards or the benchmarks. But I think you’re going to start to see efforts toward building out standards and benchmarks.
We’ve done a little bit too much assuming that everyone has to go figure out how to use these tools themselves rather than giving people ideas for how to use them.
Erik Roth: Do you think that is the mental shift leaders are going to start to have to make to take full advantage of these technologies as they come at us?
Nathaniel Whittemore: I don’t know. What’s not clear yet is where the lines between use-case creators and use-case takers are. Take vibe coding, for example. Should enterprises be creating sandboxes where everyone’s expected to do this? Or should they allow people who want to hack at these things to raise their hand? And then do they allow those things to scale if what they find is interesting? We’ve done a little bit too much assuming that everyone has to go figure out how to use these tools themselves rather than giving people ideas for how to use them.
Erik Roth: Most of us are accustomed to getting handed a product that’s ready to go. That’s not really where we are at the moment with AI.
Nathaniel Whittemore: This is the pattern with pretty much all new technology. Even with social networks, it’s a tiny fraction of people who uncover the cool thing to do with it, and then everyone follows suit. So it’s weird that we would think people would take these super-powerful new AI tools and just go off and figure it out for themselves.
Erik Roth: What is your favorite use case, for lack of a better term, for AI in your daily operations and leadership?
Nathaniel Whittemore: By far my most important use case is collaborative business strategy brainstorming. I use AI chatbots to sift through my ideas.
Erik Roth: As a thought partner?
Nathaniel Whittemore: Absolutely. I would highly encourage people to experiment with it. The chatbots have gotten a lot better.
Erik Roth: Nathaniel, thank you so much for your time today.
Nathaniel Whittemore: It was super-fun. Thanks for having me.
Related Articles
