Airside Labs
    Code Is Twenty Percent of the Thing
    Newsletter

    Every episode, in your inbox

    One email per Travel Tech Podcast episode: the conversation, the key takeaways and the use cases worth building in aviation and travel AI.

    Episode 36
    October 9, 2026

    Code Is Twenty Percent of the Thing

    Alex Ragin, Amit Ganjoo, Leo Kotil, Jim Hetzel, Íñigo Valenzuela, Christopher Olivares, Iñaki Uriz, Founder, Zoftify & Tourseta; Founder & CEO, ANRA Technologies; Founder & CEO, Overwatch AI; Retailing Strategy, TWAI; Founder, Smartvel; Founder, Elyo; Co-founder & CEO, Caravelo

    AI made writing code nearly free. But what if code was only ever 20% of the job? Host Alex Brooker pulls together seven guests from across travel software, airspace infrastructure, airline operations, corporate travel, destination content, and airline ancillaries who each arrived at that same conclusion independently, unpacking what lives in the other 80% and where advantage moves when execution costs almost nothing.

    Show Notes

    In this compilation episode, host Alex Brooker surfaces a single idea that seven guests reached independently from completely different corners of travel and aviation: code is roughly 20% of a complex software product, and the remaining 80% is domain knowledge, process, and judgment. Alex Ragin of travel software agency Zoftify and tour operator platform Tourseta opens with the number itself, illustrating it through a deceptively simple custom fields feature that consumed five days of internal and client discussion before a line was written. Amit Ganjoo, whose career spans experimental aircraft design to UAS traffic management at ANRA Technologies, adds that complex systems don't fail at the center. They fail at the seams, at the edge cases no happy path test ever reaches. Former commercial pilot and Overwatch AI founder Leo Kotil grounds that in airline operations, where disruption is the normal state and the non standard event is what crews must be equipped to handle. Jim Hetzel, now working on retailing strategy at TWAI, recalls the international pricing specialists at his first corporate travel agency who could save thousands of dollars on a ticket by knowing where to break a fare in ways the GDS pricing engine never could. Íñigo Valenzuela of destination content platform Smartvel argues that as development costs approach zero, value shifts to network effects, knowledge of customer pain, and position in the customer journey. Christopher Olivares, who had never written a line of code before building his travel product Elyo with ChatGPT in early 2025, shows the other side is real, since the 20% genuinely collapsed, while Iñaki Uriz of airline subscription platform Caravelo closes by naming what that leaves: deciding what to build, and why, is now the whole job for every team, not just the well funded ones.

    Read the takeaways

    Code Is Twenty Percent of the Thing

    Every week, someone asks a travel software agency why they can't just vibe code the platform themselves.

    Read the newsletter issue

    Full TranscriptClick to expand (2,203 words approx.)

    Alex Brooker: Hi everyone, welcome to this latest episode of the Travel Tech Podcast. Now this is another one where I've done something a bit different. I've taken one idea that several guests reached on their own, from completely different corners of travel and aviation, and I want to let each of them make the case — or their case — in their own words.

    This one is about a number, and it's not an AI number particularly. It's a number about what software and product actually is. Now, everybody knows, or most people now realize, that writing code has got a lot cheaper using AI agents and these tools. The cost of creating software has massively gone down when your engineers are actually using these tools. But what almost nobody outside this industry has really absorbed yet is what fraction of the job writing the code ever was.

    Five people told me the same thing, but in five different contexts. And one of them was willing to put an actual figure on it. Alex Ragin runs a travel software agency called Zoftify and a tour operator platform called Tourseta. He gets the same question from prospective clients every single week. Over to you, Alex.

    Alex Ragin: As a tech company, we receive so many inquiries from clients, from potential clients, coming to us saying, "We can vibe code the same thing," or "Why can't we just vibe code this?" And of course you can, but the problem here is that it will not be even close. It will not be anywhere close to what you are looking for.

    And recently I posted something on LinkedIn where I explained that our team spent around five days discussing a feature internally and with our clients. And that was a very simple feature — to add custom fields to the platform. So essentially, right now in the platform, you can define, like in a CRM, a set of custom fields, the sections and so on and so forth. The feature is not a very complex feature. I wouldn't say it's a groundbreaking feature, but it's very much needed. It's been requested by a number of our clients, so we decided to build it.

    And we spent so much time just discussing all the edge cases, all the scenarios. We created so many Jira tickets. We tested this from A to Z, like various scenarios. And imagine — how can you make sure that AI, or whatever you use to build this, is going to treat all of those edge cases the way you want it to? There are so many bits that make every single app complex, be it a simple itinerary app where you might have very specific settings or very special rules on how to process this or that situation, and you have to describe them. Otherwise, AI will just decide what to do with that.

    And right now, the problem is that you can build something very simple — and I know that some companies would, as an example, instead of paying certain percentages to Stripe for generating invoices, build an app that generates all the invoices for them automatically. And this is a really good example, but it's a very simple app. It's not even close to being an operating system for travel companies.

    So when we talk about very complex systems — CRMs, ERP systems, GDSs — you wouldn't be able to vibe code them. Because again, it's not only about code. Code is just like 20% of the entire thing. The remaining 80% is knowledge, is processes, is basically training as well, and understanding how those things should work together. So that's the reality.

    Alex Brooker: Five days of argument about custom fields. 20% code, 80% knowing what the thing is actually supposed to do. So what is that 80% made of? Amit Ganjoo has spent a whole career on systems that people rely on but never see. And he has a phrase for where they break. Amit, the floor is yours.

    Amit Ganjoo: Complex systems fail, but they fail at the seams, not at the center. So you have to find those edge cases and where those things are happening. My career arc has always been about systems that people rely on but don't see — from aircraft design to airspace infrastructure, something that's happening behind the scenes, making these complex systems work together.

    It means it's the edge cases — when complex systems have so many touch points, so many interfaces. The happy paths are what people always test and load test and build on. It's always those nuances, those edge cases, which normally you don't think about, which get you in the field. Which is why, if you look at testing a system in the lab versus testing it in the real world, you always find those edge cases. And that's been true in my career as well. I've realized, whether it's building an experimental aircraft or building an airspace management system, it's those edge cases that get you every time. They fail at the seams, not at the center.

    Alex Brooker: Which is perfectly manageable if the seams are incredibly rare. Now, Leo Kotil flew commercial jets before he founded Overwatch AI, and he will tell you that in aviation, the seam is every single day. Over to you, Leo.

    Leo Kotil: There's no standard day. There's always something that happens, usually in flight. Something doesn't work, or there's some operational issue, or you've got to change the aircraft, or the weather. And then the problem is not so much that you have a lot of content, because if everything is fine, you can quickly skim through it and you can understand what's going on. But it's more about the little things, the non-standard things, the things that you might miss. No standard day — there's always something that happens.

    Alex Brooker: Now, here is that 80% with a human face, and it's my favourite thing anybody has told me on this podcast. Jim Hetzel started out in corporate travel agency and he watched people do something that machines simply could not. Jim.

    Jim Hetzel: I started my career in a corporate travel agency and I got to work with some really fantastic people. And I found that one of the arts in a travel agency is the international pricing specialist — or just, the person. There are people who literally specialised in international fares, and it actually was an art. There were people who understood how international fares were calculated, that there could be different ways of calculating a fare, where to break the fare at different points. And they could literally save thousands of dollars on a ticket just because they were skilled at it, in ways that computers could never — the GDS itself, the GDS pricing engine, could never do those kinds of things.

    But there was an art to it. People who could look at an international fare, know where to break it, and save you thousands of dollars on a system that was supposedly already calculating the best possible price for you.

    Alex Brooker: That skill wasn't in the code. It was in a person. It took years to build that specific domain knowledge, which is exactly why Íñigo Valenzuela thinks the whole industry is about to compete on the wrong thing. Íñigo.

    Íñigo Valenzuela: The cost of developing is close to zero. The value is not going to be the code. The value is going to be a mix between the network effect of what you have, the knowledge about the needs, how do you solve different pains, and where are you in the customer journey? What kind of pain are you solving? Not the code — the network, the knowledge of where the pain actually is, and where you sit in the journey.

    Alex Brooker: Now, to be fair to the other side — take all of that too seriously and you end up concluding that nothing new can ever be built by anybody, which is clearly false. Christopher Olivares had never written a line of code in his life before starting his company. Over to you, Christopher.

    Christopher Olivares: There was that side of it, and the other side of it was that I could use ChatGPT at the time to try to build it myself. And so I had never touched a line of code before 2025, and found myself around February with ChatGPT. Complete discovery. And so, yeah, it was enabling in two directions — in the building, and also in the idea of what you can make the product be.

    Alex Brooker: Never written a line of code, and shipped a working travel product on his own. Both of those things are true at the same time. The tools really do collapse that 20%, but they don't touch the 80%.

    So if building is nearly free, what is the job now? Iñaki Uriz runs Caravelo, the company behind most — or many — of the airline subscription products that you've seen launched in the last few years. I asked him what the hard part becomes when everyone on the team can produce anything in minutes or hours rather than days and weeks. Over to you, Iñaki.

    Iñaki Uriz: Nobody knows, right? We all speculate that that's the way it is. But the more I use it and the more I go through it, my inner feeling is that execution is the thing that we are going to lose, in a way. You won't need to execute lines of code. You won't need to execute slides in a deck. You won't need to execute an Excel file. Execution is not the thing. The thing is, what to execute — what to do, why to do it, what's important — and then giving the right orders, in a way.

    So it's funny because these days I've been doing some decks with Claude, with the design feature in particular. And I remember I'm okay-ish with Google Slides or PowerPoint — okay-ish. A couple of guys here are really good. So in the past, last year, when I had, "Oh, we need to really convey this idea a bit more properly, this slide is not good" — I would literally draw something on paper, explain that to the guys, and say, "Please make it pretty and cool and Caravelo-style."

    Now I do exactly the same thing, but just with the machine. So listen, here's the scribble I have — take a picture, upload it, give the instructions: needs to convey this, link with that whole idea, do that. And then a couple of iterations later, I'm done. So it's the execution layer that is going to be lost. But the rest, I think, remains the same — why to do it and what is important.

    Alex Brooker: That is the job. Not the code, and certainly not the slide deck — I said that with a really derogatory tone. Yeah, I love slides most of the time.

    Five people, five corners of the industry, and one number. But code was always 20%. The 80% was knowing where the seams are, which fare to break, what a customer will actually pay for, and what to build in the first place. None of that got cheaper. If anything, it got more valuable, because now there is nothing else left to hide behind. If you can build anything, what you choose to build is the whole decision.

    That used to be a product management problem for a few companies with the budget to build. Now it's everyone's problem. That's another idea for the series. I hope you enjoy this content. I certainly enjoy meeting new people, talking to the innovators and founders across the industry. This episode was brought to you by Airside Labs — that's airsidelabs.com. I'm Alex Brooker. Hope you enjoyed this one. I'll see you in the next one.

    Previous Episode

    Ep 35: The Pricing Fight Inside Every Airline

    Next Episode

    Ep 37: Cancel the Flights So Nobody Misses Connections

    Related episodes

    Building AI for aviation or travel?

    Airside Labs builds working AI prototypes on your data and tests them against the edge cases of your operation. Our agent tools give Claude, ChatGPT or your own agent a catalogue of 6,700+ aviation AI use cases and aviation lookups it can trust.

    About the Travel Tech Podcast

    Hosted by Alex Brooker, founder of Airside Labs, the Travel Tech Podcast features authentic long-form conversations with leaders, innovators, and startup founders shaping the future of aviation and travel technology. No scripts, no sponsors, no sales pitches — just real conversations about building technology that matters.

    Have a story from travel or aviation tech? Come on as a guest.