Support deservesdeeper thought.
Air travel support is full of accepted compromises. We believe many deserve another look. These are the questions and convictions that shape how we think, what we challenge and the standard we want to build toward.
Seven convictions. One responsibility to the passenger.
Read our thinking01 / Customer effort
Don't push work to customers.
It's not a solution if it makes the customer work harder.
There is a tempting way to make support easier for an airline: ask the passenger to do more. Find the booking again. Explain the problem again. Repeat a credit card number or PNR already provided. Read a policy, interpret the exceptions, then contact another department. Search the website, try the app, then call when neither can help. Wait through a survey request that serves the airline, not the passenger. Each shortcut may seem reasonable on its own, but the result is the same. The airline's job gets easier. The passenger's gets longer.
This isn't usually intentional. It's what happens when support is treated as a cost center rather than an experience worth perfecting. Different teams, vendors, and systems are cobbled together across the world, each with different rules, access, training, and capabilities. Nobody has thought deeply enough about how it all works together, so the passenger ends up bridging the gaps. Fixing this requires more than better processes. It requires people whose entire job is to think about customer support, orchestrate everything behind the scenes, and obsess over every unnecessary step. That's how a dedicated specialist can deliver a step change in quality.
Delivering that experience is harder than adding another form, fallback message, or self-service option. It takes creativity, deep thought, and a willingness to see every problem from the customer's perspective. It means finding ways to solve problems rather than simply redirect them. But that's where the effort belongs. Deflecting calls, closing tickets, and shortening interactions may look efficient, but the work hasn't disappeared if it's simply been pushed onto the passenger.
The question we ask
How much work is left for the passenger after we say we've helped?
02 / Quality metrics
Real outcomes beat contrived metrics.
NPS and CSAT are not robust. We demand more.
Why do airlines let customers define the quality of their support? Passengers haven't spent years studying airline operations, experimenting with better solutions, or understanding what modern technology makes possible. They don't know the APIs, the systems, or the alternatives that could have been offered. They're asked to give a score in the moment, often while tired, stressed, or relieved that an ordeal is finally over. And what are they comparing against? Their own past experiences, or perhaps another airline. They're judging what they received against what they've come to expect, not what's actually possible.
An illustrative comparison
Good scores don’t always mean good support.
A passenger tells us how it felt. Someone who knows airline support inside out asks what we could have done better.
| What happened | NPS/ | Reality |
|---|---|---|
| Happy with a rebooking, but a better flight was missed. | Positive | Needs work |
| Happy after three calls to resolve one request. | Positive | Needs work |
| Unhappy about a cancellation, despite the best possible rebooking. | Needs work | Positive |
Higher standards, better outcomes.
Real quality demands a far more ambitious standard. A team with years of hands-on air travel support experience, a deep understanding of airline systems, and a habit of testing better solutions can judge more than satisfaction. They know when a problem could have been prevented, when a better solution was available, and when unnecessary effort was imposed on the passenger. Their standards should rise with every new capability. A 10/10 shouldn't mean the customer was happy. It should mean nothing better was realistically possible. That's an extraordinarily difficult standard to meet, and precisely the point.
Perhaps the clearest evidence that something is wrong with these metrics is the disconnect between the scores and reality. Airline support teams report remarkably high NPS and CSAT scores, while passengers widely regard air travel support as the epitome of poor customer service. Independent organizations hand out glowing ratings, and airlines congratulate themselves on a job well done. How can an industry celebrated by its own measurements be so widely disliked by the people it serves? When an experience widely regarded as terrible can consistently earn excellent scores, those scores aren't a meaningful standard of quality. They're contrived vanity metrics, and it's time to demand much more.
The question we ask
How good is an airline's support when reviewed by an air travel support specialist?
03 / AI, not human
AI simply does it better than humans.
Air travel support is uniquely difficult, and better suited to AI.
Passengers want immediate help, especially when things go wrong. But air travel support faces extraordinary swings in demand. A major disruption can turn a quiet afternoon into thousands of urgent requests, all from people who expect an instant response. Meeting that expectation with human agents is virtually impossible without maintaining enormous teams that sit idle when demand is low. And in an industry with notoriously thin margins, that simply isn't economical. Add the need for around-the-clock service, multiple languages, and complex problems, and it's hard to imagine an industry less suited to the limitations of human staffing.
The human side of this model is no better. Agents deal with stressful problems under relentless pressure, often for rock-bottom wages. Turnover is high, training is uneven, and valuable experience disappears whenever someone leaves. One passenger reaches a seasoned specialist; another gets someone new, with different knowledge, permissions, or tools. The quality of support becomes a lottery, not because agents don't care, but because the system makes consistent excellence almost impossible. These are fundamental problems with the model, not failures of the people doing the work.
AI changes the equation entirely. It can be available instantly, around the clock, in virtually any language, and scale from a handful of conversations to thousands without building a global workforce around peak demand. It can bring consistent knowledge and attention to every request, without fatigue or the pressures of a growing queue. None of this guarantees great support; the technology still needs deep industry expertise, careful design, and access to the systems required to solve real problems. But the opportunity is extraordinary. AI isn't simply a cheaper way to deliver airline support. It's a fundamentally better way to build it.
The question we ask
Why delay a genuine competitive advantage?
04 / Customer understanding
We are not automating workflows.
Customers don't think in workflows. Neither does AI.
The arrival of agentic AI has airlines asking the wrong question: which workflows should we automate first? Refunds, rebookings, baggage claims, flight changes. Each becomes another project, with its own rules, integrations, and carefully defined boundaries. It sounds like a sensible way to introduce AI, but it misses the point entirely. If the goal is to automate predefined workflows, why use technology capable of so much more?
Two ways to design support
New intelligence needs new thinking.
Pre-AI thinking
“Which workflow should we automate?”
Choose a workflow first. Make the customer fit its boundaries.
Disruption
- Ask
- Check
- Act
Rebooking
- Ask
- Check
- Act
Baggage
- Ask
- Check
- Act
The model follows a predefined script inside each lane.
AI-optimized thinking
“What does this customer really need?”
Let AI understand the situation and find the best solution.
- 01
Understand the customer
Consider their needs, context and constraints.
- 02
Plan the best solution
Explore options, ask useful questions and weigh tradeoffs.
- 03
Execute required workflows
The same workflows, chosen and combined to serve the customer.
- Disruption
- Rebooking
- Baggage
The model thinks holistically about the problem.
A horizontal approach is better for both customers and AI.
The real opportunity is to apply AI horizontally across customer support, starting with something far more valuable than executing tasks: deeply understanding the customer. What happened? What are they trying to achieve? What constraints, preferences, and circumstances matter? Often the customer doesn't know what to ask for, or even what options exist. An intelligent agent should be able to explore the situation, ask thoughtful questions, and identify the best possible outcome. Understanding the problem is the hard part. The workflow is often just a means to solving it.
Customers don't think in workflows. They have problems, often several at once, and sometimes they don't even know what they need. Yet airlines and chatbot providers still ask passengers to choose a topic before they can get help. It's perhaps the clearest sign that the system was designed around internal processes rather than the customer. Forcing a complex situation into a predefined category makes sense when building rigid automation, but it makes no sense when the technology can understand the situation for itself. Workflows should be tools that AI uses when appropriate, not boxes that customers must fit into. The goal isn't to automate a collection of tasks. It's to understand and resolve whatever the customer needs.
The question we ask
Why constrain AI's intelligence to fit yesterday's thinking?
05 / Chat, not phone
Chat will make phone calls obsolete.
Customers don't want to call airlines. They're forced to.
Nobody wants to spend an hour on hold with an airline. Yet millions of passengers still pick up the phone because experience has taught them that the alternatives are a waste of time. Websites can't handle their requests. Apps offer limited options. Chatbots send them in circles, misunderstand their problems, or eventually tell them to call. Passengers have learned to skip these channels altogether, knowing they'll probably end up on the phone anyway. They don't call because they prefer it. They call because experience tells them it's the only way to get anything done.
Chat is fundamentally better for both sides of the conversation. A phone call demands the undivided attention of two people, even when one is simply waiting for the other to find information or complete a task. Chat lets both the customer and the agent work asynchronously, managing their time and doing other things while waiting for a response. Customers can share documents, consider options, and reply when convenient. Agents, whether human or AI, can handle multiple conversations, consult systems, and resolve problems without keeping anyone on hold. Everything is recorded, making it easier to review decisions, check work in real time, catch mistakes, and maintain context. Chat doesn't just make support more convenient. It makes better, faster, and more accurate support possible.
We learned this firsthand building and operating our own travel management company. Chat became so much more effective than phone support that we eventually turned off our phone number, offering callbacks only when customers requested them. Our agents worked faster, conversations were clearer, and real-time checks helped reduce mistakes to almost zero. Customers could get on with their lives instead of waiting on the phone, while our team delivered better outcomes with less effort. We didn't force people away from calling. We made chat good enough that they rarely needed to. The future isn't about removing the phone option. It's about making it unnecessary.
The question we ask
Why focus on phone, if it doesn't make sense?
06 / Meaningful security
AI risk deserves an honest conversation.
The future is too bright to dwell on spurious risks.
For years, airlines have deployed terrible chatbots that frustrate passengers, misunderstand requests, and send people in circles. Apparently, that wasn't too risky. But now that AI can actually understand customers and solve their problems, suddenly everyone is worried about the risks. Security reviews, compliance committees, and hypothetical worst-case scenarios threaten to slow progress to a crawl. Why was it acceptable to deploy technology that demonstrably failed customers, but so difficult to embrace technology that could finally help them?
Real risks deserve serious attention, especially privacy and security. Modern AI systems can be tested more extensively than ever, with every change evaluated against thousands of scenarios before deployment and continuously monitored afterward. Sensitive information can be carefully scoped, PII encrypted, model access restricted, and systems configured to prevent unnecessary retention of personal data. We take these responsibilities extremely seriously. But risk must also be judged realistically. If an AI agent accidentally issues a refund, the consequence is a very small financial loss. Human agents make mistakes too, often without the hundreds of automated checks that can protect AI-driven operations. The standard should be demonstrably safer, more reliable support, not an imaginary world without mistakes.
The greatest risk may be losing customers to airlines that move first. For decades, customer support offered little competitive advantage because everyone struggled with the same limitations, especially during disruptions. AI changes that. For the first time, airlines can differentiate dramatically on the quality of their support, potentially more than on their aircraft or flight experience. We're already seeing a similar opportunity with Starlink: fast, reliable Wi-Fi can become a compelling reason to choose one airline over another. Exceptional AI support could be an even greater advantage. Airlines that move thoughtfully and decisively will raise customer expectations, leaving slower competitors struggling to catch up. The danger isn't just what might go wrong with AI. It's what competitors will get right while others hesitate.
The question we ask
What are the genuine risks that deserve deep focus?
07 / Industry depth
Why work with us?
You can't transform an industry you don't deeply understand.
Building an AI agent that can answer questions about air travel is relatively easy. Building one that can genuinely support passengers is extraordinarily difficult. Airline systems are complex, policies are full of exceptions, and even a seemingly simple request can involve multiple bookings, tickets, passengers, and conflicting rules. Yet general-purpose AI companies routinely approach airlines with impressive technology and little understanding of the industry they're trying to transform. Knowing how to build AI is not the same as knowing how to build great air travel support.
We come at this problem from a very different direction. We've built and operated our own travel management company, one that focused on the most complex air travel. We've dealt with real travelers and spent years understanding what exceptional support actually requires. We know the difference between answering a question and solving a problem. We've experienced the frustration of disconnected systems, inconsistent processes, and the countless edge cases that never appear in a product demonstration. That experience shapes everything we build. Our engineers don't simply connect APIs and automate requests. They work alongside people who understand the industry deeply, question established practices, and obsess over the details that determine whether a customer actually gets a great outcome. Industry expertise isn't something we add to the technology. It's what makes the technology valuable.
We believe the greatest advances come from combining deep domain knowledge, exceptional engineering, and a relentless willingness to challenge conventional thinking. That means studying real conversations, experimenting with better approaches, testing difficult cases, and improving continuously. It means refusing to confuse a successful API call with a successful customer experience, or a convincing demonstration with a production-ready solution. We're not interested in building another chatbot or automating a handful of workflows. We're here to rethink air travel support from the ground up, and build something fundamentally better than what exists today.
The question we ask
Does your airline want to lead the support revolution with us?
Next Steps
Less effort for the passenger.
More responsibility for us.
These convictions are a standard to test our work against. Tell us where support falls short for your passengers and your team, and what a better experience should look like for your airline.