PlatformOps, Told Through One Flight School

The easiest way to explain PlatformOps is to leave the data centre entirely and visit a small flight school on a windy airfield. Mia runs it. Three instructors, two trainers, a booking whiteboard and a shoebox of lesson records. It works — the way a paper map works — until the day it doesn’t. Watch what Mia does about it and you will understand, without a single line of code, what platform teams do for a living and why the word “product” keeps showing up next to it.

The flight school that kept reinventing its paperwork

Mia’s trouble starts quietly. Each instructor keeps private lesson notes. Two of them build their own booking sheets because nobody is sure the whiteboard is current. A third maintains a maintenance log on her phone that nobody else can read. When a student switches instructors, the student’s history is a legend known only to one person. Everyone is working hard. Everyone is duplicating work, too — and nobody can say what the school actually offers, because the “catalogue” lives in nine heads and a shoebox.

Software teams have a name for this condition: rumour-driven development. Spotify’s engineers described it years ago — teams rebuilding similar things simply because no reliable way existed to learn what already existed. The fix, Mia will discover, is not more effort and not more rules. It is a shared, maintained way of doing the common things, plus one named person accountable for keeping that shared way alive.

What Mia built, and what PlatformOps builds

Mia’s rebuild has three pieces, and the parallels are exact. First, a single front desk — one place listing every course, every aircraft, every available slot. In platform terms that is the portal and service catalog, the interface layer of every internal developer platform. Second, a standard lesson pack: the approved sequence for a first solo, a cross-country, a checkride prep. Instructors can improvise beyond it, but the pack is the supported route — what platform people call a golden path, an idea Spotify worked with back in 2014 and Netflix called the paved road. Third, a rule about who owns what: instructors own their students’ progress, the mechanics own the machines, Mia owns the front desk. Nobody owns everything; therefore nobody is blamed for everything.

Why PlatformOps wins without forcing anyone

Here is the part managers get wrong for a decade. Mia does not decree the new system. She opens the front desk, makes the lesson packs genuinely better than the shoebox, and lets the instructors switch when switching is obviously easier. One converts in a week; the holdout lasts a month, then quietly stops maintaining his private spreadsheet. Adoption came from pull, because the shared way was the easier way — which is the entire sales pitch of a good internal platform. Industry guidance backs her instinct: the CNCF’s own materials keep platforms optional and composable by design, and the product mindset behind platform engineering says the offer has to be compelling, because a tool people are forced into is a tool they are already negotiating against.

The promises Mia wrote on the wall

The front desk comes with commitments, written where everyone can see them. The booking system will be accurate within five minutes during office hours. If a trainer goes unserviceable, instructors hear before students do. Maintenance questions get answered the same day; syllabus questions within two. In aviation this is just competent operation. In software, these written commitments are what separate a real platform from a wish list: published service levels, stated support tiers, and a clean split of who gets woken up — application teams for their own services, the platform team for its own parts.

Mia adds one more promise that sounds gloomy and is anything but. When a lesson pack is superseded, it gets a formal farewell: an announced last day, an instructor assigned to help migrate stragglers, and then removal, so the catalogue never fills with ghosts. Platforms that never retire anything eventually drown in their own history; the polite goodbye is what keeps the front desk trustworthy.

For anyone keeping score at home, the mapping runs like this:

  • Front desk with everything listed → the portal and service catalog.
  • Standard lesson packs → golden paths and templates.
  • Written commitments on the wall → service levels and support tiers.
  • Mechanics for machines, instructors for students → the on-call split.
  • The formal farewell for old packs → managed deprecation.

What this story is not

Guard against three misreadings. The front desk alone is not the achievement — a pretty portal bolted onto chaos is still chaos with better lighting. Mia’s role is not a renamed office manager; her methods changed, which is what separates real transformation from rebranding, a distinction Thoughtworks warns about in the software version of this story. And nothing here required the school to be big: a small organisation with a well-kept handbook already has the spirit of a platform, scaled to its size. If you want the same story told in industry vocabulary — IDP layers, golden paths, lifecycle stages — the Ops Atlas keeps a plain-English PlatformOps primer that pairs nicely with Mia’s shoebox.

One last note for readers who want the aviation parallel pushed further: flight schools have run this model for decades under other names. Standardised syllabi, published aircraft hours, dispatch desks, named instructors responsible for sign-offs — the industry simply never needed a buzzword for what it already did on Monday mornings. Software is catching up to the airfield, not the other way around.

The moral fits in a sentence: build the shared thing once, keep it genuinely pleasant to use, write down who answers for it, and retire old things politely. Whether the shared thing is a booking desk or a deployment pipeline, that sentence is PlatformOps.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top