In a 1991 BYTE review entitled Citrix’s New Multiuser OS/2 I wrote:
Citrix Systems, a team of IBM refugees led by OS/2 guru Ed Iacobucci, has extended OS/2 into the multiuser territory occupied by Unix, Xenix, QNX, Theos, and a variety of DOS-based operating systems.
That’s how I met Ed Iacobucci. He left Citrix in 2000 to found DayJet, a company that aimed to make real the vision of point-to-point air travel that James Fallows had championed in his book Free Flight.
In 2007 I interviewed Ed for IT Conversations. It was one of the most memorable of the podcasts I made for that series.
Ed died in 2013.
The other day, Alan Levine pointed me to Descript, a new tool for making podcasts with transcripts; it features a technique, first shown to me by Laurian Gridinoc, that uses timecoded text as the interface for audio editing.
I’ve been waiting for that technique to appear in a mainstream podcast production app, so when I heard Descript was doing this I jumped right into a remake of the Ed Iacobucci podcast. It was a remarkable experience in many ways.
There was no automatic transcription back then, so while I did a few experiments with Mechanical Turk, there were no transcripts of my shows. I uploaded the show’s MP3 file to Descript and it did a fantastic job of speech-to-text, faster and better than anything else I’ve tried. But that was just the start.
I’m listed as the post-production engineer on that original podcast, which meant I edited the audio for clarity: eliding phrases or sentences, removing ums and likes. Doing that in a waveform editor was laborious to say the least. With Descript I was able to do much more of that kind of editing, better and faster, because text selection as the interface to audio editing is the absolute holy grail.
But I wasn’t just editing the audio, I was also tuning the transcript at the same time: correcting the recognized text (pretty rarely!), tweaking the auto-generated speaker labels, refining the flow to strike a harmonious balance between the words you hear and the words you read.
There’s much more to say about Descript, which by the way deployed a significant upgrade in the middle of my project, but for now I’ll just say: Thank you! It’s a brilliant piece of software that enabled me to revisit one of my most treasured conversations and bring it to life in a way that people can now search for and read, as well as hear, with maximum fidelity. DayJet folded in 2008; it was a remarkable tale of innovation; here’s hoping Ed’s dream will come true.
Here’s a link to the new version of the show’s MP3 file. Here’s the interview as published by Descript. And here’s the HTML export of the transcript.
Phil Windley: Up next on Interviews with Innovators, host Jon Udell talks with Ed Iacobucci, co-founder of Dayjet. Widely known in IT circles as the co-founder of Citrix, he left in 2000 to pursue his interest in aviation. In 2000 he co-founded DayJet, a company whose mission is to deliver on-demand jet travel services.
From IT Conversations.
Jon Udell: I have an ulterior and extremely selfish motivation for having this conversation. So you’ve been to Peterborough when you came to visit us at BYTE. You had to schlep to Manchester and then drive an hour, right?
Ed Iacobucci: And getting to Manchester wasn’t even easy coming out of Florida.
Jon Udell: Now, what you probably didn’t know is that 20 miles west of Peterborough is the town I live in, which has a fabulous, and of course completely underutilized regional airport. We have in Keene an airport where we have, I think the third longest runway in the state.
It can land and does occasionally land Air Force One during political seasons. And back when you visited me, at BYTE in the, what was it like the late eighties, early nineties, we had commercial air service out of Keene that was subsidized by the essential air service program of the government. And so I was actually able to drive two minutes to my airport and I could be midtown Manhattan in an hour, quicker than anyone else could get there.
Ed Iacobucci: But I bet it was hard to get to Huntsville, Alabama, though.
You can fix one degree of freedom, but not both.
Jon Udell: At some point I picked up on this emerging air taxi movement and I read the James Fallows book, and I just reread it the other day to prepare for this conversation and I’ve been basically just desperate for this to play out.
I’m just really curious to know, first of all, how you got into this and second of all how it’s going and, and third of all, most importantly for me, cause it’s all about me, you know, when can I be using my own airport again the way I’d like to?
Ed Iacobucci: (laughs) Well, that’s the objective.
I can’t tell you that we’re gonna achieve it all overnight. It’s one of those frontiers that’s never been properly addressed. You know, I spent most of my career, as you did yours, trying to elevate the computational and communication capabilities that businesses have through systems and networks and such. But one thing that we never really have addressed, on a broad scale, is the same interconnectivity from a physical standpoint.
Jon Udell: And of course architecturally it’s the exact same style, right?
Ed Iacobucci: It’s an awful lot of similarities.
Jon Udell: Yeah. It’s the peer-to-peer air travel network.
Ed Iacobucci: That’s exactly what it is. I love talking to people that understand what I’m saying when I talk technology. It’s a different world that I’m involved in now, but basically, it is peer-to-peer versus hierarchical networks and all the same issues relating to efficiency and best routing in a dynamic environment.
That’s not the only issues, but those are some of the issues. That’s what attracted me to the project. I think if you remember when I left Citrix, one of the things that I wanted to do other than retire, which lasted about two months, that’s pretty boring actually. Retirement isn’t that’s cracked up to be. One of the things that I wanted to pursue, at least intellectually was where do we go with software as a service? Which remember we started exploring back in the late nineties. And Citrix, obviously that took on a life of its own.
And with ASPs and with a massively interconnected and computationally rich web, you know, one of the things that we’ve always done as technologists is focus on the evolution of infrastructure that people use for delivering things.
But there hasn’t been as much of a focus on vertically integrated, highly specialized systems, which, in many ways we always kind of poo-pooed. I mean, real technologists work on horizontal platforms, you know, kind of a macho attitude. In reality, that’s one step, but every pendulum swings.
And I think that the place where there’s a lot of value that can be added by a lot of smart people is, instead of building a highly horizontal platform that appeals to a lot of people, is go the other end of the spectrum and build a very highly specialized, vertical solution that solves one problem really well and leverage that platform not by making it available to a broad audience, cause almost by definition there isn’t one, but focus on the needs of a fully integrated service delivery. And that’s really what drove me to something that looks so different, like a service industry.
Jon Udell: The appeal of the air taxi idea though, is extremely broad. I mean, initially it’s not going to be something that most people, I guess, will be able to afford. But the idea can go extremely broad, right?
Ed Iacobucci: Oh yeah. If we actually dive into the market itself, I like to joke with people, saying that Citrix was a lot of fun, but man, was it hard to explain to people why you’d wanna do remote computing when they didn’t think they needed it.
This on the other hand is the exact opposite. Everybody understands exactly what it is that we’re trying to achieve: seamless point to point travel on a cost efficient basis. And they grasp the concept very quickly, but very seldom do most people think deeply enough in terms of what the complexities are that are behind it, because, as a lot of appealing concepts are, it’s easy to come up with an idea, but it’s really hard to make it real.
A lot of people have had this idea for many, many years. In fact, that’s probably one tha t’s been kicked around for ages, but it’s not until recently — and by recently, I mean, I don’t know, 15 years, 20 years — that we’ve had the kind of tools that you need to really tackle the problem. Not solve it, but just tackle it.
Jon Udell: So let’s go through what the enablers are. Now the Fallows book lays out two broad categories of technological enabler. One is a new generation of general aviation craft, which are smarter, faster, lighter, and safer than what we’ve had, kind of unchanged really, for 30 or 40 years.
So that’s the airplane part of it. And he focuses a lot on that. In fact, a lot of the book is the story of, in particular, Cirrus.
Ed Iacobucci: Sure.
Jon Udell: In parallel with that, and he talks less about it, but I’m hoping that we’ll talk a lot more about it, is this notion that you also have got now this thing called the internet. And one of the things that it can do is make demand visible and aggregate demand.
That if you could go to the internet and put a bid out that says, well, I’m here, and if there’s three or four other people within a 50 mile radius that wanna go to this other place, or various places within a 50 mile radius of some destination, then we can discover one another in a very dynamic way.
That was never conceivable until fairly recently. But I think that you’re talking about also operational routing capabilities.
Ed Iacobucci: There’s a lot of complicating factors and, from a broad brush basis, I agree totally. It’s the hardware and the communications and software, obviously, that all come together. But I mean, at it’s very root, first of all, let’s talk about just general trends. What the real enabler in all this is really what we’ve been doing for 20 years.
It’s Moore’s law, it’s cheap memory and it’s good communications. I mean, that’s really what enables manufacturers to build airplanes better. That’s what essentially the glass cockpits are. I used to joke about it when we first started working on this is that it’s, you know, it’s a thin client.
It’s three pieces of glass up in the front and what used to be individual steam gauges all became applications that run an operating system. It’s highly redundant and, you know, multiple data buses and all sorts of enhancements to guarantee a certain level of reliability but at its core it’s an application running on an operating system. So the IT gains that we’ve made, or the technology gains, I should say, in the last 20 years, come very strongly to play in terms of how these new airplanes are designed.
Now having said that, let me back off for a second and say, I have a slightly different opinion that it’s not necessarily the airplane that’s revolutionary because this is something that I’ve learned in talking to a lot of manufacturers. There’s been a lot of hoopla about, you know, these VLJs (very light jets) are really what’s gonna change the world.
In reality, it isn’t really the airplane. Other than a few things like the integration of information systems, it’s still the same basic laws of physics that come to play. It’s still a wing, it’s two turbo fan engines, it’s one or two pilots. Structurally and aerodynamically it’s not a massive change from what existed before. The cost points are better. The manufacturing advantage that a modern manufacturer can do by intelligently engaging (with the) supply chain. Precision design and manufacturing are incremental improvements. But in the final analysis, a VLJ might give you 10, 20% better capability than a pre-existing design.
The other thing the VLJ lets you do is build volume because, as you’ll soon see, all these models are based on large, dense networks.
Jon Udell: You need a big fleet of taxis for this to make sense.
Ed Iacobucci: You can’t do it with one or two, which is traditionally what people have done. And the problem traditionally is when people independently buy and then let out airplanes to the charter world is that they all wind up being personal things that are all different.
And if they’re all different, then you can’t optimize, you can’t do fleet maintenance programs, you can’t scale. So those are all barriers to scaling. So what you need is a large fleet of identical aircraft that are modern, that can be acquired in fleet quantities.
Part of what this revolution’s about, be it a VLJ or turbine or turboprop, I mean, the airplane technology isn’t as important as the fact that you can make them in large quantities.
Jon Udell: And you mean literally identical, or do you mean interoperable and interchangeable?
Ed Iacobucci: Literally identical. Well, it’s the same thing. From an operations standpoint, we abstract it a little bit and say, as long as they have the same operating characteristics, when you train on one, you can fly any of them. Whereas private aircraft on charter, every one of them is different because they have different instrument sets.
So on the aviation side, airplane design certainly benefited from technology, but it isn’t a revolution. It’s more of a rapid evolution, from a manufacturing standpoint. I think there’s a good shot that, you know, we can start seeing something that looks a little bit more like a revolution in that sense, that you can make a lot of them fast. Be that Honda, be that Eclipse, be that whoever, they’re all gonna wind up in the same place.
But the real revolution, I think, is often missed because it’s least understood. It’s in the business model. You and I have spent a lot of our lives in technology and we’ve seen the effect of creating or identifying new markets, new customers, new value propositions, and how things just spring up outta nowhere.
This is something that’s been part of our lives, our careers. But has not really played out very well , in many industries, one of which is aviation. The core business model of commercial aviation hasn’t changed at all.
They’re scheduled services. You can say they went to hubs and then Southwest did something a little bit less hubby, but it’s all about building a schedule and advertising and getting people to show up. Which is kind of like defining a big freight train that goes around on a schedule and and then you try to reach everybody possible.
And you have big wars in terms of who piles in and how many seats there are. And it’s a fixed market. I like to say, it’s the same one and a half million people that fly from Atlanta or Orlando every year. They fly on different airlines and they all fight each other over who gets the share.
But it’s not a growing market. It’s essentially the same zero sum game.
Jon Udell: And everyone’s going through the same hubs and those hubs are beyond capacity.
Ed Iacobucci: And the charter world, nothing changes. The tax subsidies change from year to year, so it’s more attractive or less attractive for people of means to buy airplanes and then give them to somebody to manage them, to help offset costs.
But it isn’t a fundamental sea change in model. So what we tried to do was kind of abstract that, and I actually had a kind of a false start after I left Citrix. I feel very fortunate that I’ve been able to be in a position where I could actually buy my own airplane which to me, it isn’t so much the luxury of having an aircraft.
It’s really the luxury of being able to schedule your time. That’s really what it’s all about. It has nothing to do with carpeting and entertainment systems or gold plated anything or, cause I’m very down to earth when it comes to that. You know, I travel in blue jeans all the time and t-shirt cause that’s the way I lived most of my life.
Jon Udell: But you go when you go when and where you want.
Ed Iacobucci: But I go when and where I want. That’s the real luxury. And that appeals to everybody. You don’t have to be wealthy to see the value. Everybody has an intrinsic value of time.
So when I first left, I engaged Nancy, my uh, I think when I first left Citrix, we were just married about then, so she was transitioning from fiance to wife, um, to manage our bigger airplanes while I tried to figure out a business plan to make it, you know, like business jets for the masses. And I’ve found out very quickly that that’s a really hard problem because the deck is stacked against you.
The aircraft are extremely expensive. You know, a Challenger 604 is about 25, 26 million new . It’s a very high capital cost. Everything is handbuilt. It’s very high level of service, very customized service. You can’t mass produce it in any reasonable way.
The expectations of the customers are high. The asset utilization’s low. And believe it or not, the competition is incredible because a lot of people have bought airplanes that, throw ’em at the charter to offset costs. So it’s just like all the elements of a crummy business all rolled up in one. And coming right outta software, to me it was a shock to see that this was really such a bad business environment. So basically I hung it up there, I said, nah, I don’t think I wanna spend a lot of time on this.
Then something happened. Two things actually, since you brought it up already. I read Fallows’ book and about the same time I went to Agenda 2001 and Vern Raburn who I knew from years past, most notably from Slate, although he was at Lotus and Symantec also, Vern Raburn was pitching his new sub- million -dollar jet for the masses.
That was his idea. I was skeptical, knowing a little bit about airplanes, but in the same panel we had Dr. Bruce Holmes who was the head of the NASA, well, one of the A’s in NASA is aeronautics, right, it’s not all just space.
He was the chief strategist for the aeronautics, the small airport systems and integration and technology to utilize underutilized assets that we have in our country.
Jon Udell: Just as an aside, for people who haven’t read the book, there’s a subplot about NASA and about Bruce Holmes and about Dan Goldin and about how these guys in a very, kind of underground way, were really seriously trying to figure out how to reinvent air travel and find a way to utilize these 5,000 regional airports that we have around the country.
Ed Iacobucci: Which is a huge capital investment that our country’s made.
Jon Udell: Exactly. Exactly.
Ed Iacobucci: And yet we’re still piling everything into 15 major airports.
Jon Udell: Exactly. I think Bruce Holmes is the guy who clocked all of his travel door to door, for a period of years.
Ed Iacobucci: He still does.
Jon Udell: And he found out that …
Ed Iacobucci: Bruce works for us now.
Jon Udell: He does? No kidding.
Ed Iacobucci: I hired him this year
Jon Udell: Wow.
Ed Iacobucci: To be our chief strategist for next generation air traffic systems.
Jon Udell: That’s awesome!
So the story is that he clocked all of his trips and he figured out that for anything under 500 miles, it was like just as fast to drive.
Ed Iacobucci: Exactly.
Which was, to me was a tremendous epiphany. Of course, we all know that.
Jon Udell: Yeah.
Ed Iacobucci: We just never really actually took the trouble to measure it all.
Jon Udell: Exactly.
So what did he say at Agenda that turned things around for you?
Ed Iacobucci: So I went to Agenda, which was as you know, it’s a technology conference, always has been, first and foremost. There’s a panel there, it was about aviation and it was a real oddball panel. They had Fallows, they had Dr. Holmes and they had Vern on the panel. I was already in the middle of trying to investigate how to do aviation. I was on the verge of giving up and I was looking for other things to do. And it dawned on me that what we needed to make point to point travel a reality was a platform. That’s what we were missing. The platforms were big, expensive handbuilt creatures that were bought by very, very wealthy people and, as I said, underutilized expensive assets.
But what we needed was something that was more akin to what the PC was before we had explosions of local area network. So it dawned on me what Vern was saying about his Eclipse –which was still a concept, this was 2001, he hadn’t really built anything yet– was an approach to what he called making the model T of business jets, and in other words, being the first mass produced precision manufactured aircraft.
And so that got me thinking, so I gave Vern and his wife a lift back on our airplane after agenda, back to Albuquerque with Nancy. And we went and took a look at his new facilities and immediately started thinking of software, network operating systems, and so forth. And that set the wheel in motion.
Jon Udell: So when you use the word platform, you are not just referring to the equivalence of the plane to the pc, but you are referring to the software systems that animate that piece hardware.
Ed Iacobucci: In the early days, I looked at the airplane as the pc. I mean, this is kind of oversimplification, but let’s say that the airplane was the mass produced pc, the engines were– originally Williams and then ultimately Pratt and Whitney was the Intel making the thing that had to be mass produced to make these things a reality.
It was incredibly important, high value component. And then I looked at ourselves as some kind of a blend of an operating system/ network operating system/ application that laid on top of all that and it delivered low cost, point to point travel.
Jon Udell: Because that’s what operating systems do is resource utilization.
Ed Iacobucci: Resource management.
Jon Udell: Okay, so we’re getting down to it. Describe what kind of network operating system you are building.
Ed Iacobucci: Okay, so here’s the thing. It didn’t take very long to look at the business models and say if you take a $25 million airplane and replace it with 25 $1 million airplanes, then it fixes a lot of problems. It gives you a lot of robustness in terms of being able to service individual demands. If you couple that also, by the way, with doing it by the seat instead of by the plane. That lets you interleave packets, or payloads. It became very clear that what we need to do is build a very large self optimizing network that would take a lot of other factors into consideration.
Like the physics of the airplane, the temperature, the loads. The beauty of aviation is that it’s like physics meets business. How much you can carry depends on what the temperature is. The altitude, the runway lengths. Safety is all expressed in terms of parameters that you can operate under, right?
But now the optimizer has to take it into account as it starts shuffling around different customers. So it’s not just a straight optimization, it’s an optimization that has to be done very quickly, real time, and it has an incredible number of constraints, many, many more than you would normally have.
So that’s why it’s so important. You have to eliminate all the optional constraints. Like the customers that want black jelly beans. Well they ain’t, they aren’t gonna be our customers. Right. You have to be willing to take any flight. With no special services. Cause special services complicate things.
Jon Udell: But that’s okay. Southwest has found that people are very happy to go along with that idea.
Ed Iacobucci: It’s the same, it’s the same notion. It simplifies, but, but when you’re talking about a real time on demand network, it becomes even more important. So I had to hire mathematicians. I actually had a group of folks that I’d had previous experience with in adaptive compression algorithms.
Really smart guys. We brought them on, gave ’em a challenge of their lifetime and actually for the rest of their lives. The problem is you never find a solution to the problem cause it’s what mathematicians called NP Hard. Which means that you can take every computer on earth that’s ever gonna be manufactured between now and the end of our lives and put it to work on the project til the end of our solar system, and you’re not gonna find the optimal solution. So you have to move from traditional hard optimization into heuristics that are aided by optimization techniques. And so then we brought in Georgia Tech, who I’m alumni of, and I’m on their engineering board, advisory board, brought in the operations research group. And these are really heavy hitter guys that have been doing optimization. Actually, some of ’em were my professors when I was at school, it was funny to see ’em again. These guys did optimization routines for large air carriers. But the difference, the essential differences is in a fixed schedule, there is a fixed schedule.
And optimizing assets around a fixed schedule is a vastly different problem than trying to determine the most optimal solution in real time for something that doesn’t have any fixed schedule and that’s morphing every request that comes in. Basically their response was, nobody’s ever done this, and these are the, you know, smart OR guys, and since then, MIT guys have said the same thing, you know, a lot of people.
It really is a problem that hasn’t been very adequately addressed. So now it looks really exciting for me cause you know, if it was just like chartering airplanes, that is not very exciting. But now we’ve got new science, new math, it’s a lot of green fields in areas that we can get collaboration of some major universities.
Top-notch people want to work and want to assign PhD students to work with us. An opportunity to set up a group of really smart people. And so we did that. And that’s how we got started. And that was the beginnings of the optimization. And the idea is that every reservation that comes in, you have to figure out how how to value these things.
So we have invented a market model where the pricing is based on how you value your time. If you value your time highly, you don’t wanna waste any time at all, you want to say, I’m gonna tell you I’m gonna be there at nine and I’ve gotta be there by 10 :30. And we know that it’s about an hour and a half flight. That’s a very expensive ticket.
But then if you say, I’m willing to leave as early as six in the morning and be there by noon for something that nominally takes an hour, then we can afford to give a very large discount for that seat.
Without knowing whether we’re gonna match anybody up or not. Statistically, we can determine pricing models. And we did that through, uh, I’ll explain in a second another area of science that we have to bring in. But, with that flexible model, that gave us the degrees of freedom that the optimizers could build some really effective schedules. And so then we used Georgia Tech team to do traditional formulations and run what we would determine in real time. They would run it over the weekend on their mesh of computers and determine if they could improve upon it with adding a lot of computational time.
So we’ve kind of determined that our algorithms and heuristics are within 5% or less of what ultimately might be optimal solutions for problems that you can test. Which are smaller problems. I mean, they can’t use the traditional techniques for a typical network because like I said, it’s far too big.
So for problems of, I don’t know, 10 airports and 20 aircraft, or 50 aircraft, they can test our optimality. It doesn’t work for 300 aircraft…
That was half of the problem. I don’t wanna bore you with a lot of details here, so
Jon Udell: It’s not boring, believe me. But feel free to move along.
Ed Iacobucci: The other side of the equation was that we also had to knock the problem of mesh and how you parallelize things cause these problems aren’t really well suited to be solved in parallel. So we had to build a strategy around that and build a mesh computing platform and all that stuff. That’s kind of like nuts and bolts, but that went on in parallel. But the big problem that we had then, which was one that had been nagging me from the beginning was, if we really succeed at building a new market, which I believe we are, is, how do I know how many people are gonna use a service that never existed before? Especially if it’s people that maybe hadn’t even flown before.
Jon Udell: Yeah, this is the the web 2.0 problem of, oops, I’m too successful.
Ed Iacobucci: Or, how do you determine what is gonna be successful or not? Cause these are really big asset investment decisions you have to make, you have to lay out airplanes, you have to lay out bases and people, and you have to go places and you have to do marketing investments and, believe me, there’s far more ways you can go bankrupt than there are that you can make money because as it turns out, jump forward a little bit, that these value networks, which is really what we’re building, is a value network between regional points.
The composition of the value network determines the load of the value network. Which is, you know, I guess intuitive to somebody that looks at networks. But from a social standpoint, it’s all the same rules apply. If I have a network of nodes A, B, C, and D and I add E, that could have an impact on all the nodes at their various times of the day.
But if I add F it could have impact some nodes differently than other nodes. So it’s an interrelated loading. A problem that’s very, very difficult to determine. So for that, I thought, you know, we’ve got all these guys that are taking basically chaos and organizing it into order so we can file flight plans and make it all look organized or actually be organized on the back end.
But what I need is another group of people that creates organized chaos or organized complexity that mimics the social behaviors of a region of travelers that can be used to test how well we can reorganize that into order.
Jon Udell: So you mean you need to actually be able to model…
Ed Iacobucci: …you need to model the markets and the demand. And that’s not a simple problem either cause it’s gonna depend on pricing and value, time/ value trade offs and density of your transportation network and what nodes you introduce and what the interactions between the nodes are, cause every city has a different effect than others.
For that, I thought, you know, I did study operations research way back when and never really used it heavily, but one of the things I always loved was modeling. And, in fact, before even did Citrix, my other alternative plan was doing something similar to what Maxis did, and that’s built a platform for large scale simulations and never did that.
They did a marvelous job, by the way. I’m thinking the kind of thing that Sim City represented in a generic sense is the kind of technology we wanna bring in. And that happens to be, the modern phraseology for that is complexity science.
When I was at school it was discrete time simulation, but then it got a biological twist to it. And so it became complexity science. At one point in time it was called chaos theory, but complexity science is more, uh, accepted terminology for it. And I had some people that I had done some work with, in fact, through one of my directors , we had both served on the board of a Bios institute in Santa Fe, which was an offshoot of the Santa Fe Institute, which was biological modeling or evolutionary modeling of large complex systems.
So we got in touch with some of those folks, and really, really smart guys that have been doing this all their lives, and we offered ’em a job. We said, Hey, come on board and we’re gonna build the most sophisticated regional traveler model that anybody’s ever built that’s gonna have a great deal of fidelity.
They were gonna use it not just for postulating the future, but actually for building a business. And integrating that as part of our long term planning systems. And so they came on and worked for about four years. And the result was this other piece of technology, which gets married to our system, which is called ABM.
We call it ABM. It’s not very imaginative. It’s agent based model.
Jon Udell: Not anti ballistic missile.
Ed Iacobucci: No, no. And it’s basically, I mean, the simplest way to describe it is, it’s Sim City on steroids. A commercial grade Sim City on steroids. Very targeted on the specific problem of regional travel.
So we’ve got like nine different types of agents or sims that operate and they’re populated according to IRS statistics and 10 square mile zones in the country. And they all have different rules on how they book trips and what flexibility they have.
And then we loaded on top of that a bunch of demographic data. Some we bought, some we got from DOT, some we got from IRS. And then we loaded on top of that all the schedules for all the airlines between all the airports in the contiguous 48. And then we actually developed some driving algorithms so we can estimate driving times. Not great circle, but actual routes and added congestion, time of day congestion through various nodes when driving.
Then we added train schedules and basically had this very sophisticated, very high fidelity model of the transportation options that you would face if you lived in one 10 square mile region of the US and needed to go to another one.
Jon Udell: That’s pretty amazing, that’d be a useful service in and of itself.
Ed Iacobucci: Well, the interesting thing is after we started developing that, that’s when we started doing a lot more work with the FAA, DOT, NASA, uh, actually not only Eclipse, but Pratt and Whitney, then GE when they came into the market, and Honda and Embraer. And so, you know, we’re a little startup company, but we’ve pretty much been in the business of trying to plot out the future, and the manufacturers started sending us design specs for power plants and hypothetical airplanes.
We actually found that we had to build some real Chinese firewalls. Cause we were getting all this confidential data from competitors to see how does this airplane design work in loading, you know? So we were able to make some really quantitative assessment.
Jon Udell: Well, this leads into another question I was getting ready to ask, which is that presumably you’re not gonna be the only such service operating on the planet. At some point there comes the need to federate these things, right?
It sounds like this other piece that you’ve built is potentially part of the answer to how you are able to make that happen.
Ed Iacobucci: Well, lemme back up one second. One thing that I need to add that’s a complexity or actually a reality check, of this whole thing, that was the biggest curve that we had to handle as we went through this. As a technologist, longtime technologist, I’d like to look at the problem, break down in pieces, and then figure out, in this case, how could we build an ASP service? Or how do we get the federation of this across a bunch of people that are buying assets?
Cause I don’t really wanna buy airplanes and run ’em. But it turns out, interestingly and surprisingly, one of the constraints that we have to live with is the regulatory environment we’re in today. Which winds up being a great, big reality test. You can’t build a business that requires this kind of an investment having in your business plan an assumption that you’re gonna change regulations to make it happen.
There’s no investor in the world that’s gonna invest in that. So one of the constraints we had on our business from the beginning was, this has to operate in the regulatory domain and environment that we have today. We better get pretty darn good at understanding how that works.
So right from the beginning, I started making trips to Washington, meeting with senior FAA folks. Tell them, man, listen, this stuff that NASA’s doing that we want to do in kind of a practical implementation of it, it makes a lot of sense, and we also recognize you can’t change all the rules for something doesn’t exist yet.
So help us understand how we follow the rules so that you can at least not feel like you have to stand in our way. So that turned out to be a long road, a very rewarding road, we built some really good relationships with the regulators. But the punchline is because there is such an interwoven relationship between operating control and safety, it’s very hard to say you own the airplane, you guys over there own the pilots, and we’ll run it.
Almost to play by the rules, you have to vertically integrate across the board. So you have to own the airplanes, hire the crews, train the crews, manage all the concurrency of crews, cause there’s limitations of how much they can fly in a day, a month, a year and all that stuff. So you have to manage, that’s all safety aspects, which can be done. It’s all fairly well understood rules, but you have to kind of integrate the whole thing.
Same with the airplane.
Jon Udell: So you do have to kind of own it for now.
Ed Iacobucci: You have to own it all, from the trip formation, the planning, and you have to do it fast. Because if you’re gonna build a system where 95% of the trips are managed by a computer, instead of vast rooms of human beings doing flight plans, which is non-scalable by the way, then you have to be able to embody all those business and regulatory processes in a system that uses the planning tools, and to develop stuff that conforms. So we wound up building a very vertically integrated company. We used our relationship with Eclipse to put in place a very long term, large scale order for airplanes.
We’re working with other manufacturers as well, but mostly Eclipse right now. This has been a long, long, long road that we’ve had to travel because not only do we have to build the core technology for routing and realtime optimization, but had to develop the market models, had to run market studies to back check the market models to make sure that we’re not inhaling our own exhaust or whatever the right metaphor is, right? We had the regulatory side that we had to develop a relationship over four years, and the culmination of all that is a system that we’re about to launch, and it’s gotten a lot of attention.
I’m trying not to hype it a lot. But it’s one of those things that everybody wants to talk about because it’s such a compelling story.
Jon Udell: We’ve been hearing about it for a long time now, it’s about to get switched on in some form that people can see and experience.
Ed Iacobucci: So what this means to you is that you’ll be able to go to that airport, in your community, and be able to say, I need to go to some other airport in my value network. I don’t think it’s the model that says you can fly anywhere you want to fly, everybody flies where they want and it’s totally open.
I don’t think there’s enough scale to make that make business sense, cause there’s economics of flying the airplanes, so you have to focus on what we call day ports, which are kinda like preferred airports where we do the per-seat services between, we’ve identified five and we’re gonna grow that dramatically as we go along based on customer input, actually the reservations that they make themselves tell us a lot about what direction they want to go in.
Jon Udell: So what are those original five non hub airports?
Ed Iacobucci: They’re all in Florida. Because we wanted to keep it all close by here so we could watch the launch carefully. It’s one in South Florida in Boca Raton, which is between Palm Beach and Fort Lauderdale, basically on the east coast. Lakeland, which is between Orlando and Tampa in the central region. Gainesville, which is where the University of Florida is and the business hub, getting up towards the panhandle, and Tallahassee, state capital, and then Pensacola way out on the panhandle. So it’s kind of like a network that initially spans the center of Florida from the east coast.
Jon Udell: And these are all characteristic of that group of 5,000 airports that Fallows talks about.
The regional, underutilized, you know, you drive up to one and there’s not a parking problem and there’s Cessnas sitting on the runway and stuff.
Ed Iacobucci: Absolutely. Two of ’em actually have commercial traffic, but it’s very limited traffic , very limited service I should say. They go mostly through Atlanta, like Tallahassee and Gainesville are the two. To get anywhere else in the network commercially, you go through Atlanta.
The value in this whole thing is saving time, right? I like to say that we’re really not an aviation company or a technology company or a marketing company. We’re in the business of time arbitrage. Some people sell us their time that they don’t need. And we resell it more expensively to other people that can use it. So we use the flexibility one customer has to give , another customer that doesn’t have flexibility a service and aggregate ’em in a way that makes the business make money.
And it comes together.
Jon Udell: When do you switch this thing on? What’s the first day?
Ed Iacobucci: We actually have started turning on some customers in a controlled fashion. Because what we want to do is guarantee service level. We discovered nobody wants to use this as a business critical service if they’re gonna get rejected frequently . They may not get accepted, you know, the capacity, then we may not be able to find a solution to their request in the prescribed time, which is 10 seconds now, by the way, from the time that you tell us what you wanna do, we’ll tell you whether we can do it or not in 10 seconds. And it’s all done virtually, right? Because there’s no real airplanes, or this is all done in advance. So we have a virtual mapping of what the capacity of the network is. And so we can construct a yea or a nay answer.
Jon Udell: So if you make a promise that you can do it, you don’t necessarily have complete knowledge at that moment about all of the components that are that are gonna satisfy that request, right?
You’re making a bet and, and it’s a good bet.
Ed Iacobucci: Well, it’s more than a bet. We know we can do it. We don’t know with what assets, we know we’ve got 10 virtual airplanes that are gonna be available that day. And we know we’ve made X number of commitments for these windows.
And we know we’ve got the virtual crews. Now the reality of it is, which one is it gonna be and what the sequencing is gonna be, and when we pick you up and drop you off, and what order the airplanes fly around in that’s changing all the time.
And what the load will be and therefore what your profit will or won’t be.
And every request that comes in disrupts the previous schedule, but it gives us an opportunity to find a better schedule. It not schedule. It’s planned basically. Schedule implies that we publish it. We don’t, that’s one of the things we had to do that made this all hard, is to operate under the on-demand rules we couldn’t publish a schedule.
Jon Udell: So that also leads to this other question, which is, I heard Esther Dyson talking about this topic recently, and she said, like you said, it’s really not about the airplanes, it is though an awful lot about the customer experience. What’s your sense of how people react to this highly dynamic environment in which I’m presuming you might get a number of updates as to, well now you’re leaving at such and such time, right?
Ed Iacobucci: Well that, you know, this is what we’ve been grappling with for the last six months.
We know how to make it work. We know how to scale it. We know how to make the business work, and we know approximately the demand that we’re gonna see because we’ve been investing a lot in developing markets and customers in advance. What we’ve been grappling with is how do we take this basically entirely fluid system…
Jon Udell: …and present it to people…
Ed Iacobucci: and present it in a way that people can reliably use it. And what that means is you just add more constraints and in other words, it’s not, you’re not gonna be flipped around all over the place. We start with, essentially, you negotiate as big a window as you can accept because the bigger the window, the cheaper the ticket, right? And it’s not a departure window. It’s a window in which we will complete the mission for you. And you basically have to block that outta your schedule. So if you have a lot of meetings during the day, you’re gonna necessarily be pushed to smaller windows, but that’s okay.
You’re more productive and then may cost a little more. But it allows you to be productive. If you don’t have any hard commitments one way or the other, you can give us a bigger travel window, and then somewhere in there, the flight will be generated.
Now, obviously you can’t say to somebody, you know, you negotiated all morning to get me from A to B, which is maybe an hour flight . People don’t feel comfortable with that, so you have to shrink that down. So then the next challenge operationally is, How do you shrink those windows down over time as you get closer to flight time?
Leaving enough space in the windows for disruption recovery, and not flooding on notifications. So there’s a fine balance and we’re learning that. And really what we’ve discovered is the night before we can crunch it down pretty much.
The big windows are useful for us in the days leading up to the travel because that’s when we’re doing most of our optimizing. When we get right before the day of travel, we can afford to crunch that down pretty far. So we crunch the window down to a much smaller size and we notify the customer the night before, that they have to be at the airport by time X, and they’ll be at their destination by time Y and leaving a little room for wiggling around for us in that. That’s our commitment and it’s a reasonable window, you know? So maybe an hour flight they may have to allocate two hours to get there. Which is far better than driving or anything else.
Jon Udell: Exactly, exactly. It’s like, what’s it compared to? So this is absolutely fascinating, Ed, this is so cool.
Ed Iacobucci: Yeah, it’s fun because it’s like, I feel like we’re kind of inventing a market, not inventing, but we’re pioneering a market and a customer set and it’s all those things that we’ve learned over the years, all packaged up in something that doesn’t look like technology, but it is all about deployment and acceptance of technology.
Jon Udell: Wow. Thank you so much for making the time to do this and congratulations on getting to this point. I know you said it was a long road, but on the other hand, I mean what you’ve done in I think not much more than five years…
Ed Iacobucci: right, it’s five years…
seems like a hell of a lot.
Well, it does when you look back on it, but you know, we’ll see. And the true test is gonna be the customers and I think they’re gonna be happy with it. So we’ll just stay tuned.
Jon Udell: All right. Well, thanks so much, Ed.
Ed Iacobucci: Take care, Jon.
Yeah, you too.
Phil Windley: You’ve been listening to Interviews with Innovators, with host Jon Udell.
The post-production audio engineer for this program was Jon Udell. Our website editor was Niels Makel the series producer is Niels Makel. This is Phil Windley. I hope you’ll join me next time for another great edition of Interviews with Innovators on IT Conversations.