The drift was always there

I started as a welder.

Six years staring at a 6,000-degree arc of fire, knowing my body wouldn't last forever. I left the heat for a Compliance Technician job in the late '90s — somewhere clean, fluorescent, with monitors that didn't try to burn through my visor. I got there in time for the gold rush. I got laid off in 2001 along with everybody else.

No degree. A skillset for an industry that had just stopped existing. My dad gave me one line:

Attack what scares you the most. Conquer it and be free of the weight.

For me, that was college. The accounting degree I'd told myself wasn't for someone like me. I went and got it. I never looked back.

What followed was twenty years across fourteen companies. Public accounting, FP&A at EMC, HR Comp, RevOps. Eventually running Global Sales Comp at CyberArk. Ten of those fourteen companies closed, merged, or got acquired. Ten times, somebody in a room I wasn't in made a decision about the future of my badge. Never about performance — I want to be clear about that, because the part of me that still has something to prove needs to say it out loud — never about performance. Always about the logo on the door.

After a while you stop being surprised. After a while you stop being angry, even, which is somehow worse. You start being tired. And then the tiredness curdles into something else — a quiet, organized rage that you walk around with like it's just part of your weight now. I lived with that for a long time. I am writing this partly to lay it down.

But that's not the part I want to write down first.

The part I want to write down first is the ghost.

There's a ghost in every comp org. It doesn't matter what HRIS you're on. Doesn't matter if the ICM is Varicent or Xactly or a spreadsheet held together with INDEX MATCH and prayers. Doesn't matter if the CRM is Salesforce or HubSpot or that one custom thing the former VP built before he left. The drift is always there.

It looks like this.

A rep gets promoted in Workday. The promotion fires Monday. The ICM doesn't know. They get paid their old commission rate for six weeks before someone notices, and someone else writes the retro adjustment, and someone else's Sunday gets ruined.

A territory shifts in Salesforce. The HRIS doesn't reflect it. The accounts go to nobody. The pipeline data starts looking weird three weeks later but the dashboards don't flag it, because nobody wrote a rule that says accounts can't be unowned for thirty days. So a chunk of pipeline quietly disappears.

A rep leaves. Their employee record is terminated. The ICM still has them coded for a bonus pool because nobody told the ICM. Three months later, finance pays a commission to a person who hasn't been at the company since February.

Every operator I've ever worked with has lived with some version of this. The systems aren't broken. They each work fine on their own. The problem is they don't agree with each other, and nobody is the referee.

I built one. As a hobby. An automated reconciliation agent that crawled through HRIS exports and comp data and CRM exports and surfaced the gaps. I showed it to our SOX Director, expecting some version of neat. She looked at the design and admitted it was amazing. Then she looked me in the eye and asked:

Who's making sure the automation isn't creating gaps or missing data?

I didn't have an answer.

I remember exactly where I was standing. I remember the color of the chair she was sitting in. I walked back to my desk and tried to act normal for the rest of the afternoon, and then I drove home with the radio off because the question kept rolling around in my head and I didn't want anything else in there with it.

That question — that exact phrasing, sitting in the back of my brain for weeks — is the reason this thing exists.

Here's what I realized, slowly, after that conversation.

We are racing into an AI-augmented future where every operator is going to have agents doing the work that used to take humans. That's good. I want that. I have built agents myself, badly and well, and I have watched them deliver in an hour what used to take a team a week. But every single one of those agents creates a new gap. Every one of them assumes the data feeding it is right. Every one of them quietly bets that the system on the left and the system on the right agree with each other.

They don't.

And the SOX Director can't sleep, because she knows it. The Controller can't sleep, because she knows it. The Comp Lead is working at 11 PM on a Sunday before a Monday run, praying the data synced, because she knows it. Everybody in the building knows it. Nobody has a tool that proves it, one way or the other.

I looked for the tool.

It didn't exist.

So.

The name is OrgDrift because that's what it is. The organization drifts away from itself. The HRIS drifts away from the ICM. The ICM drifts away from the CRM. The org chart drifts away from the comp plan. Every reorg fires another round of micro-drift that nobody catches until somebody, eventually, gets paid wrong, or doesn't get paid, or signs a quarterly attestation that turns out not to be true.

The brand mark is a moon over a horizon line, with a reflection underneath. The moon is the system you trust. The horizon is the boundary between what you can see and what you can't. The reflection is what would be true if everything agreed. When the moon and the reflection don't line up — that's drift. That's the whole pitch in one image.

(I have feelings about the mark. I redrew it more times than I should admit. There was a Saturday where I spent the whole day rewriting the Canvas animation and ended that night with worse output than I started with, and I closed the laptop and didn't open the file again for a week. I am not proud of that week. I am also not embarrassed by it. The mark currently sits at four states — aligned, drift, scanning, sealed — and I will probably redraw it again. Probably soon.)

The category is Revenue System Integrity. The measurement is the Organizational Drift Index. The tribe is Drift Hunters. None of those words are perfect. All of them will probably get refined. But they are the working language, and the working language has to start somewhere.

There is a strategic argument I want to put down before it gets lost.

Platform consolidation is accelerating. Salesforce bought Spiff. Rippling is at sixteen billion dollars consolidating HR, IT, and Finance onto one stack. Workday and Salesforce are doing zero-copy data partnerships. Inside of three years, half the standalone "drift detection" tools in this space will be features inside platforms or rolled into bigger suites.

If you build a detection tool inside a platform's blast radius, you get absorbed.

The only thing that doesn't get absorbed is the referee. The independent third party. The auditor. The entity whose value depends on not being one of the platforms.

Nobody acquires the referee.

That's the strategic bet. That's why OrgDrift is not a tool. OrgDrift is an independent verification authority, and the work I am doing for the next several years is to make that phrase mean something — first to a small number of operators who feel the pain, then to the auditors who sign off on their work, and eventually to the regulators who decide what good looks like.

That is a ten-year arc.

I am writing this down because I want to remember, when I am six months in and tired, that the work is supposed to take ten years.

A few things to be honest about, because this is a journal and not a deck.

I am building this mostly alone. I write the code, or more accurately, I steer the model that writes most of the code. Some nights that feels like a superpower. Other nights it feels like I am whispering instructions to a very capable stranger who will not remember any of this tomorrow. Both things are true.

I have spent more nights than I want to admit looking at a CSS variable named --ed-paper-0 trying to decide whether a section should be #FAF8F1 or #F8F6EE. I have lost both arguments with myself. That is a real thing I do for a living now — argue with myself about hex codes at 1 AM.

I am not a designer. I am pretending to be a designer by reading about Japanese composition and Apple's product photography and copying restraint from people who actually know what they are doing. The current palette is Apple × Rolex × Japanese, and I cannot defend that lineage with a straight face except to say that it is calmer than what it replaced. Some days that feels like enough. Some days it doesn't, and I close the laptop with a kind of low-grade disappointment that is hard to explain to anybody who isn't doing the same thing.

Here are some things I'm afraid of. Not in any order.

That this takes longer than I think. That the founding-customer ask is too big. That the founding-customer ask is too small. That the moat is real but I haven't articulated it well enough. That somebody else is building this right now, faster, with more money.

That nobody pays for this.

That somebody does pay for it, and I can't deliver.

That the model that writes most of my code today writes a competitor's code tomorrow, faster than I can ship mine.

That I run out of runway before I run out of conviction.

That I look back at this entry in two years and the version of me reading it will be embarrassed for the version of me writing it. (I am going to leave the entry exactly as I write it anyway. The whole point.)

Then there are the days when the work goes well. There was one this week — a small one, nobody else would have noticed it — where a thing I'd been wrestling with for an hour just clicked, the way it does about once a week, and I sat back in the chair and thought, that's it, that's why I do this. On those days I am not afraid of much. I am certain. Not about timing. Not about price. But certain that the gap I am building toward is real, that the SOX Director was right, and that somewhere in the next two years every audit committee in the country starts asking some version of the question she asked me. Somebody is going to have to have an answer.

I would like that somebody to be us.

That's the first entry.

The next one is March 11. That's the day I stopped thinking about it and started typing.

If you are reading this and you are me, six months from now, and the thing is going well — good. Don't get smug. The work is the work.

If you are reading this and you are me, six months from now, and the thing is not going well — also good. Keep going. Most of the things that ever worked almost didn't, and most of the people who didn't make it stopped one Tuesday in a way that, in retrospect, was avoidable.

If you are reading this and you are not me — hi. Thanks for being here. I'm not exactly sure who I'm writing this for yet, but apparently it's you. I hope some of this is useful. I hope none of it is sad.

— K.