How Technology Can Make Us Better Humans: A Conversation with The Startup Stack
AE Studio is on a mission to create technology that makes us better humans. Plain and simple...sort of.
In a conversation with Louis Beryl, host of The Startup Stack podcast from Rocketplace, AE’s CTO Greg Buckner dove deep into the origin of our mission and how we work to achieve it every day, from the conversation that started it all to the processes that have led us to build some pretty awesome technology along the way.
Read the stories and insights Greg dished out below, including some interesting tidbits about how to pick an agency as a non-technical founder and the future of NFTs.
If you don’t like reading, you can listen to the full audio version on Apple Podcasts, Spotify, Stitcher, and a bunch of other places. Or just click below:
Greg: Why was the JavaScript developer at a loss for words?
Because he didn’t node how to express himself.
Louis: [Laughing] I feel like you could use that in a lot of places.
You guys are a development, design, and data science firm. You do a lot of different things. I would love to hear a little bit more about AE Studio.
Greg: We do design, development, and data science, but the best way to think about it is that we build software products. In today’s increasingly more sophisticated world, that involves living at the intersection of development and data science.
Design is involved in almost all of our projects, UI/UX, product design, etc., and engineering, whether it’s mobile, front-end, back-end, etc. But we also do data science, computer vision, and machine learning. Some projects require just one of those or two of them, and some of them have all three. We have a lot of different colors in our palette that we can paint from, but it’s really about building incredible software products.
Louis: Sometimes you launch startups yourself. Why don’t you tell me a little bit more about that?
Greg: There’s this idea that we’re working on. We’re thinking about innovating on a new model—a new way of tapping into capitalism and tapping into the way people work.
There’s a lot of people who exist in the world today who are incredible entrepreneurs, or they would be if they could tap into that, but maybe for one reason or another they’re overlooked. You can think of first-generation college graduates as an example of this, or people who didn’t go to college or have a nontraditional path. Or people who are older, have a mortgage, have a family, and maybe they need a safety net. We see opportunities to hire people like that as team members at AE, give them support to work with our clients to build incredible products, but then also give them the opportunity to develop their own ideas and tap into that entrepreneurial spirit within the support structure of AE.
We’ll take the ideas that come from our employees and actually build those startups and incubate them inside of AE using all of the same great stuff that we offer to our clients. We have incredible data scientists and designers and strategists and product managers and developers, and that allows us to incubate and develop the ideas of our team members and turn those into startups, and then support them and launch them.
I can’t talk too much about it, but I will tease that we actually had an exit last year, which was really exciting and there’ll be more information on that coming out.
Louis: It makes total sense. You’ve got this incredible team put together, you’re helping companies all the time in a whole bunch of interdisciplinary areas, but you also see opportunities from your own team members and you can put those together and launch businesses. It’s really exciting.
I’d love to take a step back for a moment. Could you tell us the story of how you started AE? What was the insight you had? What were you doing before and how did it come to be?
Greg: It’s a fun story. I was introduced to my business partner a couple of years ago and we had one meal. During this dinner, we really connected on what ultimately became the philosophy of AE and what we do and the things that we care about.
We talked a lot about how there are gaps in technology today. A lot of companies will either manipulate people to do things that they don’t want to do, or just not think about what we know about neuroscience and how the brain works. Like you can’t really multitask, but we have all of these push notifications that are distracting us. Or you’ll subscribe to a product and [companies] make it really hard to cancel.
Technology is increasingly becoming the extension of our consciousness. I rely increasingly on my phone and my notes to basically augment my memory. So we had this free-flowing conversation about how we think technology products can be built in a more thoughtful way, and also a little bit about the future—like brain-to-computer interfaces (BCI) and stuff like that.
Through that, we decided that it made sense to start a company that was focused on building technology products that increase human agency and do it in a way where we’re a consultancy. We’d work with clients to basically build startups, or for enterprises, launch new products or improve their existing products.
But we really want to increase what we call “human agency,” this idea of products and the way that they operate being aligned with the goals of the end user. So we had this really great meeting and we decided to work together. Interestingly, AE actually already existed at that time. It had come out of the ashes of my business partner Judd and my other partner Ed’s previous startup. They had a startup in the food delivery space and it had failed for some interesting reasons, but they wanted to keep everybody on staff. So they were operating the company as an agency, but nothing like what AE exists as today. So I met Judd and they brought me in as a partner in the business and the business just started growing. We kind of saw it as a re-founding at that time. But the company that exists today is much different than what existed then, and that idea of human agency is at the core of everything.
Louis: When you first joined you obviously had a lot of startup experience and a lot of technical experience, but you didn’t really have client services experience. What were some of the challenges that you faced at the beginning of that re-founding of AE?
Greg: I was lucky to have had a lot of previous experience working at a software consultancy. I also had a classical education and a lot of experience practicing agile and specifically extreme programming, which is a version of agile.
Louis: What is extreme programming?
Greg: There’s an incredible book by Kent Beck called Extreme Programming Explained. He was one of the original founders of agile. Basically extreme programming is a way of organizing teams and having business people and technologists communicate to create the best possible product. There are assumptions that the business people express the needs and the business value, and the technology people who are closer to actually solving the problems and building the software decide the implementation. Neither side fully knows the context of the other, so extreme programming and this version of agile creates a really high-level communication framework, without the business people becoming engineers or the engineers spending all their time operating the business.
I had a pretty deep background in learning agile and how to operate that. And so when I came to the business, I brought a lot of that knowledge with me and we actually adapted the process that I had learned previously to the way that AE works. Honestly there weren’t a lot of challenges there. It succeeded pretty well. We had five people when I joined and we re-founded the business a little over three years ago. Now we’ve got over 50 people.
So we’ve grown a whole lot. And I think the value there came from having good processes and knowing how to deliver a really great product and how to increase the agency of our clients and leverage things like agile, but also take baby steps.
Louis: Let’s talk about good processes for a second. What are some of the tips that you have for other companies out there about building good processes?
Greg: There’s process in terms of the delivery and management of the client and the project; we’re highly collaborative and we have an AE version of agile that we love and works really well. But in terms of scaling the business, it’s really interesting. I think it really starts with two things.
One is core values, which we care a whole lot about. The other thing is the framework that you use to run the business. An incredible framework is called “EOS,” the entrepreneurial operating system. The book that’s the starting place is Traction by Gino Wickman. I recommend [it to] anybody who’s running a business, whether it’s an agency or a startup. [Traction] is a really great book and place to start.
But one of the core parts of EOS and how we operate is with our core values. That’s how we’ve been able to scale to where we are today. It [includes] things like [taking] baby steps. We pride ourselves on making sure that as we grow the business, we never take big steps that are hard to reverse. If you take big steps, it’s hard to measure whether that step was successful or not. It’s not experimental and it’s not iterative and it’s hard to reverse it. It also disorients people within the organization. So we always want to take little steps, even if we have a big idea of where we want to go, because acceleratingly large baby steps are like interest in a bank account.
Humans are really bad at understanding compound interest, but we know that like $10 today will be $1,000 in 20 years or whatever. So we [take] the same [approach] with how we make decisions and how we operate the business.
The other thing that we do that’s really cool is called steelmanning. Are you familiar with a strawman?
Louis: Of course. Yeah.
Greg: Yeah, so a strawman takes the weakest version of your argument and uses that to argue against you. The alternative is what’s called steelmanning. So Louis, if you and I are talking about something and I want to steelman your argument, I’ll take your position and I will actually seek to strengthen it and to fully adopt it myself, and actually try to improve upon your argument and stance.
When I come from a position where I actually try to improve your argument, it means that we, together, become a better thinking machine. We’re more likely to either realize that my perspective is right, realize that your perspective is right because I strengthen your argument and convince myself, or you and I together arrive at some third thing that neither of us even realized was an option.
Louis: I love this idea about steelmanning. At Andreessen Horowitz, we used to do something very similar where people would take both sides. Very often it was Marc [Andreessen] himself. He’d be advocating for one thing, and he’d be like, well, let me take the other side of the argument. And I remember this one time he takes the other side, it’s very compelling, and then he’s soliciting feedback. And everybody is completely convinced we should do this thing. It’s like, dead silent—everyone’s completely convinced. And then he’s like, “Well, no, I still want to do the other thing. I was just making the other side of the argument.” He was just too smart for us.
I love the concept of steelmanning. I’m going to completely, unapologetically steal that.
Greg: Please do. You can literally imagine a scarecrow and then imagine it being sheathed in armor and it being the best version of itself. It doesn’t even have to be an argument, it can be you and I together on one side. Myself and my business partners and everybody at AE, we do this all the time. We’ll just say, let’s steelman the other side, and then we’ll rattle off a bunch of points on the other side. And we might say, “Actually, I think we should do the other thing.” It’s a powerful and compelling way to think about it.
Beryl: This is great. Baby steps and steelmanning, those sound like good ways to make decisions and build a company. But I imagine that the EOS is much more than that. How do you think about judging outcomes for yourself and your clients?
Greg: Whenever we’re starting a new project, we want to think about the metrics for success. For us, it’s about deeply empathizing with whatever the client wants to accomplish and then helping, however we can, to bring our own expertise to their vision and work with them to co-create what the best possible outcome is.
We’re in a position where we work with a lot of entrepreneurs who don’t have tech backgrounds themselves, maybe they’re a first-time founder or they’re not an engineer. Our ability to help them understand what’s possible, knowing what the north star is that they’re trying to accomplish, is really important. Every entrepreneur or “intrapreneur” inside of an organization, or CIO or CTO at an enterprise, has a big goal. But we also want to figure out what the intermediate goal is that we’re working towards right now. And how do we make sure that we hit all of the marks that we need to get there?
So for a startup it’s launching an MVP, for an enterprise it might be building a proof of concept. We measure success by making sure that we’re headed in the right direction from the get-go. Sometimes there’s metrics that back that up. Sometimes it’s like, at the end of this first three-month engagement, we want to have 10 people use the proof of concept and validate whether it works or not. Or, at the end of this data science project, we want to roll out the beta of this thing and have 1,000 people pre-pay for it. So the metrics can change, but we always want to make sure that we’re working towards either a really solid concrete goal that’s a step in the right direction or some particular metric that we want to measure against.
Louis: I don’t meet a lot of agencies that do data science projects and software dev projects. Can you tell me what are some of the unique challenges that you face for data science projects?
Greg: Looking back at the agile framework I was talking about previously, the challenge with data science is that you often don’t know if the thing you’re working on or the goal you have is actually accomplishable. So it’s a lot more of a research and development thing.
If you tell me, “Greg, we want to build an awesome app for Rocketplace,” I can tell you with a decent degree of confidence, we can build this thing. There are things that are similar, and we have a rough idea of how long it’s going to take to build. Maybe we’ll be plus or minus five or 10%, but software technology is so sophisticated that it’s pretty clear we can or can’t build something. On the data science side, you have to do research and development and there’s risks because you don’t know if the thing can actually be built.
Now, the benefit there is that with all of our data science projects, whether it’s computer vision or machine learning or NLP, we’re creating valuable proprietary intellectual property, and that actually creates a moat around the business of our client. Because the thing we’re building has never existed before, and it requires smart people working on those things to create it. But the challenge is over-communicating with the client and making sure they understand that there’s R&D involved here. But we’re going to do our best to “yes, and” the thing we’re trying to build into the best possible thing. The way that we do that in terms of the nuts and bolts of the project is we always operate off of hypotheses.
When you’re thinking about building software, you’re usually building features. When you’re thinking about building data science solutions, you’re thinking about answering hypotheses. [For instance], we say, “Okay, we want to build this thing. Here are the five questions we need to answer to get there. I’m going to focus on this question first.” And then that answer is going to lead to this question or that question, and it’ll lead to this or that. As we get closer, we then can start building the proof of concept and actually figuring out what the thing is that we’re going to put into the world. It starts with a lot more questions than answers.
We have a rigorous way to think about that. It kind of takes the abstract idea of data science and makes it more concrete and easier for our clients to understand— and for them to know what we’re building and how we’re working towards that thing, [as well as] what’s possible. That creates a ton of value.
Louis: I want to transition to advice for other companies out there, or potential clients that you might work with.
When they’re thinking about working with a design, development, or data science shop, what are the questions that they should be asking when they meet a firm like yours? And how do they even know they’re ready to work with a firm like yours?
Greg: I think it’s important to understand what you’re hoping to accomplish and what your goal is. I also think it’s really important, when you’re talking to agencies or vendors, to get a sense for their philosophy, their opinion, and view of the world.
For us, as I mentioned earlier, we want to build products that increase human agency. That’s a really compelling vision and it guides the way that we think about things. We truly believe that through using that perspective, we’re going to build better products than if you’re just building without an underpinning to how it works. We also think a lot about the future. One of the things that we’re doing—and this is a 20- or 30-year project—is beginning to do research into BCI. We actually want to play a part in building the operating system for the brain in the future.
Louis: So BCI being brain-to-computer interaction?
Greg: Brain-to-computer interface. So Elon Musk’s Neuralink is a good example of this. There’s also Kernel, which is here in Los Angeles. We’re thinking about the future and we think a lot about human agency. And as you mentioned earlier, we have these internal skunkworks projects that we work on. Those are all things that point to our ethos. That gives clients a lot of confidence and helps them understand that we’re thinking about the future and we’re going to bring that same kind of care and thought to their product.
Louis: I meet founders all the time who come to us and say, “Hey, do you have an agency that can help me build the first version of my app?” How should they think through areas of quality, cost, speed? How would you advise founders out there, especially non-technical founders, on evaluating firms and agencies?
Greg: That’s a great question. So one, obviously, look at the work that the company has done. References are always great. But also, honestly, just ask those questions. We have clients who just ask us, “I’m non-technical, I’m evaluating what makes the most sense here. How do you guys think I should evaluate you? What sets you apart from another agency that I’m going to work with?”
Louis: I’d love to hear how you answer both of those questions.
Greg: The way that I typically answer something like that is: You can speak to the other people that we’ve worked with, but we want to help increase your agency. And we want to work with you to build the best possible thing that you can.
For example, everybody at AE is client-facing. We literally become your team. Something that a lot of agencies do that we think doesn’t lead to the best possible product is they’ll ask you for a 35-page Word document. They’ll throw that over the wall and then they’ll come back in six months and maybe they built the thing, or maybe they didn’t.
Here’s a real-life example. We had a client who said, “I worked with an agency, gave them this document, it came back, and there was no logout button [in the product]. I asked where the logout button was and they said, ‘It wasn’t in the document.’” Instead of doing things that way, we actually just become your team.
Louis: That example is particularly frustrating. I think that a really good agency understands that the client might have one particular goal, but there’s a lot of other user behavior that happens when you’re building technology. You have to sign in and you have to reset your password, for example. The client might not have specified that they want to build a “reset password” functionality. It’s not the goal of their company, but it’s something that is critical to building technology products or account management or other things. Stuff like that— logging out, signing in—is stuff that sounds like table stakes if you’re going to be high-quality designers and developers.
Greg: Yeah. Ultimately, I think you want to find somebody who’s going to be a thought partner with you and who almost feels like they’re actually a co-founder or your first hire. They’re just as invested in the business as you are. That’s how we operate because it’s kind of the only way we know how to do things.
Louis: Can you give us an example of a client project that was really challenging but that ultimately you’re very proud of? I’d love to hear a story of one of these projects that AE outperformed.
Greg: We have a company who’s a client of ours, they’ve been a long-term client, and they’re in the biotech health space. This is interesting because it was a development and data science project. They needed help developing their computer vision algorithm. They essentially have a pee stick [which] you pee on, on either a daily or weekly basis. Then you point your phone at it and it measures key biomarkers that you can get from urine analysis. So things like keytones, vitamin A, vitamin D, hydration, etc. They needed help developing their computer vision algorithm. We started there and we were able to get them to a place, working together, where they were actually able to raise follow-on funding because of the work that we did. And they were able to launch their beta all with us serving as their data science team.
Louis: Did the technology already exist? I imagine that’s traditionally done chemically, not with vision. Was there some question that it was even possible to do it with computer vision?
Greg: It was known that [the analysis] could be done with computer vision, but the algorithm wasn’t there yet, and it wasn’t performing at the level it needed to. So the question was, what are the technologies we need to develop to actually get there?
One of the hardest things was white balance. Most people are doing this in their bathroom. Some people have daylight, some people don’t, some people have a warm light, some people have a halogen light, some have very little light. So understanding and baking that into the algorithm so that it would work in a wide variety of circumstances was super important.
I’m proud of that relationship because the early work was difficult, but we’ve also continued to expand the relationship now where almost two years later, we have multiple data scientists and multiple developers and engineers working on the project. We kind of feel like we’re a partner in the business. I talk to them myself on a weekly basis and we work together to help figure out the best way for this startup to move forward in terms of strategy, but also in terms of the day-to-day. Sometimes there’s things about the process that could be better and I give advice on that. We consider ourselves a real stakeholder in the client and sometimes that involves giving the client advice or input on various things and also accepting advice and input from them on things we could do better.
We ultimately think that we’re doing a great job, but we’re also 1% as good as we could be. And that’s a perspective that we use with ourselves and with our clients. I think things work really well when you have that humble idea of like—we’re great, but we could be even better.
Louis: In 2020 so much changed. Now we’re a quarter through 2021. What do you see changing in 2021? What are your clients asking you for?
Greg: One really interesting thing is NFTs.
Louis: For our audience, what are NFTs?
Greg: NFTs are non-fungible tokens. It uses the same underlying blockchain technology that exists in things like Bitcoin but it enables you to have a file that can only ever be represented once. The interesting thing is it’s getting a lot of traction, for example, in the art world. Because even though there are infinite versions of a digital piece of artwork, there’s only one specific file that is tied to that NFT token. That means that you can buy and sell that specific file as though you’re buying and selling actual art. There’s a lot of other use cases as well, but the art world and this idea of actually having files that can’t be infinitely reproduced is a huge thing right now.
We’ve had a lot of clients who want to see how they can leverage that in their projects.
Louis: Can you give some examples outside of art? Where do you think NFTs could be really useful and proliferate?
Greg: Gaming is becoming an increasingly sophisticated and big area. You can imagine use cases where you sell goods [in a game], but there’s only one version of that particular sword or upgrade or armor, spaceship, whatever. That’s one interesting example.
Another is adjacent to art, and this actually ties in a lot with the pandemic. We have all these digital experiences right now, so you can imagine having a digital experience and you take a screenshot or you capture a little moment in time from a concert that you’re [attending] remotely. That can be turned into an NFT. Really for any kind of digital experience, which are happening increasingly today, you can have an NFT that captures it and you can sell that thing as a memento of that experience.
Louis: It’s fascinating. NFTs [are] this unique thing of one. I feel like a unique thing of one, people are unique things of one.
We live in this world of technology where we have all these deepfakes and all sorts of ways to reproduce people. I’m wondering if we could use NFTs for identity verification. Your personal NFT could follow you around the internet to verify it’s really you.
Greg: There are super interesting ramifications for things like identity or in the future, things like thought and the original version of a thought or an idea. If you can use something like an NFT to prove provenance for that particular thing.
That could be applied to things like the reproducibility crisis right now in academic research. If you can actually prove and verify through the blockchain or through NFTs that this is the actual data that was used in that study and then reproduce it, it helps create greater ground truth for all of science. Or for things like deepfakes that you’re talking about, proving that this is the actual version of the conversation Louis and Greg had and there’s no different version where they were saying stuff that didn’t make sense.
Louis: If you could go back to before you joined AE, what advice would you give yourself when you were just starting?
Greg: I think that it’s really important to make sure that you surround yourself with people who you can think well with. At any given moment, there’s a thinking machine that is the people in a particular conversation. The way that you think and the people that are around you have a big influence on the ideas that are going to come out of those conversations.
We talk a lot about groupthink as something to be avoided, but I think there’s also positive versions of that where you can have the best possible group of people. I think myself and my two business partners, Judd and Ed, are a really, really strong thinking machine. And we actually spend time making sure that we protect and foster that.
With other people at AE, we try to create really good thinking machines in various different meetings and various different relationships and opportunities. That’s a thing I didn’t know was as important a few years ago. I think [it] would’ve influenced the way that I think about interpersonal dynamics and co-founders in past businesses. [You have to ask], are we a good thinking machine? And if not, how can we use tools like steelmanning to improve that? I think that’s a really valuable thing to think about, which also probably spills into significant others and personal relationships and friendships.