Episode 13: The Laws of Business and A.I.
Rob May and Brooke Torres interview Brendan Kohler, the Co-Founder and CTO at Sentenai and Co-Founder at Hyperplane Venture Capital, an AI-focused venture capital fund. Tune in for his perspective on hiring for A.I. talent, lessons learned investing in A.I. companies, and how to separate real from hype when applying A.I. at your business.
Rob May, CEO and Co-Founder,
Brooke Torres, Director of Marketing,
Brendan Kohler, Co-Founder and CTO, Sentanai, and Co-Founder, Hyperplane Venture Capital
Rob May: Hi, everyone. Welcome to the latest edition of AI at Work. I'm Rob May. I'm the co-founder and CEO at Talla. Talla builds a knowledge base automation and chat bot full-stack solution for sales and support teams that you can deploy to your reps or to your customers, so that you can answer questions faster or you can automate a lot of your tasks. It's a really, really cool product. Everybody we demo it to gives us a wow at a couple points in the demo. If you like the podcast, I hope you'll check it out and take a look at Talla. I'm here with my co-host for many of the episodes, Brooke Torres.
Brooke Torres: Hi, everyone.
RM: Today, our guest is Brendan Kohler. He's the co-founder and CTO at Sentenai and also a co-founder at Hyperplane Venture Capital, which is an AI-focused venture capital fund here in Boston, Massachusetts. Brendan, welcome. Why don't you tell us a little bit about your background and then what Sentenai does in the AI space.
Brendan Kohler: I got my start as a practitioner and researcher in AI. Over the past decade, I've done a little bit of everything. I jumped from there into entrepreneurship, investing and, lately, helping companies accelerate their efforts in realizing the potential of using AI through Sentenai.
RM: Tell us a little bit about Hyperplane. I know you were one of the original founders. Are you still pretty involved over there?
BK: Hyperplane is sort of an interesting story. When I came out of my last company, where we were doing predictive maintenance, driven by sensor data, we realized very quickly that the infrastructure that needs to support using AI in the enterprise and in industry-- it effectively is still very nascent. A lot of investment needs to go into the core technologies that will power the next business revolution, centered around AI.
I co-founded Hyperplane for the purpose of putting investment dollars where they hadn't been going, which is into hard tech companies that solve a lot of challenges around applying AI. Talla is one of them. There are many others in the portfolio. All of the companies share a common focus around solving really difficult challenges using data and for the purposes of deploying AI.
BT: We have a good range of guests on the podcast and I know you're pretty technical. Let's talk a little bit about how hiring for AI talent is different. What should business leaders be looking for? What's different about building a data science team and where do you see that team roll up in a company?
BK: I think a lot of listeners are probably constantly thinking about this. How do I improve? How do I deploy, or even hire for these kinds of positions? The thing that we've seen at Sentenai is that data science teams generally aren't built the same way that teams of software engineers are.
A data science team is generally full of people who have deep expertise in certain kinds of statistics and certain kinds of programming that don't apply across other parts of an organization. We see it as being best for an organization to roll up those teams within the business units they serve. Instead of having a core IT-centric data science or AI team, we see most successful companies deploy multiple smaller teams within different parts of their organizations. Or, if they need to centralize the full team to make use of data across an organization, they'll form a data science team under a digital team, and they will recruit people across the spectrum, from data engineering to visualization experts to experts in statistics and machine learning.
RM: You've invested in a lot of companies now. You're running an AI company. What are some of the most interesting lessons you've learned, advising these companies? If some of our listeners are in the process of buying AI products or building or deploying AI products, what are some of the things that you've seen that have worked and some of the mistakes that you've seen people make?
BK: I think the primary mistake I've seen in companies, that I've advised and invested in, make is for smaller companies to assume that they understand the complexities of the business they're trying to help. In our world, AI is this amazing malleable tool that we can use to solve many problems.
Recognizing, within a large company, what problems are actually solvable and what problems boil down to issues of companies structure or processes or the other variety of potential blockers that exist in an organization that's actively trying to execute on a business model. The business owners that engage these smaller companies really need to do a lot to bring along the smaller technology company into the organization and understand how they can help. If they don't do that, these technology-focused companies generally don't have a chance. What we see as being most effective is for smaller companies to engage with companies that are willing to invest the time and resources, learning and bringing these smaller AI-focused companies into the fold.
RM: One of the things I've noticed in the ecosystem in general is that you had this explosion of things people were trying with AI-- applications and use cases and everything else, and tons of startups doing that. I think one of the things that's happened over the last year in particular is, many of those companies have shifted and moved. Some of the moves have been big. Some of the moves have been very nuanced. Some of the companies are starting to look more like each other. I think they're doing this because we've found that intersection of the Venn diagram. Everybody's stumbling upon the same handful of use cases that are both technically feasible and economically viable for where we are. Are you seeing a similar thing?
BK: I think we're seeing it to a certain extent in various verticals. For example, in industrial IoT, many of the companies that were heavily focused on many different areas of the problem-- everything from sensors to the AI platforms-- they're all converging on a business model that involves more of a vertically-integrated solution. That's a reflection of the maturity and the technical capabilities of the businesses they're engaging in.
Similarly, our investments in financial-oriented technology companies that employ AI to help larger banks and investment organizations, they're also converging on certain business models. I think, you're absolutely right that there are certain areas where we're all finding AI can truly help, but a lot of this is because of the maturity of the markets and because businesses are unable to properly evaluate all the potential options with AI. It's much easier for them if the companies that are pitching to them frame everything the same way.
BT: What do you think about this idea that there is this flywheel effect where, if you don't adopt AI early, you miss the boat and you can't catch up?
BK: I wouldn't say that. I would say that there are certain inflection points within an organization that's focused on executing, as well as possible, where they have opportunities to adopt a different process, a different model, different technologies. Outside of those areas, where that inflection point happens. It's very difficult for them to be able to get approval up and down an organization.
Those inflection points happen every few years, every time different parts of a business stop working and corrective action has to be taken or every time IT technologies start failing to support the execution of a business as it grows. We see that a lot, especially on the software side. I think, it's important that companies at those inflection points evaluate all of their options, because they really only get those chances every few years within an organization, especially if they're a large organization. They just can't afford to move as fast as a startup or a small business can.
RM: One of the interesting things about companies that are deploying AI is that the last wave of technology, which was the social-mobile cloud wave, it was not difficult to understand if you were a non-technical person. It was relatively straightforward. Now, you have this wave of IoT, blockchain, and artificial intelligence. These are technologies that are hard to understand, sometimes, even if you are technical, depending on how technical you are.
When you think about being a mildly technical executive at a large company somewhere and you're looking at AI and you see all this hype and you read about the failed projects that Watson took on and you look at these things and you're bored and your CEO is pushing you to do AI, we've got to be an AI-first company, etc., and you are seeing successes in the news and you're seeing successes at Google, how do you start to make sense of it all? What do you look for? Do you have any mental frameworks or heuristics for starting to separate the fake stuff from the real stuff and figure out how to apply it to your business?
BK: It's a fascinating question. Even from an investor lens, it can often be difficult to separate companies from the ones that truly have the talent and the ability to execute on AI strategies and those that are marketing themselves as AI, because they think of it as a good way to get investment dollars. I think that also applies to selling into businesses and dealing with board objectives and adoption of technology within an organization.
What I would say is that the laws of business are not suspended for blockchain, AI, and machine learning. What a business actually has to do is evaluate the offerings of the companies based on business outcomes, cost, and the time-to-value. If an organization sticks to those metrics, then it doesn't matter which of the technologies are hype or not hype. They can evaluate these businesses through traditional frameworks.
RM: There are a lot of areas that AI hasn't been applied to yet. That happens for a couple of reasons. Sometimes, the data sets aren't there and we don't have the best small data algorithms yet, like we do for large data sets. Sometimes, nobody's taken a look. Are there any glaring opportunities in the AI space that you think, we really want to fund something there? Or something that you haven't seen that you just say, if I had time to go do another company, I would go do this? Is there a opportunity that nobody's tackling?
BK: It's difficult to find areas of business where AI has not been applied from a startup perspective. You see lots of attempts at creating innovation, oftentimes where there isn't need for innovation. What I would look at, for opportunities that aren't being addressed, is in places where there's already a huge amount of data within these large incumbent organizations and they're not actively leveraging it.
It's not so much that with Hyperplane, or even with Sentenai, we're looking at new areas where you could apply AI. Actually, we see the lowest hanging fruit as being within the enterprise and industrial spaces, where these companies have decades of data that they're not actively leveraging. A lot of our investment strategy is predicated on investing in companies that help large organizations sort out that data and create value from it. It's such low-hanging fruit, it's really hard to look at other opportunities that are, to be honest, significantly harder to generate value from, when you have all of this data and nobody actually is leveraging it. Even if they say they want to.
RM: I think, in general, web software and consumer software opportunities are overfunded and over-pursued, because that's what a lot of the tech media focuses on. For example, there aren't a lot of industrial design blogs or embedded software blogs. If you read stuff about it, you read Hacker News, it's all web software and that kind of stuff.
BK: The edge compute or, to use the buzz-word now, fog compute ecosystem is filled with companies that are effectively selling to CEOs. They're not solving value from the ground upward, where startups often see a lot of their best successes. If I look at where I was in this area, going from research into industrial IoT 10 years ago, and look at where everything is now, there's hardly been any advances in technology and technology adoption whatsoever. I think, at this point, there still needs to be a lot more investment in, for lack of a better umbrella term, industrial IoT and industrial AI. What we're seeing right now, given the size of the opportunity, is chronic underfunding.
RM: This is why, as a former hardware guy, my favorite tech periodical online has become this website called Next Platform. They write a lot about changes in hardware platforms. They do deep dives into a lot of the neuromorphic chips and AI chips. It's just super interesting stuff.
BK: Only a few companies are even leveraging or even know about these advances as they come out and how they might apply. From my perspective, it's really something that executives at large organizations should spend more time doing deep dives on. If they can yield 10% more efficiency or reduce risk in a large organization by 10%, 20% on critical business processes, that merits adoption of these technologies. You can't be afraid of it just being too academic or too complex to dive in and get an understanding of how it can help.
BT: We've got time for one more question that we like to ask. We get a lot of variety in answers here. Where do you fall on the Elon Musk versus Mark Zuckerberg spectrum for the future of AI?
BK: I wish I could believe we're at the point where it's already too late and AI's taking over. If that were the case, I feel like, as an industry, we'd have a much better understanding of how to apply machine learning and AI. I hate to say it, but I feel like the reality is that we're still doing fairly basic things with statistics.
While we have a lot of data that helps it mimic what we can do, we still don't understand the processes for thought and action. I think it'll be a long time, decades, before we actually get to the point where we can really worry. I think worrying is going to be a luxury that we might be able to have at that point.
RM: Brendan, thank you for joining us today. Those of you listening, if you have guests you'd like us to have on, if you have questions you would like us to ask, please send those to firstname.lastname@example.org. Thanks for listening. Check out talla.com. and we'll see you next week.