Software in the Airlines
some strategies for building
In important respects, the American Dynamism thesis rests on two different questions:
First, can we create wholly new technologies that dramatically change the world of atoms?
Thanks for reading Verticalized! Subscribe for free to receive new posts and support my work.
Let’s call this the hard-tech problem. It revolves around innovations in manufacturing, biotech, AI, and more; wholly new innovations that make the old ways of producing energy, creating pharmaceuticals, and automating factories obsolete.
But there’s another aspect of the thesis that maybe doesn’t get as much airtime, but is vital:
Can we revitalize core American sectors and enable increased economic and productivity gains?
This aspect is concerned with enabling old industries with wholly new software, financial technologies, and labor marketplaces.
On the one hand, we have American Dynamism via disruption, and on the other we have a less sexy, yet highly important task of ensuring that core industries can continue to provide services well into the next century.
These industries come in all sizes with very nuanced workflows, a network of partners, and needs.
And this second aspect is in part why I am such a huge believer in vertical technologies. Because often the key to digitizing certain industries and meaningfully revitalizing some important aspect is only possible through vertical solutions that can unlock these ultra specific workflows.
Which brings us to a sector where we would really like there to be meaningful productivity gains:
Some thoughts on the airlines
I have some friends that are fairly well-positioned to opine on what’s happening at the airlines. And while we’ve talked about the airlines and software off and on over the years, the recent mishaps at Southwest and the FAA seemed like an opportune time to dive a bit deeper into some of the problems.
It was pretty fun to hear their take.
One of the things they noted was the gains in efficiency around most of the trip experience.
The efficiency is two-fold. First, we have far better digital user experiences. Second, we have reduced friction in many sorts of hassles.
The most significant advance is probably around the time it takes to get to and from airports. Uber and Lyft injected dramatic liquidity into the supply of rides with great app experiences and the result is that the car ride back and forth is about as frictionless as it will get.
That alone probably saves 40 minutes or so. And while the airport experience is often far from delightful at least you can bypass some of that through Clear. And ticket booking has never been better. That all adds up to heavily reduced friction throughout the experience.
Much of the inefficiencies now stem from the airlines themselves and their operations, revolving around everything from boarding processes to flight delays to crew scheduling. And all of that adds up to time wasted and real economic losses.
And before we start to blame these problems on the inherent complexity of running an airline, we should step back and remember that in no other industry do we plan for or told to consistently expect delays and disappointment.
Amazon doesn’t care if a package is delayed as a result of weather, a warehouse processing issue, or an act of God: it’s still unacceptable and they’ll invest in correcting the issue accordingly. Bezos and Amazon are obsessed with the customer after all.
And sure airlines have their own set of complexities and tradeoffs, but in stark contrast: when an airline cancels a flight altogether, fails to rebook, and wastes massive amounts of your time, at best you’ll get some additional travel points. And because those points lock you in, the airline knows they’ll make them back, the operations never improve and things continue to fall apart.
But at some point, this is going to change. And it may come sooner than you think. Congress has to reauthorize the FAA by September. With technology being the root cause of the FAA and Southwest debacles, there’s a really good chance we see provisions attached that lead to technology transformation at both the airlines and FAA. I don't want to speculate on what that looks like, but I will note that there's precedence for this. Logistics innovation found its footing with the ELD sensor regulations, fintech likewise came to prominence following regulatory reform post-2008. This could be similar.
So with all that said, I thought it would be fun to pinpoint some areas where there may be opportunity. It’s probably best to view this as a sort of thought experiment in developing software for the airlines.
At the end of the day, better technology will ultimately win. It will just take a force of will, a great product, and the right strategy. So we might as well get cracking on figuring out exactly how.
There’s three important questions that you have to ask:
How do you make it venture backable?
There is no bootstrapping an enterprise ready solution for the airlines.
How do you derisk the technology portion?
Every single airline is going to look at you and want to know why they should trust your team with highly important technology.
How do you derisk industry dependence?
Domestically, there’s 18 major carriers. You should probably design a solution that you can adapt into other verticals or determine a path to expanding into the global airlines.
And one more prerequisite:
I don’t think you need to build new core databases to meaningfully impact the industry. That task seems off the table and it’s not even really that important for where a lot of the issues are.
If you take a look at some of the software companies that provide core database infrastructure for the industry like Sabre, Amadeus, and Travelport, much of their attention is focused on the front office with relatively few strides into back office management. Amadeus has spent time on “clean-sheet scheduling” - basically route planning and in fact has Southwest as a major customer. But that doesn’t touch many of the workforce management issues present.
In fact, Sabre recently divested their crew scheduling software to CAE whom historically has been focused on pilot and crew training. The below tweet is right, and the core databases heeded the call. There simply isn’t a ton of R&D going into back office issues.
So, the problem for a startup is less that you have to compete with the giants… or even the airlines’s IT teams on back-office workflows. The task instead is to reinvent what airline back office operations software should look like. The incumbent risk is not as present as one might think.
Crew scheduling is still mostly spreadsheets. Maintenance occurs in a whole different software suite.
And in fact, these are core issues I think you could tackle. So here are some approaches:
At the outset, there’s some immediate workflows that any system would need to solve. First, pilots and crews have a home base that they needed back and forth from each day. Second, there are negotiated agreements with the unions and FAA regs around how long crews can work - that includes ramp agents, pilots, and flight attendants.
Lastly, the system should have some notion of “surge capacity,” that allows crew additions in the most efficient way possible. Hiccup in the flight journey and the crew is now going to be over their allowed hours? We need to streamline the process of adding crews to flights in order to minimize the delays and cancellation.
Here’s what CAE thinks is crucial for a crew management solution.
What’s interesting is how much overlap there is with broader workforce management. So what if you wanted to make it a bit more ambitious? Could you build crew scheduling as a module on top of a broader workforce management solution?
There’s some advantages to this approach - you dramatically increase the number of employees you serve - jumping from 10,000 pilots to 64,000 employees at Southwest.
And second, it could increase organizational buy-in. Streamline reporting, HR functions, payroll, and solve crew scheduling? That’s a hugely valuable company.
But there’s one big problem: building an enterprise-ready workforce management solution is hugely time-intensive and to the point where it’s not feasible.
But what if you took a lesson from Salesforce’s relationships with Veeva and Vlocity and found a partner? You get the infrastructure, they get a cut of the revenue.
The ideal partner?
Probably Rippling. I don’t think you can build a better framework for HRIS systems than what Rippling has built.
Rippling’s core insight was that the employee graph should be at the very core of everything. And when you build with the employee graph at the outset, you can extend that into all sorts of workflows that weren’t traditionally provided by an HRIS.
And I think that core insight is also what would allow you to unlock industry-specific versions and build an HR version of Vlocity, focused on the trickiest workforce management verticals where building integrations into core industry databases will be mandatory.
And since you aren’t spending tons of time on the overarching workforce management software, you can spend far more time in each vertical, understanding the problem and pain, and then integrating into the core systems to enable complex scheduling, vertical specific provisioning, and more.
The core of the system for the airlines would revolve around extending the employee graph into industry-specific roles: pilot, flight attendant, and ramp agent with their own scheduling components, rule sets for working, and integration data built in.
From there, the task is somewhat known: automate scheduling, feed that information back and forth between the flight data systems and the HRIS, and compound the benefits through automating further aspects of payroll, reporting, and administration.
Besides the technology, there’s a couple other key reasons I think you want to partner with Rippling. First and foremost, with an 11B dollar valuation and plenty of money in the bank, Rippling isn’t going anywhere.
That’s going to be ultra important for the customer base - who will want a lot of guarantees about the underlying infrastructure, especially if they’re going to trust you with hugely important technology.
Second, Rippling is built for these types of industry specific scenarios. They’ve built with one key insight in mind: the product is the integration. And when the product is the integration, integrating into industry specific data sets becomes far more feasible.
Why would Rippling do this?
Verticalized human capital management that can contextualize industry specific data is the future, especially at the largest enterprises. It’s just highly complex to build.
And since Rippling still so much meat left on the horizontal SaaS bone, it probably doesn’t make sense as an internal project this early in the company’s lifecycle.
But the story of Salesforce and their industry partnerships is instructive; it takes time for industry specific instances to reach their full potential and it seems best to start relatively early.
With verticalization amongst the horizontal giants (including Workday) happening faster and faster, it makes sense to get cracking and have beachheads into verticals. Align the incentives properly and Rippling has a strong industry presence that only compounds later on.
In theory, this also cures one of the biggest risks for any venture backable startup tackling the airlines. By following a similar model as Vlocity did towards attacking verticals, you mitigate industry dependence, while at the same time operating from core infrastructure that allows you to meaningfully innovate in different verticals at a high velocity.
So assuming this was actually feasible, where do you go next?
Lookin for a promising path to industry adoption? David Neeleman is probably your guy.
Neeleman is a legend of sorts amongst airlines operators. A 5-time airline founder, he's one of the only guys whom has consistently figured out how to thrive in the industry. He tends to be a first adopter, whether its new airline routes or digital ticketing. He can't help but build great airlines. And after JetBlue and Azul in Brazil both became industry leaders, he’s back with a brand new airline: Breeze Airways.
Breeze is a really interesting design partner because they're branding themselves as a tech company… who happens to build an airline. Get him on the cap table, build with Breeze as a design partner, and you have a really good shot at continuing to go into the more modern airliners before tackling the legacy ones.
MRO (maintenance, repairs, and operations) software is one of those categories that is about as unsexy as it gets and likewise, probably one of those categories that you can bet on getting reinvented this decade. And funny enough, there’s already signs that it’s going to be a large investment category.
MROs occupy an interesting spot inside the stack - sort of as an intermediary source of truth for maintenance and procurement for repair orders, inventory, asset management, and audit trails.
In theory, an MRO system allows a team to log when a part is replaced or machinery that needs repair and surface these repair orders to their relevant partners. Ideally this would then inform capital allocation decisions, procurement cycles, and more.
But it mostly doesn’t happen this way in practice. Or at least not as seamlessly as you would hope. The main pain is around double entry - companies on both ends of the supply chain have to re-enter the same repair or parts order into their system. In turn that creates huge reporting issues, room for error, and massive time sinks - especially when many of these operations are still taking place over fax machine.
While this sort of problem crops up in all sorts of heavy industrial industries, the airlines have it perhaps the worse. Parts delays, faulty procurement, and more adds up to tens of millions of waste and inefficiencies per year.
Within this space, startup activity has been primarily focused on building out network layers that centralize the data between airlines, maintenance and repair teams, and suppliers.These meta-layers then allow all involved parties to have a history of the parts supply chain. It's a really interesting approach and it will be interesting to watch if these sorts of network layers attempt to build out new MRO software over time.
MROs and Other Industries
And while this piece is about airlines, there are plenty of industries with MRO issues.For an alternative approach, I think you could learn some lessons from Rippling, build out a new MRO/EAM platform with the right sort of primitives, the right sort of architecture, and the right sort of integration layer and build a solution that helps a wide number of industries better contextualize their operations
It does seem like a really good time to build. Equipment is getting more complicated. Next-gen manufacturers coming home are going to rethink every process. And as the space industry takes off, so too will the need for really good and real-time data on maintenance operations.
There’s another huge reason that vertical software is a core part of the American Dynamism thesis and one that I will write about in the future: the ability for vertical software to enable businesses to transfer ownership successfully.
And of course, there are certain verticals which will have far more intense human captial management issues, where you can really thrive.
Neelemen is incredible at finding constraints within the current airline industry and then serving them with customer obsession. He’s the closest thing we have to a Bezos in the airlines. The current bet with Breeze is that customers want to be able to travel direct from their place in Indianapolis to their STR in Jacksonville. So Breeze is building direct routes between destinations with no real competition along the route.
MRO, EAM, and CMMS softwares all start to blend together even though they have some distinctions.