There is a category of cost that doesn't show up on your P&L. It's not a line item. Your accountant can't see it. But it is eating your business alive, hour by hour, year by year. We call it the invisible maintenance tax: the compounding toll of systems you built once, assumed were running, and never scheduled time to inspect.
We've watched this pattern destroy otherwise solid businesses for 25 years. Not dramatically. Quietly. Slowly. The way a filter clogs, not the way a dam breaks.
Seth Godin makes the point simply: when you buy something with a filter, create a calendar event. Put it in your schedule. Write down where the spares are. It sounds almost insultingly obvious. But obvious and practiced are two different things. Most founders we talk to have never scheduled a single maintenance event for any of their business systems. Not the operational ones, not the digital ones. Not the ones they built themselves three years ago when they were half this size.
Now consider what one developer built to track his housekeeper's hours: a manual ritual involving driveway camera footage, a spreadsheet, arithmetic, and more concentrated attention than any payroll task deserves. He solved it by building something small and purposeful. But notice what came before the solution: he had been tolerating a deeply broken process for long enough that building a whole tool felt justified. That's the tax. Not the inefficiency itself. The tolerance of it.
Tolerance is expensive. Not in a single burst, but in a drip. Every week you manually copy data from one place to another, every payday you reconstruct hours from security footage, every quarter you re-explain your pricing because no one documented it properly last time, you are paying the maintenance tax twice. Once in time. Once in the cognitive weight you carry around all day.
Here's the thing nobody in the AI conversation will say plainly: the forecasters shortening their AGI timelines are not predicting that your operational chaos solves itself. Intelligence, artificial or otherwise, cannot fix systems you haven't built. A brilliant assistant working inside a broken process just executes the broken process faster. The leverage only shows up when the process is clean enough to hand off. You have to do the maintenance first.
And maintenance, real maintenance, is a design problem before it's an operations problem. OpenAI's head of design talks about what humans still do better than AI: contextual judgment, taste, knowing which thing matters right now. That judgment is exactly what you burn when you spend it on low-value repetitive tasks. Every hour a founder spends scrubbing video footage for timestamps is an hour of irreplaceable human judgment thrown into a furnace. You don't get it back. And AI won't save it for you if you haven't freed it first.
The artists understand something about this that operators often don't. Open Mike Eagle and Kenny Segal's new record is built on a deceptively simple thesis: well-worn territory, mined with enough honesty and craft to produce something original. The insight isn't new. The execution is. That gap, between a familiar idea and a disciplined revisit of it, is where real work lives. Your operational systems are the same. You've heard "automate the repetitive stuff" a hundred times. The question is whether you've actually scheduled the calendar event to go do it.
So here's what we'd tell you to do this week. Not a framework. Not a roadmap. Three concrete things.
- List every manual process you repeated last month. Not the ones you think you'll replace someday. The ones you actually did, with your own hands, last month. Time tracking, reconciliation, reporting, follow-up sequences, client onboarding, anything you did by hand that you've done before.
- For each one, ask: is this broken because no system exists, or because the system we built hasn't been tended? Those are different problems and they have different fixes. The first needs a build. The second needs a calendar event and 90 minutes.
- Schedule the maintenance before you schedule the build. We cannot count the number of clients who hired us to build something new when the old thing just needed its filters changed. We'll always take the work. But the honest advice is: clean up first.
The invisible maintenance tax is regressive. It hits hardest when you're growing fastest, because growth adds surface area to every system you already have, and systems already straining under last year's load don't get better by adding this year's volume. They fail. Noisily, or quietly, depending on how much margin you've built up.
Most of the founders we talk to are at exactly this inflection point. They got to $1M or $2M on hustle and improvisation and raw founder energy. The tools they built or bought in year one are still running in year three, unexamined. Their team has grown around them like vines around a fence post. What worked then is now the constraint. The business hasn't hit a ceiling. It's hit a filter that nobody replaced.
We build the new systems. We also audit the old ones, the ones you haven't looked at since the year you launched them. Sometimes the best thing we do for a client isn't a single line of code. It's a calendar event they didn't know they needed, and the discipline to actually keep it.
If your business is running on systems you haven't reviewed in over a year, the tax is already being collected. The question is just whether you're ready to stop paying it.