I’ve read an unholy number of Chief of Staff job descriptions. This is what happens when you spend two decades in and around the role and then make the catastrophic decision to interview a hundred and fifty people about it. You start to notice things. Mostly you start to notice that the documents meant to define the most powerful coordinating role in a modern organization are, with depressing regularity, god-awful.
Not god-awful like a typo. God-awful like a confession with a bow of terrible on top.
You know the tells. "Wear many hats." (Read: we have not decided what this job is, so we are hoping you will.) "Strategic thought partner." (Read: we want the prestige of the title without naming a single decision you get to make.) "A rockstar who thrives in ambiguity." (Read: the ambiguity is ours, we are simply choosing to make it your problem, and also your personality.)
Nobody says this part out loud at the hiring committee, so I will. "Thrives in ambiguity" is not a candidate requirement. It is a design failure that looks alarmingly like Carmen Sandiego in a trench coat. If the role is genuinely ambiguous, that is not a quality you screen for in the applicant. That’s a decision you failed to make before you opened the Word document to start drafting the damn thing.
The job description is where the lie starts
I have a thesis, and I have been repeating it to anyone who will hold still. Corporate America took the Chief of Staff role, which at its best is a coordinating function modeled on the way the White House actually runs, and quietly sanded it down and hollowed it out into glorified administrative support (wrapped in event planning) with a better LinkedIn headline. The dilution did not happen in one dramatic moment. It happened one vague job description at a time.
Because a job description is not a formality you write at the end. It is the first artifact of how clearly you have thought about the role. And most of them fail in exactly three ways, every time.
They over-broaden the role, so that the person is theoretically responsible for everything, which means they are accountable for nothing. They demand the wrong things, screening for an Ivy League education (and as someone who has one, I can confidently say this is window dressing at its finest), or a stint at "The Big Four" (because nothing says "Chief of Staff" like spending two years owning nothing and generating slide decks), or a degree in finance with a concentration in budgeting, forecasting, and winning a Nobel Prize, while calling it a Chief of Staff. And they lean on words so ambiguous they evaporate on contact with reality. "Drive alignment." "Own the cadence." "Be the connective tissue." Lovely. What does this person decide, on a Tuesday, without checking with you first?
That question, it turns out, is the whole ballgame. And almost no job description can answer it.
I once read a posting (a real one) that asked for a Chief of Staff who would "operate as a strategic thought partner to the CEO while owning day-to-day execution across all functions and maintaining a bias for action in a fast-moving environment." Read that again slowly. It is asking for a COO, an executive assistant, and a meditation app, in one body, for one (low-ball) salary, with no decision rights named anywhere in eleven paragraphs. Some poor candidate took that job. I think about them a lot. I hope they’re okay.
The reason is not that executives are bad writers. Some are, some aren't, but that's not the variable. The reason is that the job description is downstream of the thinking, and the thinking usually hasn't happened. The executive is not designing a role with intention. They're describing a feeling, and the feeling is, "I'm drowning, I need a competent adult to do something, nay, anything, to help, and Todd down the hall has one and says it's been amazing." Which, fair. But "I am drowning" is not a job. You cannot write your way out of a role you have not actually decided to create.
There is an old adage that on your first day in a job, you can throw two things straight in the trash: the job description and the org chart. I think that’s only half right. The org chart has value, though a great chief of staff learns fast that the people holding the real levers are rarely the ones at the top of it. The job description is a different story. That one we do throw out, because no one wrote it with any intention in the first place. And when a job description is built from a pile of vague, meaningless words, you are practically obligated to toss it in the circular file.
So we built the thing I kept complaining about
At some point you have to stop diagnosing the disease in the group chat and go build the cure. This is, not coincidentally, the entire job of a chief of staff. You do not get to lament a recurring problem for months and call it “thought leadership.” You find the persistent friction, you design the fix, and you ship it before anyone asks you to. (I’m a sucker for my own medicine.)
So that's what we did. The CoS JD Architect Tool lives at coulandgold.com/cos-jd, and I want to tell you what it does without telling you how the sausage is made, because the "how" is our special secret sauce, made with love by both Suzi Coul and me.
One honest flag before you dive in: the tool is in Beta. It is live, it works, and we feel genuinely good about where it is. It also means you may hit the occasional hiccup while we keep refining it. If something breaks or behaves strangely, tell us. That is exactly what Beta is for, and your feedback is what makes the next version sharper.
It is not a template. It is not a fill-in-the-blanks Mad Libs that spits out the same four paragraphs with your company name dropped in. I have written enough of those to know they are worse than nothing, because they give you the comforting feeling of having done the work while you skipped the work entirely.
It is a weighted diagnostic. You answer a series of deliberately pointed questions about the role you are actually trying to fill, not the role you would fill in a kinder universe where you had already figured everything out. The questions are designed to be hard to answer vaguely. That is on purpose. Vagueness is the enemy, and the tool is built to make vagueness expensive.
Then, with Claude doing the heavy lifting on the back end, it takes what you gave it and produces a job description that reflects the role you actually designed. Two executives who answer honestly and differently will get genuinely different documents, because they are describing genuinely different jobs. It does not flatten you into a category. It reads what you built and writes that back.
The part I am actually proud of
This is a Coul & Gold Group value we treasure here: it does not flatter you.
Where your thinking was thin, the tool tells you, in plain (and slightly unsparing) language, exactly where it was thin and what to do about it. It does not bury the weakness in the polished document where you will never find it. It hands you the document and then it hands you the truth, separately, so you cannot pretend you did not see it.
And if you genuinely have not done the work, if the role is still just a feeling in a trench coat, it will not produce a glossy job description to help you paper over that. It declines. Politely, but it declines.
And that's really the ethos behind everything we do. We're honest, transparent, and committed to building tools that actually help you. We are not interested in fluffy courses that deliver little, or templates that hand you the feeling of work without the work.
I cannot tell you how satisfying it was to build a tool with the spine to say "no, not yet, go think harder." Most software is desperate to please you. This one would rather be useful.
I have watched executives hit that wall and go quiet for a second, because no one had ever made them say what the person actually decides. Not "owns the strategy." Decides. The specific call, the one they make alone, on a Tuesday, while you are on a plane. The moment you have to name it, the role either snaps into focus or it dissolves in your hands. Both outcomes are a gift. The second one just costs you less when it happens on a screen instead of nine months into a bad hire.
That refusal is not a bug we forgot to fix. It is the entire philosophy. The job description was never the product. The product is the thinking the tool forces you to do before it will write a word. A clear job description is just what clear thinking looks like once it has put on a collared shirt. If you give the tool muddled thinking, the most honest thing it can do is refuse to dress it up.
It also, quietly, strips out the vocabulary I have been yelling about for years. No "force multiplier." No "right hand." Not because those phrases are ugly, although they are, but because they are how you skip the decision. Every meaningless word in a job description is a place where someone chose not to be specific. The tool notices. The tool is annoying like that, on purpose, because I am annoying like that, on purpose.
About the one-per-day thing
Here’s the part where I ask you to be a little patient with us.
You get to generate one job description every twenty-four hours. One. Not ten, not "let me just run it again with a tweak," not spray-and-pray until the output flatters you. One per day.
I want to be straight about why, because I would rather tell you than have you assume we are being precious about it. There are two reasons, and they are both true.
The first is the boring one: every single run is powered by Claude, and that is a real cost that we, a two-woman firm and emphatically not a company with a server farm and a war chest, are choosing to absorb so that the tool can stay free and excellent. The alternative to a daily limit was not "unlimited and free forever." The alternative was a paywall, or a version throttled down into something cheap and useless. We picked the limit. It seemed like the most honest trade, and it keeps the lights on.
The second reason is the one I actually care about. The limit is a forcing function, and it is doing you a favor whether you wanted one or not. When you only get one shot a day, you cannot outsource the thinking to volume. You cannot generate fifteen versions and let the tool decide which role you are hiring for. You have to slow down. You have to answer the hard questions as if they are hard, because they are. You have to treat the role like it deserves an actual decision, because it does.
The constraint is the feature. The scarcity is the point. One good answer beats ten lazy attempts, and the limit makes you give the good answer.
Come yell at us tonight
Speaking of doing the thinking out loud. Tonight is our Ask Us Anything, and Suzi and I will be live, unscripted, and fully prepared to be wrong out loud about the Chief of Staff role, the tool, hiring, or whatever you bring.
Bring your worst job description. Bring the one your last company posted that made you cackle. Bring the question you have been too embarrassed to ask anyone, like "is the role I am hiring for even real, or did I just describe my own to-do list and add a salary." That is a great question. We have answers, and where we do not, we will say so.
Register at coulandgold.com/aua. It is free, it is tonight, and it is the most fun we are legally allowed to have on a Tuesday.
Then, when you have a quiet minute and a role you actually need to fill, go to coulandgold.com/cos-jd and put the tool to work. Answer the questions honestly. Let it tell you where you were thin. Use your one shot like it matters.
And then write back and tell me where it got you wrong, or where it got you uncomfortably right. Both are useful. The uncomfortable one is usually more useful, but we’ll happily hear both.
See you tonight.
Rachael






