For this Deep Tech Dive, I had the pleasure of sitting down with Drew Bellcock, CTO of Pipedream Labs. Pipedream is a network of underground tubes that offers near-instantaneous delivery of objects to and from homes and businesses. They are a Pre-Seed robotics and hyperlogistics technology company that is seeking to create an autonomous delivery network capable of sub-minute delivery speeds for consumers.
- A good CTO can communicate between different types of engineers, and still stay focused on the product and the system as a whole.
- We need economic systems that incentivize environmental responsibility, ethical behavior, and planetary-scale innovation. This will empower more founders to build robust companies that have a positive global impact, rather than something that simply makes money.
- Deep tech companies should focus on having unique applications/implementations of their idea, and not necessarily creating a legal mote around their technology; the focus should be on really great uses of technology. Competition is great as long as we recognize that we’re all part of an ecosystem and we need each other to succeed (to keep investors confident, to build on each other’s technology, to push for new laws/regulations, etc). Deep tech companies could change the world, and it takes more than one big winner to do that.
- For products that are more expensive to build like deep tech products, consider building with extensibility. Identify different features of the product, then design elements of the system to perform only that function. This allows for fully modular products, meaning that you essentially have your own unique lego kit that allows your sales team to tailor the deployment in a very optimized way. This also allows you to build the key features today, but still, add onto them in the future — very important freedom for capital-intensive startups.
- Be on the lookout for robots-as-a-service and AI-assisted decision making, think Jarvis from Iron Man but more like an extension of the user’s brain, rather than a separate entity.
- We need to shift educational thinking from proof of knowledge to proof of work. Someone’s degree shouldn’t dictate whether they are qualified for a job, how they present their thinking through real-life projects and problem-solving are what matters.
“If you never change your mind, then you’re not really learning, adapting, or participating in reality. It’s so important to find kind people who think differently than you, and have real conversations about ideas.”
This interview was edited and condensed for clarity
Can you tell me a little bit about yourself and how this got you into the field of robotics?
I’ve always been interested in building and designing things from an early age; I was doing that all through high school and decided to go to college to be an engineer. I chose mechanical engineering for my bachelor’s and then I went into aerospace engineering for my master’s. I just like building things at a system level that solve actual problems. I am very much a first principles thinker, identifying problems and then finding creative solutions to those using technology. That’s what I enjoy doing and that’s what ultimately got me into Pipedream, underground logistics, and the robotics to go with that.
I had the luxury of an actual pure mechanical engineering degree, which is basically just an intro to all other forms of engineering. Some modern physics, some aerodynamics, some fluid mechanics, some material science, and a lot of electrical stuff. I like to learn everything I can about technology.
What’s the story of you coming to be the CTO at Pipedream?
My co-founder, Garrett McCurrach, was my roommate in college and we’ve been best friends since. We’ve actually done a couple of software ventures together and a few smaller projects. We really like working together and we have complementary skill sets. Pipedream was an idea he originally pitched way back in 2015. It’s just the ability to move things around really fast using tubes, but I kind of made fun of him for it at the time. It wasn’t as refined as it is today. But little did I know he has been thinking about this and building the economic use case for it ever since.
Last year, right as the virus was starting to go around he shared with me a small prototype. He had refined the business use case and got me hooked! So we started having conversations about it more regularly, till March he formally incorporated the company and asked me to come on as a Co-Founder in the CTO role. He knew the business side of things and I understood the engineering design, from circuits to systems integration, so we were a good fit for each other.
I have a pretty good depth of knowledge in mechanical, aerospace, electrical, and intermediate knowledge on almost every other form of engineering. So I think that puts me in a unique position to be a really good CTO because I can communicate between different types of engineers, and still stay focused on the product and the system as a whole. It’s really important that you get along with your co-founders, so I believe it was important for him to bring on someone that he knew and could work well with. We just have a good history together.
What is the goal you’re trying to accomplish at Pipedream?
What Garrett pitched me back in 2015 was to be able to teleport things. I mean if we can figure out how to teleport things, we’ll be doing that instead. I’ve personally worked on it a little bit, (I study quantum physics on the weekend for fun) but that kind of innovation won’t be feasible for a century. So we decided to do something on the decade scale, and that is moving things around really fast, efficiently, autonomously in an environmentally friendly way.
On the economic side, we (everyone, not just PipeDream) are going to have to restructure how we think about value; to not just be about the money, but also about the social implications and the environmental implications. We have to have a more holistic economy.
How will you guys help with achieving this holistic economy?
From a high-level perspective, a lot of people turn to drones. With my background in aerospace, I’m very familiar with that. I’ve designed full systems from the ground up and that works great for high-value items that need to go to locations on the fly without the route being planned ahead of time.
However, drones can’t be the backbone infrastructure for delivery because if you think about the volume of things being delivered, the skies would be literally covered with drones. Also, it costs 25x more energy to send something using a quadcopter versus driving it in a small robot on the ground. Once you understand that, you realize ground transportation is obviously how things have to move around. That’s how things are currently moved around today, but in a large vehicle with a human operator on the roads. If we wanted to go even faster we would need to move things in individual packets, instead of batching them inside full-sized vehicles. Packetizing deliveries using current systems is just not possible though; congestion at the surface is already too high.
That’s the main constraint logistics robotics companies likeStarship Labs have with their sidewalk robots. It’s useful but again, cannot be the backbone of our delivery network. So that’s how we arrived at subterranean tunnels. By putting robots in an enclosed environment it allows us to move very quickly, and also safely. We can bury things underground where there currently isn’t a lot of blockages (besides sewers, electricity, etc. — utilities that everybody already has).
I use Amazon almost daily, probably five days a week I get a package, and it’s only going to get more pervasive in the future. So if you think about deliveries as a utility it should be underground with all the other utilities, in its own dedicated channel. That’s how we arrived at the concept of Pipedream.
I would assume that’s a similar thought process to how Elon Musk came up with Hyperloop & The Boring Company.
Underground tunnel for The Boring Company
That’s an interesting analogy. I believe they arrived at the tunnel conclusion for similar reasons as us. There’s congestion at the surface so if you want to move at high speeds, over long distances, you need a dedicated channel underground to do that.
However, there are a lot of fundamental differences, such as scale. We’re in a different order of magnitude than their tunnels because we don’t have people in our tunnels. That removes a lot of constraints, such as having the right levels of oxygen and completely different safety requirements. Our system will still be safe but if a robot crashes underground we can send in another robot to get it out. If a human crashes underground, that’s really really bad. So yeah, it’s a whole different level of complexity, and honestly, even the drilling technology that goes into that size of tunnel vs what The Boring Company needs to do is completely different.
What makes now different from 6 years ago when the idea for Pipedreams was first conceived?
Oil and gas companies have really pushed drilling technology and the installation of fiber optic networks for the internet have scaled-down drilling innovations cost-effectively. And that’s honestly one of the big differences, between five or six years ago when the idea was first discussed
Great to hear that the success of other companies like Hyperloop, helps you guys out.
Yes, the greatest gift that Hyperloop or The Boring Company would give us is less regulation for tunnels from cities, governments, and utility operators. Some people are scared to dig underground, especially in an urban or suburban environment. They don’t want to tear up the underground infrastructure or create risk unless there’s a high value and it’s very well understood. Elon is trusted in the crazy idea space. So when we go to governments, we can say if this is important enough for your existing utilities,, then it’s far more compelling to let us take a much smaller tunnel that would be used more frequently. This is particularly true in places like Arizona, Nevada, or California where they’re already letting the boring company operate.
It’s cool to see how Elon’s impact is shown every day. Would more competition in deep tech mitigate the struggles within the industry?
It’s interesting that you say that we need more competition in deep tech and I agree with you. I think one of the major struggles of deep tech companies is getting enough funding. It requires so much capital because it’s such a long play and if it doesn’t work out there’s a risk that it could end up being like, like the biotech category or Big Pharma. Where they waste Billions of dollars and don’t know, for five years whether or not it’s going to be a good investment.
The key for deep tech companies is not to stay away from having competition on the technology side but on the application side. We should be competing to improve the technology, but when somebody wins, I think it’s great if they can apply that directly to a specific use case and own that use case. Then somebody else can figure out some other way to achieve the technology and apply it to a different use case. That’s something that doesn’t happen in Big Pharma, if somebody wins at a certain drug, they try to dominate a bunch of different use cases. It almost guarantees that everyone else will have just wasted their capital, which makes investors more standoffish in the future. It’s become a vicious cycle. Competition is great as long as we recognize that we’re all part of an ecosystem and we need each other to succeed to keep investors confident.
This ascertains my thought that we should start to look at deep tech like a community.
That’s a really good point, thinking about it as a community where everyone can win. When something good happens, we can all benefit as opposed to a strictly capitalistic competitive environment. And I know there are benefits to that, but in effect, we’re trying to change the future of humanity, and I think that’s gonna take a lot of people being successful to achieve.
My dream is there’s going to be so many things built off of our idea if it works. So we question ourselves: how can we start fostering community and people who are going to build the peripheries off of our system to make the world even better? And how can we start doing that now? Because that’s gonna take time to grow that community too.
So how is the companies roadmap looking like?
We have a half-scale prototype and a test track that we have used to figure out component placement and how we’re going to integrate things and what the functionality of the system looks like, from a product definition standpoint. Earlier today, I was at a meeting with some of our engineers and Garrett, to define what our next step looks like, which we call our technology demonstrator. Right now we have multiple elements to our system:
- Pods that carry things through the tubes.
- Payload bay inside of those pods.
- Human interface locations
- Pipes and switching points
- Communications and control system
Our roadmap is well-defined so we can build our product to be extensible.
What do you mean by building extensible?
Right now we’re launching the design phase for our technology demonstrator, which demonstrates our core features, the ability to move something, the ability to autonomously interact with users while demonstrating safety, efficiency, etc. From there, we’re gonna start adding on additional functionality that deals with the overall infrastructure. It’s so different depending on where you’re installing it and what it’s going to be used for. We’re trying to find really good standards for how we build things that allow them to be expandable in the future and customized to people’s specific needs.
A lot of industries design a great product that has every possible feature and then sells that same product to everyone, which costs a lot of money. I don’t believe that’s a good economic model. It’s hard for them to make the sale to as broad of an audience as they’d like, because some people are going to be basic users, and they’re going to want basic features. They don’t want to pay the premium price for the premium features that they’re not going to use. So we’re trying to figure out how this can be something everybody uses. So how can we identify different elements of the system and build them in such a way that if somebody just wanted to do a specific thing, it can do that? But then if they change their mind in the future, we can expand it also, without the initial costs in the initial build. So that’s what I mean when I say we want to build it in an expansive way.
Any big wins so far?
The clarity in our roadmap. This allows us to start approaching partners, investors, and cities so they can warm up to the idea and not think we’re crazy. Other than that, we got accepted into Techstars. Obviously, there’s a lot of clout that comes with joining one of the world’s top accelerators. But the biggest thing is the network of people that we get to interact with daily; brilliant founders, amazing investors, great mentors. For instance, we have two people who are high up at Amazon, and really understand the logistics space. We also have access to people who have accomplished crazy hardware ideas in the past that we can learn from.
How has the networking been?
We’re at a really interesting time with the pandemic causing people to switch to all remote. Twitter has taken off in the startup and investment world, so there’s a lot of great casual interactions between founders and investors there, along with Clubhouse and Slack communities. When you find alignment with people, things organically happen. It’s a continuous relationship, we’re getting to talk to investors but we’re talking to other founders, so we’re building momentum behind the idea.
What are some interesting companies you know of in robotics and AI?
There’s so much to say about robotics. Two of our mentors are former execs at Sphero, the spherical-like robotics company, that provides children with STEAM-based educational tools to learn from. One of them, Ian Bernstein, has gone on to found a robotics company called Misty Robotics. Their goal is a robot in every home to help out with your daily tasks. Also, his co-executive Jon Carroll started his own company named Company Six, and they do reconnaissance and surveillance robotics.
One of the interesting things that John’s doing is providing robots-as-a-service instead of as a product, so they’re changing the financial model that goes into a robot. He sells people the robots as a subscription and offers warranty support, but I think that is an interesting point to make these really complex robots that are capable of doing like, world-changing tasks more accessible to people. Really good robots like those made from Boston Dynamics tend to cost $70k to $500k, which limits access to them.
There are also some interesting research areas going on in the way of powering quadcopters and electric aircraft. There’s a company called Forcyte that’s working on wireless power technology over long-range, to power things like drones, or anything that consumes electricity where you cannot have direct access to a power grid but the battery is not enough. It’s really interesting because that will allow more robots to be placed in flexible locations. It doesn’t have to go back and recharge every 10–15 minutes, for example. The most compelling daily use cases of AI beyond where they’re already integrated into Google as AI-assisted decision making. Presenting people with a lot of information, but also the best context for that information possible so still allowing humans to make the final decisions on things. Robots that can do a lot of background research and preemptively answer a lot of the questions that people would have in order to be able to make a decision.
Are you talking about Jarvis from Iron Man?
Yes, but less interactive and more predictive. Instead of having to ask for something it could predict when you would want to know something, what questions you might ask based on what you’re currently doing, or what you’re planning to do in the future. That could apply directly to engineers who could set some key parameters that they want for a design and then let an AI, iterate through and find the best possible solution to those parameters. Then, humans can double-check it again. That type of collaboration is going to be the most compelling use for AI in the future and I think that scales into robotics. This could look like robots performing daily tasks such as cleaning around the house, folding laundry, possibly even some basic preventative health care tasks that can be automated with AI.
Robots should do things that will be more helpful to humans so that people spend less time doing low-value things and get to spend more of their time doing high-value things. I believe there’s gonna be a shift away from task achievement in our economy, towards creativity. Humans are the only beings that can truly be creative, and AI is creative only within the parameters of its creator. And it can never exceed that. This is why I’m not worried about a robot uprising unless people programmed them to uprise. AI will be a great tool to extend a person’s capability, not replace it.
Which could happen right?
Yeah, I think that could happen. But it will allow us to automate the low-value repeatable tasks like I was saying, and humans will be freed up to spend their time on creative things rather than moving into a self-indulgent state, like in Wally. What I hope will happen is people will be more financially free, they will have more time in their lives because they won’t have to work as much. And they can focus on doing creative things, I think we’re going to see an economy that starts to value that more, and we’re also going to see hopefully, happier people and a more successful society.
What’s something that could happen externally that would benefit your company?
The continuation of miniaturization of electronics, components, and figuring out how to subdivide complex tech into interconnective blocks more like Legos; more inter-compatible options like what allowed computers to be so successful. Initially, they were all built on one board, but then people defined a framework for what memory looks like and how it interfaces with the other components. Then they did the same thing for processors, graphics cards, etc. and that’s what made the internet possible. Seeing that happen for components in robotics like batteries or motors would be huge.
Reminds me of molecular nanotechnology (MNT), nanoscaled robots that can self-assemble.
Yeah like the replicators from Stargate. Where they have these micro-robots that can self-assemble into other things. They can become ships, humanoids, or whatever else they want. It may not happen to that degree but it’s possible. MIT had an experiment with robot cubes that self-assembled with a few basic functions built into them, but they weren’t microscopic level.
To me, the paradigm shift is in the way we think about robotics and things that are going to be industry-changing. We standardize every interaction or every type of interaction, then it allows everyone to build a product that’s compatible with everyone else’s product and you get customized solutions to different applications much faster, much more efficiently. And the innovation starts to snowball from there, I think.
Where should someone without a degree start if they want to be a part of the robotics industry?
At the surface level, I believe there already is a shift right now, away from formal degrees being the prerequisite to everything. Certificates or alternative education has become an even better modality of education, especially for technical things; it’s just about doing interesting projects on your own. Instead of proof of knowledge, proof of work for your skills that’s the top thing I ask candidates about in my interviews.
I don’t care where your degree came from, what’s it in, or even if you have a degree. People need to be focused on identifying small problems, building interesting solutions to those, and then stair-stepping that and building out a portfolio of interesting things that they’ve done to show creative thinking. So if anyone wants to work at pipe dream, have a really interesting portfolio of projects you’ve done in the past.
Originally published at https://www.datadriveninvestor.com on April 20, 2021.