Reed Sturtevant sees a lot of untapped potential in “tough tech.”
As a General Partner at The Engine, a venture capital firm launched by MIT, Reed and his colleagues invest in companies with breakthrough technology that, if successful, could positively transform the world.
It’s been about 15 years since I’ve last caught up to Reed—who was CTO at a startup we worked at together—so I’m so excited to welcome him on this episode of Experiencing Data! Reed and I talked about AI and how some of the portfolio companies in his fund are using data to produce better products, solutions, and inventions to tackle some of the world’s toughest challenges.
In total, we covered:
- How Reed's venture capital firm, The Engine, is investing in technology driven businesses focused on making positive social impacts. (0:28)
- The challenges that technical PhDs and postdocs face when transitioning from academia to entrepreneurship. (2:22)
- Focusing on a greater mission: The importance of self-examining whether an invention would be a good business. (5:16)
- How one technology business invested in by The Engine, The Routing Company, is leveraging AI and data to optimize public transportation and bridge service gaps. (9:05)
- Understanding and solving a problem: Using ‘design exercises’ to find successful market fits for existing technological solutions. (16:53)
- Solutions first, problems second: Why asking the right questions is key to mapping a technological solution back to a problem in the market. (19:31)
- Understanding and articulating a product’s value to potential buyers. (22:54)
- How the go-to-market strategies of software companies have changed over the last few decades. (26:16)
Quotes from Today’s Episode
There have been a couple of times while working at The Engine when I’ve taken it as a sign of maturity when a team self-examines whether their invention is actually the right way to build a business. - Reed (5:59)
For some of the data scientists I know, particularly with AI, executive teams can mandate AI without really understanding the problem they want to solve. It actually pushes the problem discovery onto the solution people — but they’re not always the ones trained to go find the problems. - Brian (19:42)
You can keep hitting people over the head with a product, or you can go figure out what people care about and determine how you can slide your solution into something they care about. ... You don’t know that until you go out and talk to them,listen, and and get in to their world. And I think that’s still something that’s not happening a lot with data teams. - Brian (24:45)
I think there really is a maturity among even the early stage teams now, where they can have a shelf full of techniques that they can just pick and choose from in terms of how to build a product, how to put it in front of people, and how to have the [user] experience be a gentle on-ramp. - Reed, on startups (27:29)
- The Engine: https://www.engine.xyz/
Brian: Welcome back, everybody. This is Brian T. O’Neill, you’re listening to Experiencing Data. I’ve got my pal, Reed Sturtevant on the line from—ah, it’s been, like, 15 years since we talked. You’re now in venture capital and supporting the tough tech startup economy at The Engine. Tell us about The Engine, Reed, and how has it been? It’s been a long time.
Reed: Sure, Brian, been a long time. The Engine is a ton of fun. We’re about four years old. We were put into business by MIT, and we are a mash-up of a venture capital firm, and a specialized co-working space with facilities like machine shops, biology labs, chemistry labs. We invest into companies that are working on technology-driven businesses that, if they succeed, could have a really strong positive impact on the world.
Brian: How do you balance that, the financial aspect and being a viable company and the returns that venture capital—you know, often want to see on their businesses, but also having some kind of social good aspect? How do you guys balance that? I can see those being at odds, sometimes.
Reed: Some social impact funds have an explicit expectation of less than traditional venture returns in exchange for the positive impact. We’re not like that. We are looking for companies that, in the fullness of time, could be amazing businesses. We’re fortunate that if you actually look at some of the companies that we’ve invested in, for instance, Commonwealth Fusion—which is working on commercial fusion power—it’s incredibly challenging, it will take a relatively long time to come to market, but at the end of the day, there’s no market risk. If you can have low-cost, fusion-powered electricity, the world wants that. So, in some ways, you’re balancing technology risk with lack of market risk. So, that’s kind of one way we look at it.
Brian: So, just for our audience, part of the reason I’ve had people that are in the investing space, come on to talk about how we’re using artificial intelligence—data more broadly—in the different portfolio companies. And I thought, after our pre-call, kind of talking about the people aspect, and kind of the difference between just understanding the technology piece versus understanding how to build a viable product for the market. And what happens sometimes when we have only a focus on the technical capability and not always on turning something into a viable product. And so—and I think this is important, even for our applied data science leaders and analytics leaders because they’re often trying to drive a change, they’re trying to get people to use the stuff that they’re making. They’re not selling it to their employees, but they kind of are selling it; they’re often trying to drive a change, and there’s a lack of adoption of a lot of these machine learning solutions in particular.
So, I’m kind of curious to hear about—tell me about the leaders that come in that maybe have very, very strong PhDs or very, very strong technical capability there, you had mentioned that some skills, sometimes, they need to unlearn [laugh] in order to become successful with getting people to care about the work that they’re doing, and the solutions, and the technology. Can you talk a little bit about those traits, and—
Reed: Sure. You’re right, it has been fun working with a lot of PhDs and postdocs who are becoming first-time founders, where they’re incredibly smart, incredibly dedicated, they put in a ton of hard work to accomplish defending your dissertation and becoming a PhD. So, they’ve been trained to be incredibly deep in their understanding of their particular research. At the same time, they, as entrepreneurs, sometimes don’t realize that you don’t have to go all the way down to the ground and understand every last little detail. You’re just trying to work towards an outcome.
And so the examples of that, where they’ve been trained to want to understand and be able to explain every last aspect, that’s not necessarily relevant to just sort of, like—look, we just want to figure out what customers want and how to get it to them. We don’t need to go all the way to the ground. And then there’s some really funny tactical things that they've learned along the way. And as PhDs, time is free and—
Reed: —money is scarce. And so you—it’s, like, the exact opposite on a startup [laugh] where you’re like, “Go faster, go faster. Raise money and spend it to go faster.” So, for instance, we had two founders who, the first week after we invested, they drove from Boston to New Hampshire to buy their laptops because they would save sales tax. And we’re like, “Why did you spend a whole afternoon buying laptops when you could have—[laugh] on click to buy?” [laugh] it’s kind of backwards. But it’s great to work with these people.
Brian: Do you ever see, especially because you guys, it’s marketed as ‘tough tech,’ and I’m curious if there’s ever a situation where the founders came in with, quote, “Tough tech that we’ve developed,” but the product doesn’t—in order to get the product into the hands of the people, the tough tech version of the product didn’t necessarily need to be made. There is an easier way, or as you said, and there was an outcome we could accomplish faster by not necessarily using all or some of the tough tech that was made. Do you ever find that’s the case? “We could water this down, do something faster. It doesn’t use my dissertation work, but it does solve the problem.” Is that ever something that occurs?
Reed: That’s actually a really interesting point. We have had a couple of times when I’ve taken it as a sign of maturity, that, for instance, to self-examine whether your invention is actually the right way to build a business. For instance, we invested in a team, a company called Cellino Bio, that’s working on advanced manufacturing for tissues, like human tissues to use for therapies, and when—the very first meeting I had with the founders, they were showing us their roadmap, and they had a particular milestone in it. I was like, “Wait. What’s that particular milestone, ‘we’re going to go out and look for technology?’ Haven’t you already inve—the thing you were describing that you invented in your Harvard lab, isn’t that what you’re going to use?” And they’re like, “You know, it might not be the best way to accomplish what we want, so we just needed to put a marker in the timeline that says, ‘let’s take a second look at what else is out there.’”
And that I was really—it’s a very mature way for these—this team at Harvard had invented a way to use lasers bouncing off of nanostructured metalized pyramids, to get stem cells to open up little holes in the cell membrane so that they could feed in new genetic material to create neurons from stem cells in three days instead of 30 days. So, they were extremely technical, but at the same time, they realized that, “Okay, what we’ve invented and gotten our PhD on, there may be a better way now that we’re trying to build a product.”
Brian: What do you think contributed to that ability to see it that way and to focus on the problem space, and to kind of not carry forward a tactic—that, “I will use this tactic at all costs. I will use this hammer for this job.” [laugh]. Is there something about a personality? Is it a learned behavior? What do you think that is?
Reed: In this case, the team was absolutely focused on the outcome. They had a strong belief that, to be able to cure certain types of diseases—like, they’re currently on a path where their first product will be used to solve age-related macular degeneration. So, blindness that comes from deterioration of cells at the back of the eye, and so their mission was to help people with problems and the technology was a means to an end for them, but they really were focused on how can we get support from investors, and people we need to hire, and companies we need to partner with to accomplish that mission?
Brian: Mm-hm. But is it—there’s a drive to start a business or to have a business maybe, not just use my technology that I’ve created or that I’ve discovered through my work? Was that kind of the theme there?
Reed: Yeah, it is. And it’s really the—to achieve that outcome at scale, to help many, many, many, many people, then having a successful business, kind of the capitalistic drive to succeed as a business and grow as a business is really, for them, it was a mechanism to have outcome, the outcome at a large scale.
Brian: Got it. Got it. Tell me about some of the other companies. I know you guys—the work, as you mentioned, the fusion and stuff, it goes outside the realm of software, but tell me about some of the companies that are leveraging machine learning, and AI particularly. That’s kind of obviously the focus of this show.
Is that a heavy part of what you’re seeing in the software space, or in those software portfolio companies that you guys have, or is that not necessarily something that’s you’re always looking for, or you care if it’s there? It’s more about whether there’s viable business. Can you talk a little bit about that?
Reed: Yeah, it’s funny, different forms of AI, some machine learning, but also other forms are almost, kind of… tools of the trade across a lot of these companies. Even if they’re not part of their externally facing service or product offering, they are foundational behind the scenes. And so one example, a company we invested in recently, called The Routing Company, it’s a transportation product. They—research was done at MIT, and MIT’s Computer Science and AI Lab, but Alex Wallar, one of the founders—working with Professor Daniela Rus, who runs that lab—and he had worked on research around optimization algorithms, where you wanted to be able to understand how much time it would take to run an optimization and understand what quality you could get, what improvement you could get, given a certain amount of time. So, some of these optimization problems are incredibly time-consuming to run, so you don’t necessarily need the perfect, ultimate answer; you’re just trying to get an improvement.
And so his research was really that time-quality trade-off, they did one paper which used historical data from a day’s worth of New York City taxi rides, and said, “Hey, if we could optimize fitting people who wanted to go from this place to that place into shared rides, could we have way fewer cars on the road without sacrificing how much time you have to wait to get picked up, or how much time you have to—your ride was longer because other people are dropped off.” It’s kind of like, anybody who’s taken an airport shuttle knows the feeling that, “Why is my hotel the last one on the route?” So, they were able to show that, without affecting quality for the riders—wait times and ride times—that they could remove three-quarters of the vehicles off the road. So, if you think of that level of solution, in terms of reduced congestion, reduced air pollution, by sharing rides, you can make it cheaper for everybody who’s trying to get somewhere. So, the social equity of longer and longer commutes where people are trying to use public transit, that’s the problem they set off to solve.
So, they use AI, and machine learning in some cases for… hey, let’s understand as we operate in a certain location, where we should have cars waiting, at what parts of the day. So, they do use machine learning to sort of improve the service in a certain geography, but the core algorithm comes from a different part of the AI world, in terms of the optimization as their target.
Brian: And the goal with that is to apply this to public transit, not to taxis and all of that? Because I would see, well, what about the driver incentive? “I want to pick up rides. There’s fewer rides, but I’m still driving out there trying to find them.” But, like, whereas with the city, it’s like, “We’re just going to schedule less buses.” Or, “We’ll know exactly where to schedule them, and we’ll just redistribute our employees to other work,” or something like that. There’s no individual bus driver that can just go [laugh]—
Reed: [crosstalk 00:12:45] you’re right.
Brian: —so there’s an incentive there. Is that what it was intended to, is to apply the routing optimization to the public space?
Reed: That is their go-to-market. They recruited a CEO, James Cox, who ran the product side of Uber Pool, and they are taking a go-to-market where they’re running a pilot in Houston right now, as an example. They’re going to run one in Jacksonville, Florida, where they’re serving the public transit authorities. So, they said, “Hey, we can help you with your last-mile problem.” They have software. It looks like Uber Pool, where you have—they have software that runs on the phone for the rider to call for a ride, for the driver to agree to pick one up, and then a dashboard for the transit authority to understand, kind of, overall what’s happening.
Brian: Got it. So, on the social good thing I think we touched briefly on this on our pre-call, but talk to me about the—you know, this is a great example. It’s like when we think about the communities served by public transit, for example, are we accelerating the inequities that may already be at play? Does the algorithm need to account for this somehow if we’re trying to actually improve the distribution of public transit availability, et cetera? Do they account for this with intention? Or how do they look at that, right? B
Because if we train the model based on where all the routes are now, does that mean it’s possible when we remove the transit—because it sounds like we’re going to be removing stuff—that we could end up removing stuff where there’s already a short supply, and if we’re cutting it in half, are we really potentially hurting a community that’s already underserved without extra-compensating for it in the solution?
Reed: That's an interesting angle. In this company’s case, they’re really talking about bolting on what the industry calls direct response transit, where you call for a ride rather than a fixed route, you know, a light rail or a bus route with scheduled stops. So, they're looking at—there’s a last-mile problem where not everybody can afford to live right next to a subway stop or bus route. And so as you try to serve a more distributed population, they were trying to fill gaps, or essentially not kind of take out existing forms of—the fixed-route transit, if you look at their models, work really well for dedicated commutes, kind of rush hour, but off-cycle and more rural areas are not well served by those forms of public transit.
Brian: Can you tell me about the Scotland aspect?
Reed: Yeah, thanks for the reminder. They are all working—
Brian: With an accent, please. [laugh].
Reed: [laugh]. [unintelligible 00:15:19]. The Scottish Highlands have asked them whether—they already have a ride service. You schedule a pickup, and it’s for senior citizens. And it’s a pretty thinly populated region with not a lot of density of roads.
But the government agencies didn’t really have any data about how ridership and routes were being served, so they came to The Routing Company and gave them a grant to build a dumbed down—just essentially a data collection version of their app. So, a phone or tablet on the dash, the driver pushes a button when they set out to do a pickup, pushes a button after they pick someone off, and pushes a button when they drop them off. And then it’s GPS, geolocation enabled. And it’s just logging all of the routes and pickups, and providing a data set for the government agencies to do planning. “Should we add more cars? Should we change the size of the car? Should we do shared rides rather than point-to-point?”
And so it really was an example where apart from providing the rides, just providing the data had super high value. So, it was quite a great on-ramp to the market for eventually The Routing Company would like to help that agency automate the scheduling and pickups, but as a starting point, the agency said, “We don’t even know what we’re doing right now. We don’t have the data, we don’t have the logging.”
Brian: Right. Right. Interesting. Tell me about the—I’m just curious if any of your companies, and maybe this is less relevant with a fusion reactor—but is design a part of the beginning of some of these companies? Do they either through a trained practice or people that they have on their teams, is that relevant to some of the work that is happening in tough tech?
Reed: It is. It’s interesting, it shows up in different ways. For instance, a lot of the technologies that we encounter and get involved with are platform technologies, for better or worse, where they have multiple use cases. So, one example is Via Separations. The founders Shreya Dave and Brent invented a new kind of membrane. So, it’s material made out of graphene-oxide that’s linked together. So, it’s a filtration membrane. With a platform technology, there’s lots of—they said, “Oh, maybe dairy plants would use it for separating protein. Oh, maybe pharmaceutical companies would use it after a chemical synthesis to separate out the material they’ve created.” So, they really are walking the desert looking for a kind of the best and highest use, and to do that, they have to put themselves in the shoes of the industry that they’re trying to serve. And so that’s a quote-unquote, “Design process” to understand the experience and to place their solution in that lived experience. So, the first pilot that they’ve signed is actually in the, of all things a pulp and paper industry.
And a large paper plant might be a billion dollars to build, so incredibly expensive to expand. And when they’re making paper, if you’ve ever driven by one, maybe up in Maine, it really stinks. And there’s a thing called black liquor, which is a liquid waste byproduct. And they run it out of the plant into these giant evaporation ponds and boil it off. And the capacity to make paper or cardboard is limited by, “Can we hold this stuff?” The waste.
So, Via Separations is like, “Oh, wait a second. We can bring in a system which would take water out, you know, filter the black liquor.” And that essentially is—the plant owner conceives of it as giving him or her extra capacity in the ponds, thereby letting the plant create more cardboard. And it’s all—paper is being driven by e-commerce and all the corrugated cardboard that’s being used these days. So, in order for Via to go in and talk to a plant manager, they have to viscerally understand what the experience that plant manager is living and being able to plug their concept into that. And that really was a design exercise.
Brian: Yeah, yeah. Can you tell me a little bit—I think this is sometimes there’s something here to dig into because some of the teams that I talk to, I think they’re often not handed a great problem, and they’re hired to—I’m thinking about the data scientists I know, particularly with AI, when executive teams are mandating AI without really understanding what problem they want to solve. It’s just kind of like, “That’s what everyone else is doing. We don’t want to be left out.” So, then it actually pushes the problem discovery onto the people that know the solution side.
But they’re not always the ones trained to go find the problems. So, I’m curious, did this company, how did they get into the mindset of going out to find a problem? They already have a solution, but now they have to go map it back to a problem that exists in the market. Did they hire help for that? Is this something where you guys coached them?
Do they have a natural ability? I mean, I can even just think, “How do I find people who need a membrane-based filtration?” How do I go and find out who needs that? I don’t know about everything in the world that needs filtra—or maybe they do. I don’t know. How do you go, “No, the paper mills need that?”
Reed: Yeah, yeah.
Reed: That’s a great question because if you’re interested in entrepreneurship or innovation, you go to a business school, they’ll be like, “Go figure out what the problem is first and then work on the solution.” Whereas a lot of these science breakthroughs are the opposite. “Hey, we’ve got a hammer. What can I do with it?” In this particular case, they were driven—I talked about mission in a [therapy 00:21:24] team.
This team’s mission was reducing energy use, in kind of a climate focus. They wanted to work on reducing energy use. And so the logical stepping stones were kind of like—oversimplified, but what can this membrane do that others can’t? We can control the size of the pores down to—I forget exactly what. Seven nanometers.
So, we can make really small pores, so we can filter out things down to a very small scale. Number two, it’s more robust; it’s stronger. So, it takes more physical pressure for cleaning, or you can have high pH acidic environments that would damage other membranes, and ours is better in those situations. From an energy uses point, they’re saying, “How can we save energy with this membrane?” Well, a lot of industrial processes boil stuff off using heat.
And ours sort of filters, which is less energy. So, the analogy they hit upon in the early days was imagine you were cooking spaghetti? Spaghetti’s done, and you’re trying to get rid of the water? Would you boil all the water off? How long would that take?
But pouring it through a colander takes a lot less energy. So, they then set out to do a walk through industries saying, “Okay, what industries are using heat to separate liquids?” And that’s where they ran into the paper and pulp. And as they were talking to people, just kept asking people, “Hey, any ideas? Any ideas? What industries use heat?” So, that was kind of how they landed on that.
Brian: I bet there’s a magazine or a conference for that. I’ll bet money there’s probably, like—[laugh].
Reed: [laugh] there probably is, yeah.
Brian: Yeah, that’s a great thing about the web. It’s like, “Oh, look, there’s 10,000 people in this group just on that.”
Reed: Yes, exactly.
Brian: If you are asking the right questions, you might find it. That’s great. Do you think the ability to map the solution there—because the value I’m guessing to the plant owner, it’s like, “I can get rid of some machinery, I can spend less money on heat.” And so it’s not like a cost savings kind of approach. That’s more of a sales approach. But thinking about the value to the person that would get it, they don’t really want a membrane, right?
Reed: Yeah. Yeah, you’re right. And in fact, they tried. They thought that was going to be their pitch. Like, “You’re spending a lot of money on the fuel that you’re using to evaporate you out of your ponds. It can save you money.” And that didn’t resonate nearly as well as, “We can help you produce more paper. Your plant’s running at capacity because the ponds are always full. If we make the pond less full, you can produce more cardboard or paper.” And then—you’re right, it was a really interesting thing to sort of figure out what people actually cared about.
Brian: How did they arrive at that? “Oh, it’s not that, it’s this?”
Reed: Yeah, that was at the stage they went out and networked into paper industry and started talking to people who are actually running those plants. And MIT has a great group called the industrial liaison program, they have a few 100 large corporations that pay a membership fee to be plugged into what’s going on at MIT. And any MIT company can go and say hey, “I’ve looked at your list. I see that you have a relationship—you’re the relationship officer who works with, you know, Georgia Pacific. Can you help me get someone on the phone?” So, that’s the kind of thing that they had to do.
Brian: Got it. And are they, kind of like, hanging in the paper crowd now? Is that their heads—that’s their world-space right now, for go-to-market?
Reed: It is.
Reed: Yeah. It is. It is. They thought it would be dairy at first, and then they, once they hit paper, that looked much more promising. So.
Brian: Yeah, yeah. Well, I hope, for the data leaders that are listening to this, this story here, it’s like you can keep hitting them over the head with the membrane, or you can go figure out what the people care about and figure out, how do we slide my solution into something they care about? Making more paper, or cutting costs on heat, or whatever, but even just knowing it’s like, “I don’t care about my cost, but I would love to make more paper.” You don’t know that until you go out and talk. You have to do that research, you have to go out and talk to them and get in their world.
And I think that’s still something that’s not happening a lot with data teams. They’re just focused on, I’m going to make the membrane. Done. Next project. And it’s like, the membrane just stays on the bookshelf, and it doesn’t get used, and we’re wasting time and talent. And so that’s good to hear that that’s happening. [laugh].
Reed: Yeah, and it takes—it’s peeling the onion. So, these—it often helps, in my experience, when the team comes in with a hypothetical, and that’s something concrete that the business side can react to. This company Via, they started with plant managers like, well, “Where does it fit in my process?” And it’s like, “Oh, well, maybe we plumb it in between the pond and the evaporators.” And they’re like, “Oh, my God, what if it breaks or goes offline?
My entire plant shuts down.” They’re like, “Oh, good point. What if we bring it in in a cargo container, and it’s sort of a side loop, and if it breaks, you just stop using it?” So, the progressive layers of what do you need? How is it going to be implemented? Does it mess up your—
Brian: What’s my risk?
Brian: What risk does it add? Yeah. Yeah. Got it. Got it. Just kind of in closing here, I was curious; is there a change in—I know you’ve been in this space for a while, investing in companies, working with startups. Say in the last decade, are you seeing a change in either the teams or the methods to get new products to market, particularly more on the software side? But you could talk, generally, Is there something different, or is it really just—it’s still fundamentals? Not much has changed. The technology advances, the fundamentals of productizing doesn’t change? Or has it?
Reed: It’s, especially on the software side and B2B enterprise software side, it’s changed dramatically over the last couple of decades, in terms of pay-as-you-go on the cloud, open-source, and kind of assembly rather than having to code everything from scratch. And then the go-to-market has become much more consumerized, so if you look at some of these huge developer platform tools like GitHub that are adopted by individuals, and then work their way up an organization, that, even the enterprise software world has learned to use consumer-style experience for discovery, trial, sort of upselling. So, I think there really is a maturity among even the early stage teams now, where they can have, kind of, a shelf full of techniques that they can just pick and choose from in terms of how to build your product, how to put it in front of people, and how to have the experience be kind of a gentle on-ramp. So, I really do think it’s different.
The other thing for us is, a lot of people who are getting into the entrepreneurial world early in career are kind of like, “I’m not going to go to work at Goldman Sachs and make rich people richer by writing code. I’m not going to go to Facebook and work on targeted advertising, no matter how much they’ll pay me because I know ml. I want to do something that I can be proud of in terms of what it did for the world.” So that, especially in the clean tech and climate-related stuff that we're working on, is radically different than the companies we were investing in 10 years ago.
Brian: Got it. Got it. Reed, thanks for coming on Experiencing Data. It’s been great to just catch up with you again.
Reed: Yeah, it’s good to catch up with you, Brian.
Brian: Yeah. Yeah. All right. Well, take care.