Field Notes

When the Expert Leaves, So Does the Answer

Alex Matson

Chief Operating Officer, telos

Aviation safety regulations are written in blood. It’s a proverb you hear from pilots and accident investigators, and it means what it sounds like. Most of the rules currently keeping you alive at 35,000 feet exist because someone, somewhere, didn’t come home. The fuel tank inerting requirement traces back to TWA 800. The minimum crew rest rules trace back to Colgan 3407. You can walk the regulations and read them as a list of dates and names.

Underneath the proverb is a system. The industry has spent decades building a process to capture, analyze, and codify what it learns from failures in people, process, and technology. Every incident gets pulled into that process. Two years later, the lesson is sitting inside training material, checklists, regulations, aircraft design, dispatcher procedures – the next person runs into it without ever needing to know whose mistake put it there. The improvement doesn’t depend on the right investigator being in the room. It lives in the system.

You can see the same kind of scars written into business processes everywhere else, just less deliberately. The reason finance has a second signoff on wire transfers, the reason engineering runs blameless post-incident reviews, the reason that one clause in your standard contract reads strangely until someone explains it – all of those are responses to something that went wrong once. In aviation, the scar tissue is the system itself. Most other places, it’s furniture people inherit without the story attached.

I’ve spent my career on the commercial side of aviation – forecasting demand, setting prices, responding to what competitors do. The safety operation learned to institutionalize what it knows. The commercial operation mostly didn’t.

When a competitor in a market we both fly starts behaving erratically – bouncing prices around, dropping cheap seats into the market and yanking them hours later, doing things that don’t match their usual pattern – it shows up in our booking data within a day. The revenue team uses that data to forecast how each flight will fill up and how many seats to sell at each price between now and departure. Those forecasts assume the world will keep behaving roughly the way it has been.

The question when a competitor breaks pattern is whether to update the forecast or hold it steady. If you update and they were just running an experiment that reverts in three days, you’ve contaminated a model that needs months of clean data to recover. If you hold and they’ve actually made a permanent shift, you miss it and watch your bookings drift wrong for the rest of the season. The right call depends on which competitor, what they’ve done in past skirmishes, how long their last few experiments lasted, whether they’re behaving this way across the network or just in this one market, and whether anything in their public news or earnings explains the move. None of that is in the data.

That kind of judgment lives mostly in the heads of experienced operators. The analyst who has watched a competitor for three years and learned their tells. The revenue manager who remembers what the network did the last time it looked like this. When those people leave, the understanding leaves with them. Nothing catches it on the way out the door.

This is the problem that’s been sitting unsolved inside the travel industry for as long as I’ve been working in it. When we started telos, it’s the question we kept coming back to: what happens to the reasoning around a decision after the decision is made, and where does it go? Almost everywhere we looked, the answer was nowhere.

The design problem became: how do you capture the reasoning chain around a decision – what the operator saw, what they weighed, what they chose, what happened next – and make it available the next time someone faces a comparable situation? That kind of record only exists if the system lives where the work actually happens. You can’t reconstruct it from after-action interviews. It builds through use.

That’s what we’re building inside a major commercial airline. Every resolved situation becomes searchable precedent – what was decided, how the team reasoned through it, which conditions mattered, which ones got discounted, and whether the call held up. A first-year analyst can open that record and see how an experienced team actually thought through a comparable situation. The institution’s reasoning starts to compound across hiring cycles instead of resetting every time someone takes a new job.

The accumulated judgment of an institution is the part you can’t hire in or buy on the open market. It either gets built into the way the work happens, or it leaves with whoever last understood it. Institutional judgment doesn’t need to be written in blood, but it does need to be written down somewhere it survives the next departure. A flywheel that captures it as decisions get made, and codifies it back into the way the next decision happens. That’s where we’re going to start.

telos • 2026