Here is the transcript of the conversation between Fund/Build/Scale podcast host Walter Thompson and Ozzy Johnson, Director of Solutions Engineering at NVIDIA:
Ozzy Johnson: Some things we’ve talked about a bit earlier is how you don’t necessarily need to be a fully technical founder. You don’t need to have that background because AI is enabling so much in the way of, again, anyone who can solve problems. Anyone who can think in a structured way can be to some degree, a developer.
Walter Thompson: That’s Ozzy Johnson, Director of Solutions Engineering at NVIDIA. In his role, he’s a bridge between the company’s internal product teams and its global developer community. The team he leads also offers technical guidance to NVIDIA’s Inception program for startups.
We talked over Zoom about where early-stage AI founders need the most help and how people from academic and research backgrounds more easily shift to an entrepreneurial mindset. And we also spent some time exploring strategies for nurturing a successful AI developer community. He also shared some thoughts about balancing initial spending, with the need to drive early growth, and which trait successful AI founders have in common. (Spoiler: they aren’t all developers.)
Welcome back to Fund/Build/Scale. I’m Walter Thompson. I am talking today with Ozzy Johnson. He’s Director of Solutions Engineering at NVIDIA. Ozzy, thanks for being here.
Ozzy Johnson: Thank you. Glad to be here. Looking forward to it.
Walter Thompson: Based on what you’re seeing: developers who are trying to turn themselves into early-stage AI founders, where does that cohort need the most help initially?
Ozzy Johnson: The thing I think about most there is really a notion called fundamentals. I see when a lot of people are founding, they get really, really determined, really focused on a vision, or maybe a particular clever idea that they miss out on these fundamental habits that are required to get there — how to really execute how to create a valuable, saleable, differentiated product.
And a bunch of that is, in a way, is not exactly the fun part. It’s not the visionary work, it’s just sort of showing up. And in doing what can sometimes feel like mundane iteration day after day to get gradually better, that’s actually going to get you to that goal to to realize that vision.
Walter Thompson: Do you meet many founders who are not developers? I mean, it’s one thing to contribute to a project, but transforming that into a profitable business is something else entirely. So how hard is it to make that transition? And how do you see people making that shift in mindset?
Ozzy Johnson: Yeah, yeah. So I do, I do definitely see founders, who are not developers, you know, the very typical classic founding duo, CTO, CEO, CEO may not be a developer, sometimes they are. I think, though, in modern times, this is increasingly rare. You know, code generation has made development extremely accessible. And it’s really a matter of just having structured thinking and problem-solving more than knowing a specific language or framework. Really, the transition from taking a project to a product can be quite hard, but I don’t think it has to be.
I think the fundamental difference there is that when you’re defining a project, you’re essentially finding your own criteria for success. It’s like, what do you want to do? What is your vision, whereas a product success is absolutely and ultimately determined by the market. So those are a bunch of factors and things that you can’t control the chapter, observe the chapter, listen to and adapt.
Walter Thompson: So, from your perspective, you’re saying you don’t need a deep bench of AI machine learning talent on board to put together an idea and approach an investor?
Ozzy Johnson: No, I don’t, I don’t think you do. But you do need to build something, right? I tie back to what I was saying earlier about good ideas, that you’re not likely to be particularly popular if you’ve got the idea. and you’re looking for folks to realize that the world is full of good ideas. The multiplier that really makes something successful is the execution.
So at the very least, you know, with these enabling technologies, you know, go out there, prototype it, build it, demonstrate what it is, and then, you know, use that to recruit the folks that can refine it, they can scale it, they can extend it. I don’t think there’s, there’s much of any reason in 2024, for not being able to rapidly get to that prototype, you know, that’s not necessarily going to be your saleable, you know, scaled production product. But, you know, you should be able to build something.
Walter Thompson: Most of the founders I’ve come across so far since starting this project are coming out of academia and research, which means they don’t have a lot of fluency with bizdev, and sales and marketing, or building a brand. But VCs tend to say the best storyteller on the founding team is the de facto salesperson. What’s your perspective on this?
Ozzy Johnson: I actually really strongly agree with the spirit there, right? I think that having a story is absolutely essential, and being able to tell that story really concisely, in a way that resonates with, frankly, different audiences and people is essential. You need to be able to tell that story, both to folks who are potentially going to invest in you, you need to be able to tell it to your customers, and depending on who your customers are, you might need to be able to tell it to line of business leaders, an executive, an engineer, you know, a practitioner, an end user.
So you have to be able to just be really dynamic. For that reason, I think the person who can really weave and refine that story doesn’t necessarily have to be the one who goes around telling it. And ultimately, with all of this landing, whether that’s landing your funding, landing your customers, is something that involves a lot more than just creating or just telling the story. So kind of overall, that story is absolutely necessary, but it’s not sufficient.
Walter Thompson: Which leads into my next question: if you don’t have that skill set on the team, where do you start looking? I mean, you can always look on LinkedIn to find somebody who has experience with enterprise sales. But that seems like a really general approach.
Ozzy Johnson: Let’s sort of take that in two parts, right? Where to find people initially, yeah, it absolutely could be a developer community. But I would kind of warn against going and treating that like, you know, a direct recruitment effort or job postings, unless, of course, you know, it’s a community that kind of welcomes that sort of thing. There are places that do monthly or weekly threads about, you know, looking for help, right, trying to, for exactly this purpose. But if you’re not dealing with that, and say, you are like a non-or less technical founder, and you’re looking to pull people in, or the other way around, go be part of a community just have conversations.
I think it’s really a common mistake to think about. Recruiting and building a company is something that’s transactional. When you’re joining a company, when you’re creating a team, particularly a startup, you know, it’s not a transaction, it really is a collective relationship. And it’s a journey: you want to know the people that you’re going on with, you don’t want [someone] who just responded to an ad. The kind of second part of that is that there’s real discipline in, say, enterprise sales is very different from, you know, going and marketing a product that is looking to scale and be direct-to-consumer.
That’s really a matter of timing of knowing if that is the market you’re going after, and if you’re ready to do that, you really want to try to bring in the folks that have done it before. But even in that case, again, you know, you don’t want to make a transactional, you want to try to have that conversation, start that relationship, and grow from there.
Walter Thompson: Pretty much everyone I’ve talked to has emphasized how critical it is to start digging a moat early. but is there such a thing as doing that too soon? From the outside, it seems likely that a team will need to pivot at some point, or maybe not pivot, but continually iterate. So can you carve too deep a moat too soon?
Ozzy Johnson: I think it depends on the type of moat, right? Like some of it is just implicit, right? You can’t, you can’t leverage something that is built on you know, locked in or network effects until you’ve got the user base to really support that, otherwise, you’re just adding friction to growing it in the first place. So to kind of run with the notion, if you’re digging the moat before the castle was built, it’s just putting you on an island. That may not be that interesting to begin with, and this whole thought I really tie back to a bit of what we were talking about earlier with fundamentals and storytelling. If you know why you’re doing something, what sort of information went into those decisions, how you’re different, and really what you expect to happen, then you’ve really got everything that you need to know, not just to start, but to kind of, you know, feather the throttle there.
So, yes, start digging early, and speed up, slow down, or change the route based on that new information. And that sort of loop is what I am talking about when, when I’m saying fundamentals — just sort of day after day. What are we doing? Why are we doing it? How are we different? Do we need to speed up? Do we need to slow down? Do we need to change course? If you’re doing that, I think your moat is going to be fine.
Walter Thompson: Given that we’re in the midst of this hype cycle, do you think enough people are really spending enough time with the customer discovery process? Or are they just kind of rushing so they can bring something to market because technology moves so quickly? Is that OK, or is that a problem?
Ozzy Johnson: So I think it is fundamentally a problem, if it’s something that you’re that you’re not doing. I think when things are moving fast, you do kind of have to, you know, keep your head on a swivel, right? You need those loops to be to be short, or at least short or shorter. But I do think this is something that can be missed, to literally just talk to people and understand where they’re challenged.
One of the things that I think about a lot with this is very often the thing that is most valuable, that is most saleable to folks, the things they will buy? It’s not the things that they are not doing now that you are potentially going to introduce them to. It’s the things that they have to do that they are doing now, but they find kind of miserable, or are so important that in a way they can afford to do them well or not do them well. They can afford to, you know, do this with people in a way that’s really arduous. It’s really manual.
And if you’re finding a problem like that, you’re solving it, you’re making it better, you have you have gold there. If you’re not talking to folks, you don’t you don’t know what they’re really, really doing. And you could be solving a problem that feels interesting and feels clever to you. But isn’t actually something that anybody is looking for or isn’t necessarily going to benefit from.
Walter Thompson: Personnel costs in an AI startup seem like they’re usually the largest expense. But R&D is pretty expensive, as you know. Investors and founders, they both want to scale quickly, but can you share some advice and how to balance initial spending while driving early growth? My sense is that a lot of founders were really happy to spend on tech, but less so on software, like you know, people, that kind of thing.
Ozzy Johnson: This is a huge topic. So I’m going to try to take just a small piece of it or a few aspects, right, but what I would prioritize is what I think is kind of most important here. So kind of the first part of this, the I would say just practice looking through or looking at costs through several lenses, I think it’s really easy to fall into a mindset, that one category of costs is good or bad, or strictly preferable to another.
For example, it definitely resonates with what you say there about a lot of attention being put on personnel costs, but a true gem of a person is going to return a huge multiple. So, right, you want those folks you want to spend on them. The second part of this is really putting your resources, whether they’re direct or indirect costs, into what I call saleable differentiation, like, really, how are you different? What can you sell, as opposed to say, plumbing or things that could be real commodities there. And that can also be really about enablement, as well, to tie it back to the first point, it’s one thing to have a gem of a person, but you might need structure around them, you need to enable them, you need the infrastructure that’s actually going to get the most out of what they’re capable of.
And this too kind of goes back to the same big idea earlier, that it isn’t just about the idea. It’s also about the execution there. So yeah, I’d say, always just look at these things, not as buckets, but as an overall number. And what is that number turning into, in terms of value, you can sell, if you can’t tie those things into some expected value for your product for your growth in a relatively direct way you really shouldn’t be doing.
Walter Thompson: What are some of the top traits — it’s a hard question to answer I know, because there’s no checklists for who is a good AI founder. But if we were looking back at founders who you’ve come across who successfully launched their own open source projects, whether they are independent or VC-backed, what do they all have in common? Is there a common thread that connects these people?
Ozzy Johnson: I would certainly say there’s some, I don’t know if it would necessarily be specific to open-source projects, but just sort of successfully founding in general, I do think you have to have a certain amount of hubris and kind of unrealistic belief in yourself, right? Because you are doing something that is necessarily hard, you’re doing something that’s truly innovative, it’s not something that has been done before.
So that’s almost like a bit of a baseline there to have that. Confidence, though, again, that’s one of the things that is common, and I think necessary, but it’s not enough. Because if you are going directly at some goal, not necessarily executing, it’s not enough just to be motivated. So these folks also are going to be intensely reflective, right? Because this is getting into that iterative loop that you need to succeed. And it’s also the point where you might need to pivot in terms of your product, or that you might need a different team, because the folks who can build the thing, initially are not necessarily the ones who can sustain it or accelerate it. So I’d say it’s, it is a really important balance of confidence, and reflection, and agility there, right? Those, those are probably the three things.
Walter Thompson: Given where we are in this hype cycle, are you seeing people just kind of launch interesting projects to put themselves on the map? And is that a viable strategy?
Ozzy Johnson: Say it’s viable, but it’s not really sensible, right? If something’s viable, it can work, but generally the thing that or at least the way I try to think about this, I want to do the thing that is most likely to give myself the best chance of success. So, you know, really creating a project for the purpose of getting on the map is like, what’s the goal there? Wouldn’t it be far more efficient or just more sensible to make a real product that solves a real problem?
If you’re making a product for press or for notice — if that’s the goal that’s driving it, I just have to question whether that is really going to be representative and useful. Why not spend that time literally solving someone’s problem? And right there, you’ve potentially got your first hook for your first product.
Walter Thompson: What is your role with NVIDIA’s Inception program? Give us a brief overview.
Ozzy Johnson: My team sits here in developer programs. Inception is one of the major parts of the dev program, essentially, it is our strategy for startups. So the way my team works with Inception is really learning from what’s out there in the market by working with startups, if there is something that many startups are trying to do, we want to learn from that.
We want to understand, you know, what is difficult to adopt, what is really resonating with them, what’s going to help them build the next thing, because ideally, we want certainly all the most successful startups to be building on our platforms to be using our stack. And really, the best way to do that is ensure that they all are. And the way we can do that is to know what they need to have those things ready. And to support them in a way that really scales and again, provides value. So yeah, it’s a whole loop we learned and we try to then provide examples.
Walter Thompson: I’ve talked to a couple of people who are students at this point who are interested in starting up. And one of the things that they’ve come back to me and said a couple of times is that they’re looking for a place where they can see what other people are working on and share what they’re working on, but in a kind of a non-competitive environment. Is that any aspect of what you’re offering?
Ozzy Johnson: There’s certainly places for that, you know, also in our developer program is access to our community forums, there’s all sorts of discussion there. Through Inception, we do highlight the work of our members, we’re always welcoming folks who want to talk about their solutions, built with our tools, built with our products, [they] can publish those things through some of our blog platforms. And that is a really great way to do that, it gets you to a large audience. And then, of course, the folks that are consuming it are seeing how, you know, others have been successful.
We also organize a certain number of community events as well. Generally, we might be talking about a particular technology there that is new for us that we think may be helpful, interesting, etcetera. But it’s also time for folks to just mingle, share experiences, you know, talk with folks from my team, potentially others who are in attendance. So, yeah, this is absolutely one of the things that the program facilitates.
Walter Thompson: Does this also include, I mean, just not to be mercenary, but can you facilitate connections with investors or help someone connect with a potential customer?
Ozzy Johnson: Yeah, absolutely. So both connections to investors, sort of mutual matchmaking is absolutely part of the program, we call this Inception Capital Connect. Connecting to customers is as well, but the way we really think of this is just the general category of go-to-market help. So it would be doing this through visibility, right? This is being featured in a blog, or things of that sort. It can be much more direct. But that really depends on the product, on the needs.
And really the best thing anyone who is in the program can do for that is to really keep us informed. What are you doing? What does your product do? So that you’re on the radar, so that when we are talking with, say, you know, an enterprise company, or perhaps another startup, we’re able to say, “hey, actually, these folks over here do exactly what you need.” So let’s make that introduction.
Walter Thompson: Which leads to my next question, which is about actually landing those customers landing your first enterprise customers, it’s got to be really exciting. But it also comes with, you know, some liabilities and some opportunity costs. What are some of the downsides of working with a big company? And how can AI founders mitigate them?
Ozzy Johnson: The challenge with working with a big company is really, they’re kind of, I wouldn’t say special, but specific expectations. When you work within an enterprise or say, a consumer product. One, just the start, they tend to have a lot more stakeholders in the room, because you’re not just talking to, you know, maybe a practitioner that can work with total autonomy, you also need to convince the manager, you might need to convince a chief architect and executive, etc.
So that is where to tie this back to what we talked about storytelling, you need to be able to tell your story to everybody, you need to be able to present your product in a way that resonates with the different wants, needs and goals have all of those folks, and you likely need to do it in a way that that will work, when they’re all in the same room together. Without making one or the other feel any more or any less important, then once you’re working with them, you know, there’s a big challenge with an enterprise, they may have feature requests, there may be things that they expect that they need. And it can be really tough to pick exactly how far you’re willing to go, to land the deal, without letting the company essentially dictate your roadmap.
And this, I think, is a really high power skill of a good founder or a great product leader — to be able to understand that a certain amount of this information or what you’re getting from this company, may be gold, right? Because if they are in a particular space, whether it’s, you know, maybe they’re in finance, HLS or something else, very likely the things they are asking for are going to be representative of folks in that industry of that size. And it’s a great thing if you’re building to features or expectations that are then going to be saleable to the next two or three folks that fit that pattern.
At the same time, if it’s not saleable. If you can’t scale that, then you’re sort of putting yourself potentially in a world of hurt. So yeah, I’d say those are things to really be aware of. It really is different messaging, you need to have a clear and different story than you would otherwise. And yet there’s a lot to learn, but you have to be really protective of your own resources and your own roadmap when you’re doing that.
Walter Thompson: This is a transformational moment in tech that’s really lowering barriers for people who want to build something. From your perspective, what is this opening up with regard to expanding and diversifying the developer community?
Ozzy Johnson: Yeah, I think that’s a great question. And I think you’re exactly right. And that’s kind of the core point that it is lowering barriers. One of the things we’ve talked about a bit earlier, is how you don’t necessarily need to be a fully technical founder, you don’t need to have that background. Because AI is enabling so much in the way of, again, anyone who can solve problems, anyone who can think in a structured way can be to some degree, a developer. It really is democratizing access to information, generation of code, etcetera, etcetera.
What’s really exciting, what’s really great about this, to me, is basically, if you are a subject matter expert in something — whatever background you’re coming from, whatever that expertise is — you now have this opportunity to build something, potentially just starting with a team of one by yourself. And this isn’t a thing that we really had before, you needed to have a certain kind of pedigree, you needed to speak the language of code to do this. But now, it’s really anyone who has experience, an idea and a will to do that now has an avenue to do so.
And with the tools that are out there, you can do so really quickly, not at a huge cost. And what I think we’ll see come of that is just a huge increase in the diversity of people who are starting things who are building things, and frankly, of the solutions that are available. Some of the things that are most interesting to me to see come out of the current kind of revolution with generative applications are things that really help people who may have been previously underserved in ways that are kind of invisible, right? If you have tools that help you communicate better, help you phrase or rephrase or change your tone, or that you can self educate with, that is like truly kind of democratizing our future in a way. So, absolutely: I’m extremely optimistic and have really high hopes for what this is opening up for the future.
Walter Thompson: My final question, do you have any advice for someone who is interviewing for a job with an AI startup in 2024? If you were sitting across the desk during the interview process? What kind of questions would you ask the CEO to make sure this company was on the right track? And was something you wanted to take a bet on?
Ozzy Johnson: Yeah, that is a great question. Probably the biggest, biggest thing, asking back if it is a startup is, first I want to hear the story. Right, exactly. “What is it? Why is it? What are we doing here?” Very often, there are folks who can speak to this. And in a way that sounds good. But it’s like, “why is someone literally going to buy this product? How is it truly different? Is it better than what exists? And then as an extension of that? Where are we headed? What’s the goal?”
You know, it’s one thing if we’re saying “yes, this is how big we think we’re going to be, this is how we’re trying to exit, this is who we think is going to require us?” It’s another? If someone is saying, “well, yeah, this is my vision,” because vision is something that will necessarily change and can really only be understood by the person who has it.
So I’d be looking for real concrete direction and goals. And, “how are we different? Where are we headed? How is this different?” Essentially, just play me the movie of how this is all going to work and where I fit into it. “What is my role in making that successful?” That’d be the core.
Walter Thompson: Ozzy, thanks very much for a great conversation. I really appreciate the time.
Ozzy Johnson: Yeah, thank you. It was wide-ranging. Interesting. You really made me think in several places. I hope somebody benefits from it.
Walter Thompson: I’m sure they will. Thanks again. Take care.
Ozzy Johnson: All right, thank you.
Walter Thompson: I’ll be right back with some show notes after a word from our sponsors.
Thanks again to my guest, Ozzy Johnson, Director of Solutions Engineering and Nvidia. For my next episode, fundraising from both sides of the table, I interviewed Jorge Torres, CEO and co-founder of MindsDB and Vijay Reddy, AI Start investor at Mayfield.
If you’ve listened this far, I hope you got something out of the conversation. Subscribe to Fund/Build/Scale so you’ll automatically get future episodes, and consider leaving a review. For now, you can find the Fund/Build/Scale newsletter on Substack.
The show theme was written and performed by Michael Tritter and Carlos Chairez. Michael also edited the podcast and provided additional music. Thanks very much for listening.
This transcript was edited for space and clarity.