The Operating System of a Company

Five years inside founder’s offices taught me one thing I had to unlearn early: the job isn’t to add process. It’s to get rid of process faster than the org can pile it back on.

Most founder’s offices start out lean. Close to the founder, moving fast, everyone in the loop. Give it eighteen months and you’ll usually find a review deck nobody reads cover to cover, a cadence calendar with more meetings than the team has hours to sit through, and a growing reputation as the place where decisions go to slow down. Somewhere along the way, the founder’s office turns into the exact layer of weight it was supposed to prevent.

I’ve come to think of a founder’s office as an operating system rather than a department. It has three real jobs. Decide what gets reviewed and how often. Decide who gets to act without asking permission. And keep information moving at the pace the business needs, not the pace the org chart happens to allow. Get those three things right and the company genuinely moves faster. Get them wrong and you’ve just built a second layer of management, minus the accountability that comes with owning a P&L.

1. Review cadence is a filter, not a ritual

Almost every company defaults to weekly reviews, mostly because weekly is easy to put on a calendar. It’s also usually wrong. The right cadence follows how fast the underlying number can actually move, not what’s convenient to schedule.

Cadence	What it should carry	Who it’s for	What goes wrong when it’s mismatched
Daily	Leading operational signals: volume, uptime, collections, escalations	Function heads, ops leadership	The founder gets pulled into noise that was never theirs to react to
Weekly	Progress against the committed plan, blockers that need someone else to unstick them	Leadership team	Turns into status theatre, everyone reports out, nobody decides anything
Monthly	Trend lines, resourcing calls, tradeoffs between competing priorities	Founder and leadership	The real conversation gets buried under recap slides
Quarterly	Strategy, big bets, org design, capital allocation	Founder, board-adjacent	Quarterly becomes the only room where hard calls get made, so it runs six hours long

Here’s a test I’ve used for years: if the number on the slide couldn’t plausibly have changed since the last time this exact group met, the cadence is wrong. A weekly review of something that moves once a month isn’t discipline. It’s a meeting that outlived its purpose and nobody had the authority to kill it.

This is where a founder’s office actually earns its seat. Not by inventing another review, but by going through the existing ones every couple of quarters and cutting the ones that quietly became theatre. Every organisation drifts toward more cadence, more decks, more people in the room. Somebody has to hold the opposite instinct, and that’s genuinely most of the job.

2. Decision rights are the real org chart

Org charts tell you who reports to whom. They almost never tell you who can actually say yes to something. That gap is where a lot of the slowness in a growing company hides. A decision with no clear owner doesn’t speed up because you added another meeting about it. It slows down further, because now there are five people in the room and none of them wants to be the one who commits.

I’ve found it useful to break this down by decision type rather than by function:

Decision type	Who owns the call	Founder’s office role	Founder’s role
Day-to-day execution inside a function	Function head	Stay out of it	Stay out of it
Tradeoffs across functions, within agreed strategy	Function heads, together	Help them converge if they’re stuck	Only steps in if it’s escalated
Resourcing and prioritisation calls	Leadership team	Lay out the tradeoff clearly, with real numbers	Makes the final call
Strategy, org design, capital, new bets	Founder	Frame the options, pressure-test the assumptions	Decides

The pattern to notice: the founder’s office frames and prepares at every level, it doesn’t approve. The fastest way for a founder’s office to turn into a bottleneck is to start acting like a sign-off layer for decisions that were never theirs to sign off on. The second a function head has to route something through the founder’s office to make it official, you’ve added drag to a call that used to happen in the room, in real time.

The discipline that actually helps is naming which of these four buckets a decision falls into before the conversation even starts. A lot of meetings run long simply because the room never agreed on what kind of decision it was making. Treat a resourcing call like a strategy call and it’ll take three times as long, and still won’t land anywhere.

3. The founder’s office is a compression layer, not a filter

There are two versions of this function, and the difference between them comes down to what happens to information as it passes through.

A compression layer takes a messy, thirty-slide functional update and turns it into the three numbers and one decision the founder actually needs, without losing what matters underneath. It moves information upward faster and cleaner than it would have travelled on its own. It also moves the founder’s intent back down without twelve rounds of people guessing what they actually meant.

A filter layer does the opposite. It sits between the founder and the rest of the org and slows both directions, whether by gatekeeping access, re-opening decisions that were already made in the room, or becoming the mandatory checkpoint every initiative has to clear before anyone treats it as real. This version is heavier than having no founder’s office at all, because now people are routing around a person instead of just working with the founder directly.

The tell I’ve learned to watch for is how a function head describes the founder’s office when it’s not in the room. If the answer is “they get things unstuck,” it’s working. If the answer is “you have to go through them first,” it has quietly become the thing it was supposed to prevent.

How much gets through that arrow, and how fast, is basically the whole value of the function. Everything else (the decks, the trackers, the cadence calendar) exists to keep that arrow wide and quick. The moment those artifacts start existing for their own sake, the operating system has stopped serving the organisation and started serving itself.

The principle underneath all of it

An organisation doesn’t get faster by adding a layer whose job is to check on the other layers. It gets faster when review cadence actually matches how fast the underlying number moves, when decision rights are named before the meeting starts instead of fought over during it, and when the founder’s office judges itself by how little friction it adds, not by how much visibility it has.

Adding weight is easy. Any founder’s office knows how to add one more review, one more approval, one more dashboard. Speed is the harder discipline to hold onto, and it’s the only one of the two that actually compounds.