Subscribe now

You go to work. You use your expertise to deliver value to your employer or your clients. You get paid a salary or a rate. Your employer or your client captures the margin between what they pay you and what your work is actually worth.

That margin is often substantial. And for most of our careers, we don’t think too hard about it because that’s just how employment works.

But AI is quietly eroding the rationale for some of that margin — and creating a window, right now, for people who are paying attention.

Here’s the shift: the expertise has always been yours. What employers and firms historically provided was the infrastructure — the clients, the systems, the support staff, the overhead. But AI is compressing infrastructure costs dramatically. The tools that used to require a team are increasingly accessible to an individual. The research that used to take hours takes minutes. The deliverables that required coordination and production time can be generated, refined, and packaged by one person with the right knowledge and the right prompts.

This doesn’t mean everyone should quit their job tomorrow. It means the calculus is changing, and women in accounting and finance are in a particularly strong position to take advantage of it — if they’re willing to think like owners instead of only like employees.

What does that look like concretely?

It looks like identifying the highest-value service you deliver and asking: could I offer this directly? It looks like building workflows with AI that let you deliver that service at a margin your employer currently captures. It looks like starting small — a few clients, a few hours a week — and letting the proof of concept build before you make any dramatic moves.

It also looks like asset-building, which is the part of the wealth conversation that gets left out entirely when we talk about women and money.

Income is what you earn. Wealth is what you own. And most of the wealth gap isn’t about income at all — it’s about assets. Property. Investments. Equity. The things that grow while you sleep.

A business, even a small one, is an asset. A content library is an asset. A client base is an asset. An audience is an asset.

AI makes it faster and cheaper to build all of them. A newsletter, a template shop, a consulting practice, a course — these are assets, not just income streams. They’re things you own, that can grow, that you could eventually sell or license or hand off.

For women in finance who have spent careers helping other people grow their assets, this is worth sitting with.

You know how this works. You understand compounding. You understand the difference between something that generates income once and something that generates income repeatedly.

The question is whether you’re willing to apply that thinking to yourself.

The expertise is already there. The tools have never been more accessible. The window is open.

The only thing left is the decision to walk through it.

Women own less wealth than men. Not just earn less — own less. The gender wealth gap is wider than the wage gap, more persistent, and far less discussed. And it doesn’t just affect women who are struggling. It affects women in professional careers, women with graduate degrees, women who have done everything right by every conventional measure.

A woman can spend 20 years in accounting, earn a solid salary, and still retire with significantly less wealth than her male counterpart — because of time out of the workforce, because of the wage gap compounding over decades, because of the invisible tax of being the default caretaker, and because of the simple, stubborn fact that women are less likely to negotiate, invest aggressively, or bet on themselves financially.

AI is not a magic fix for any of that.

But it is something we haven’t had before: a legitimate, low-barrier way to create leverage.

And for women in accounting and finance specifically — women who already have the domain expertise, the analytical instincts, and the credibility — it might be the most significant income opportunity of our careers.

Here’s what I mean by leverage. AI doesn’t just help you work faster. It changes the math of what’s possible at a given level of effort. A fractional CFO who uses AI to automate client reporting can serve more clients at the same quality without burning out. A CPA who builds AI-assisted tax prep workflows can reclaim hours that used to disappear into admin. A financial analyst who learns to build automated dashboards can offer a deliverable that would have taken a team — and price it accordingly.

None of this requires you to become a developer. None of it requires you to quit your job. But it does require you to stop waiting for someone else to hand you the roadmap.

The people quietly building real income with AI right now aren’t the ones with the most technical skills. They’re the ones who understand a specific problem, for a specific person, well enough to solve it efficiently and charge for the outcome. That description fits a lot of women in finance who have been underpricing their expertise for years.

The wealth gap is structural. AI won’t dismantle it overnight. But there’s a version of this moment where women in accounting look back and say: that was when I stopped being good at someone else’s business and started building my own.

I’d rather we be in that version.

There is a story being told about women and AI, and it goes something like this:

Women are behind. Women are hesitant. Women need to be encouraged, onboarded, reassured. Women are the cautious adopters who will come around eventually, once the technology is proven and the path is clear.

It is a condescending story. It is also, for a specific generation of women, almost comically wrong.

I want to talk about the women who were online in 1997.

Not the ones who had it set up for them. The ones who listened to a modem scream for forty-five seconds and waited, with genuine patience, for a webpage to load one horizontal line at a time. The ones who figured out dial-up internet before broadband existed, before Wi-Fi existed, before anyone had made it easy or obvious or intuitive.

Those women are in their late thirties and forties now. They are accountants and analysts and CFOs and controllers. They are the people running the numbers, managing the close, advising the clients, and keeping the financial infrastructure of this country functioning.

And they are being told, with a straight face, that AI might be a little too complicated for them.

I want to offer a different frame.

You Have Been Here Before. Many Times.

The women who grew up with the internet arriving in real time did not have a tutorial for any of it.

They learned AIM by using it. They built MySpace profiles by copying HTML from forums, breaking it, fixing it, and eventually making something that looked exactly the way they wanted. They navigated the first iPhone with no manual. They figured out Facebook, then Twitter, then Instagram, then Snapchat, then TikTok — each one a new interface, a new logic, a new set of unwritten rules that had to be decoded through experimentation.

They did all of this without being called early adopters. Without getting the LinkedIn post. Without anyone noting, in a profile or a performance review, that they had demonstrated exceptional technology adaptability across three decades of digital transformation.

They just did it. Because that’s what you do when something new arrives and you’re curious and you want to participate.

That is the biography of a generation of women in professional services. And it is the exact skill set that the AI moment rewards.

The MySpace HTML Principle

Here is something worth sitting with: the methodology for learning AI tools is identical to the methodology a fifteen-year-old girl used to customize her MySpace page in 2004.

Find something you want to do. Look for how someone else did it. Try it yourself. Break it. Figure out why. Fix it. Try something slightly more ambitious. Repeat.

That’s it. That’s the whole curriculum.

The women who taught themselves HTML from a forum post — not because anyone encouraged them, not because it was a career strategy, but because they wanted their profile to look different from everyone else’s — already know how to learn technology this way. It is in their bones.

What’s different now is that the stakes are higher. Not higher in terms of difficulty — AI tools are, in many ways, more intuitive than early web development. Higher in terms of opportunity. The thing you figure out this time isn’t a prettier MySpace page. It’s a competitive advantage in a profession that is changing faster than at any point in its history.

What Napster Taught Us About AI Ethics

There’s another piece of this generation’s technology biography that doesn’t get enough credit: they already lived through a moment when a transformative new tool arrived and broke all the existing rules simultaneously.

Napster didn’t ask permission. It just showed up and made every existing music industry assumption obsolete overnight. And everyone who used it — which was essentially everyone — had to figure out in real time where they stood. Was this okay? Was it theft? Was the industry’s outrage legitimate or self-serving? What did it mean that something was technically possible but legally and ethically contested?

Those were not easy questions. And a generation of people navigated them without a framework, relying on instinct and values and a gradually developing sense of how new technology intersects with existing systems of ownership and fairness.

That is exactly the conversation happening around AI right now.

The people who have thought the longest about what it means when technology disrupts incumbent industries, when the rules haven’t caught up with the tools, when something feels powerful and slightly uncomfortable at the same time — are not the 24-year-olds who grew up with smartphones. They are the people who were online when the internet was still figuring out what it was.

The Wealth Piece Nobody Is Saying Out Loud

Here is where I want to be direct, because this is ultimately what Carry Forward is about.

Women in this generation — the dial-up generation, the MySpace HTML generation, the figured-it-out-themselves generation — are facing a retirement wealth gap that is real, documented, and not going to fix itself.

They earned less, in most cases, than their male counterparts. They took time out of the workforce for caregiving at higher rates. They were passed over for promotions at higher rates. They negotiated less aggressively, in part because the penalties for negotiating as a woman are documented and real. And all of those individual inequities compounded, year over year, into a wealth gap that is significantly wider than the wage gap that gets all the attention.

AI is not going to fix systemic inequality. I want to be honest about that.

But it is creating an income opportunity that is specifically well-suited to people with deep domain expertise, strong analytical instincts, and three decades of proof that they can figure out a new tool without anyone holding their hand.

That is a description of a lot of women in accounting and finance right now.

The template shop, the productized service, the newsletter, the consulting practice built on an AI-augmented workflow — these are not fantasies. They are things women in this profession are building right now, quietly, in the margins of their existing careers. They are assets, not just income. Things that compound, not just things that pay out once.

The window is open in a way it may not be in five years. The tools are accessible. The expertise is already there.

And the women who taught themselves to build something online before anyone told them they could — they have done this before.

The Throughline

Every platform. Every interface. Every tool that arrived without instructions and required patience and curiosity and a willingness to break things and figure out why.

You navigated all of it. You didn’t wait for the tutorial. You didn’t wait for permission. You didn’t wait for someone to tell you that you were a tech person before you started behaving like one.

AI is the next chapter in a story that started with a modem screech and a blinking cursor.

You already know how it ends.

You figure it out.

It doesn’t have a flashy name. It’s not on your resume under “skills.” But it is, arguably, the thing that makes the whole profession work. The standards call it professional skepticism — a questioning mindset, a habit of considering contradictory evidence, a refusal to accept something as true just because it was presented confidently.

You don’t sign off until you’re satisfied. You don’t take a number at face value. You look for what’s missing, what’s inconsistent, what doesn’t quite add up. You’ve been trained, evaluated, and licensed on your ability to do exactly that.

So when research tells us that women in accounting are more likely than men to have ethical reservations about AI — that we’re slower to adopt, more cautious, more inclined to ask hard questions before we trust a tool — I want to propose that we’ve been reading that finding entirely backwards.

The profession built us to do this. And right now, the profession needs us to do it more than ever.

What the AI moment actually demands

Let’s talk about what’s happening with AI in accounting right now, because it’s moving fast and the stakes are real.

AI hallucinations — convincing, confidently stated information that is simply wrong — are sometimes hard to detect precisely because generative AI presents them in an authoritative tone. This is not a minor edge case. Recent academic research found that approximately 4.5% of new public-company disclosure text in 2024 can be attributed to AI-generated content — and that figure is only growing. These errors are polished and confident, and often live in footnotes, scope exceptions, and judgment calls rather than obvious miscalculations — exactly the places where an accountant’s eye is supposed to catch them. (Journal of Accountancy)

The ICAEW has flagged over-reliance on AI as a particular danger in financial and regulatory reporting, where there are prescriptive rules to be met and no tolerance for errors. A Harvard researcher found that professionals using higher-quality AI actually became less accurate in their assessments — because the tool’s confidence made them less likely to apply their own judgment. (ICAEW)

The AICPA has explicitly urged auditors to maintain professional skepticism and avoid automation bias — the tendency to trust AI outputs without critical assessment. (XBRL)

In other words: the thing the entire profession is being told it needs most right now is the thing women in this field have been building for their entire careers.

The ethics concern, reconsidered

Research from Lean In found that women are 38% more likely than men to have ethical reservations about AI. That statistic gets cited frequently as evidence of a problem — women holding back, women hesitating, women needing to get more comfortable and just start using the tools. (Lean In)

I’ve sat with that framing for a while, and I think it’s wrong. Or at least, it’s incomplete.

Ethical reservations aren’t the same thing as fear. They’re not the same thing as technophobia, or a lack of confidence, or an inability to adapt. Ethical reservations are what you have when you understand enough about something to know what the risks actually are.

Consider what those reservations tend to look like in practice. Questions about data privacy — who has access to the information you’re putting into these tools, and what happens to it. Questions about accuracy — how do you verify that the output is correct when the tool presents everything with equal confidence? Questions about accountability — when an AI-assisted analysis goes wrong, who is responsible? Questions about bias — whose data trained this model, and whose reality does it reflect?

Researchers have pointed out that assigning responsibility for incorrect or unethical AI outcomes — whether to developers, accountants, or deploying organizations — underscores the need to re-evaluate accountability frameworks across the entire accounting profession.

These are not fringe concerns. These are the questions regulators, standard-setters, and firm leaders are actively wrestling with. The AICPA has published a checklist for implementing AI responsibly, covering governance, ethics, documentation, audit readiness, and transparency. The questions women have been asking informally — often dismissed as hesitation — are the questions the profession is now formalizing into standards.

Professional skepticism was always the point

There’s something worth naming about the timing here. For years, women in male-dominated professional spaces have been penalized for asking too many questions. For wanting to understand something before committing to it. For raising concerns that disrupted the momentum of a room full of people who were already convinced.

That dynamic didn’t disappear when AI arrived. If anything, it accelerated. The cultural pressure around AI adoption has an urgency to it — move fast, adopt early, don’t get left behind — that can make careful, questioning engagement feel like the wrong posture. Like you’re falling behind when you should be keeping up.

But AI cannot make judgments that require human experience, ethics, and intuition — and accountants are needed most in situations where decision-making is complex and ambiguous. That’s not a consolation prize for the humans who didn’t get replaced. That’s the actual value proposition of the profession going forward. And it’s a value proposition that requires someone to be the person asking hard questions.

The Journal of Accountancy has written about a new frontier opening up for CPAs as AI system evaluators — noting that professional skepticism, adherence to rigorously developed standards, and the habit of using judgment and questioning mindset are exactly what this emerging role demands.

The women asking the most careful questions about AI aren’t falling behind the profession. They’re walking directly toward where the profession is headed.

What ethical leadership actually looks like here

I want to be precise about what I’m arguing, because I’m not making a case for avoidance. Staying on the sidelines entirely is its own kind of risk — for your career, and for the profession. If the people with the sharpest ethical instincts step back from AI entirely, the people shaping how these tools are built, deployed, and governed will look even less like us than they already do.

What I’m arguing for is engaged skepticism. Use the tools. Learn them. Get good at them. And bring your full professional judgment with you when you do.

That means verifying outputs the way you’d verify any source. The Journal of Accountancy recommends approaching AI responses with the same skepticism you would apply to any tax or audit situation — asking the model to cite sources and then fact-checking against them.

It means staying alert to automation bias — the subtle pull toward trusting a confident-sounding answer because generating it felt like work.

It means asking, every time: who is responsible for this output if it’s wrong? Not to be obstructionist, but because that question is the one that keeps the profession honest.

And it means bringing those instincts into the conversations happening at your firm right now about how AI gets used, what gets checked, and what guardrails are actually in place. Those conversations need people who will ask the uncomfortable questions. That’s a leadership role. It has your name on it.

The longer arc

The accounting profession was built on a premise that I think about a lot: that independent, skeptical, professional judgment is a public good. That someone needs to be the person who looks carefully at the numbers and says I’m not satisfied yet. That trust in financial information depends on people who take that responsibility seriously.

AI doesn’t change that premise. It raises the stakes of it.

The women in this profession who have been quietly asking hard questions about these tools — who felt their caution dismissed as hesitation, their concerns reframed as fear — were doing exactly what the profession trained them to do. They were right to ask. And the profession is, slowly and formally, beginning to catch up to the questions they were already asking.

That’s not falling behind. That’s being ahead.

Stay sharp. Stay current. Carry forward.

— Kelsey

A researcher, when asked about women and AI, said something that stopped me mid-scroll: women are in a position to find gaps with AI because they didn’t build this system.

Read that slowly. Not in spite of. Because.

She wasn’t being consoling. She wasn’t offering a silver lining to soften some harder truth. She was making a structural argument — that the distance between women and the systems being built in Silicon Valley is not purely a liability. That the very experience of being an outsider to a technology, of approaching it without the assumptions baked in by years of building it, of having never been the intended user, creates a particular kind of seeing.

I want to make that argument fully. Because I think we undersell it, and I think it matters.

What insiders can’t see

There’s a body of research, much of it coming out of MIT Sloan and Harvard Business Review, on what academics call “outsider innovation.” The core finding is counterintuitive: in study after study of problem-solving contests and R&D challenges, the winning solutions are more likely to come from people outside the field in question than from the credentialed insiders.

Sociologists have a name for what gives outsiders their edge: focused naivete. It’s a productive ignorance of entrenched assumptions — the ability to approach a problem without already “knowing” what’s possible and what isn’t, what’s worth asking and what’s a dumb question.

Insiders get captured by their expertise. They have too much invested in the current framework to see outside it. They’ve stopped noticing the assumptions because the assumptions became invisible somewhere around year three of building inside them.

Outsiders ask the questions that make insiders faintly uncomfortable. Why does it work this way? Who decided that? What problem was this actually solving, and for whom?

Those questions feel naïve from inside a system. From outside it, they’re the beginning of every useful critique that has ever changed anything.

The system women didn’t build

Now apply this to AI — and be honest about what that system looks like.

Less than 22% of AI professionals globally are women. The training data these models learned from reflects the writing, the priorities, and the perspectives of whoever produced that data at scale — which has never been an even distribution of humanity. The teams that decided what problems were worth solving, what outputs looked “correct,” what a helpful AI response should sound like — those teams were, and still are, predominantly male.

This is not a conspiracy. It’s just compounding. When the same demographic builds the tools, the tools reflect that demographic’s assumptions about what’s normal, what’s useful, what’s worth optimizing for. AI isn’t neutral. It’s a mirror of whoever held it longest.

Which means that when a woman accountant sits down with an AI tool for the first time and thinks something feels off here — she might be right. Not because she doesn’t know how to use it. Because she’s noticing something the builders couldn’t.

When she asks why does this output assume my client is male? or why does this prompt template write in a register I’d never use with a colleague? or why does this tool handle exceptions worse than edge cases that happen to women more often? — those aren’t beginner questions. Those are auditor questions. Those are the questions that happen to produce better products.

Fierce ambivalence as a professional stance

The same researcher offered a phrase for how she thinks women should approach AI: fierce ambivalence.

She was careful to define it. Ambivalence doesn’t mean not caring. It means holding divergent attitudes at once — using the tools, believing in their power to help you, and holding the builders to the highest possible standards for safety, equity, and honest design. Participating and critiquing simultaneously.

This is not a contradiction. It’s actually a precision instrument.

The women I know who are doing the most interesting things with AI are not uncritical adopters. They’re not the ones who believe every tool is revolutionary and every output is trustworthy. They’re also not the ones who’ve opted out entirely because the technology feels compromised.

They’re the ones doing both at once. Using the tools hard, pushing on their edges, noticing when the outputs are subtly wrong in ways that would be invisible to someone for whom “default” settings match their own experience. And then building something better with what they find.

That is the outsider advantage in motion.

The accounting angle

This matters particularly for women in accounting and finance, and I want to name why.

Our profession has a culture of precision. Of trusting the system because the system has auditors and standards and checks. Of being skeptical of output that hasn’t been verified. Of asking show me the work.

These are not obstacles to AI adoption. These are exactly the instincts that make a good AI user.

The accountants who are going to get into trouble with AI are the ones who trust it too completely — who let the tool fill in a number without asking where it came from, who accept a generated narrative without checking it against the underlying data, who mistake fluency for accuracy.

The accountants who will build the best AI-augmented practices are the ones who bring professional skepticism to every output. Who treat an AI draft the way they’d treat a first-year staff return: useful starting point, needs review, I’m not signing off until I’ve looked at the work myself.

That skepticism? Women in this profession have been trained in it for decades. We’ve had to verify our own work more carefully because the cost of being wrong was always higher for us. We’ve had to ask the questions that made us look cautious, or slow, or difficult — because we knew that our errors would be remembered longer.

That’s not a liability. That’s earned quality control. And it turns out to be exactly what you need when working with a technology that is confident, fluent, and occasionally completely wrong.

The question you’re positioned to ask

Here’s the practical challenge I want to leave you with.

The outsider advantage is real — but it only activates if you actually engage. Focused naivete only produces innovation if the outsider sits down with the problem and goes deep. The research on outsider innovation is clear on this: what makes outside thinkers effective isn’t that they stay at a distance. It’s that they get close enough to ask the question nobody from the inside thought to ask — and then they don’t let go of it.

So the question is: what do you notice that the builders missed?

What happens when you run your work through an AI tool and something feels slightly off? What assumptions does it make that don’t match your clients, your practice, your professional reality? Where does it smooth over complexity that you know from experience is not actually smooth?

Those are not failure moments. Those are data points. And they’re yours specifically — because you’ve been working in these systems from a vantage point the builders don’t have.

The women who will matter most in the AI transition are not the ones who waited to be invited in. They’re the ones who walked in, looked around with fresh eyes, and said: I see something you missed. Here’s what it is. Here’s what we do about it.

You didn’t build this system. That means you can see it.

Use that.

Buildher is a newsletter for women navigating the AI transition — with clarity, skill, and no patience for being left behind. If this piece landed for you, forward it to someone who needs the reminder that their outsider perspective is worth something.

If the salary she’s earning is the ceiling. If the title she’s working toward is worth the wait. If the expertise she’s built over a decade of hard, often thankless work is ever going to translate into the financial security she thought it would by now.

This letter is for her, because I think she’s at exactly the right moment — and I don’t want her to miss it.

Here’s what I’d tell her:

The expertise you have right now is worth more than your employer is paying for it. Not because your employer is necessarily acting in bad faith — but because that’s how employment works. The margin between what your work is worth and what you’re paid for it is how businesses function. You knew that. What you may not have fully internalized is that this margin is now increasingly capturable by you, if you decide you want it.

AI has changed the math on starting something. The infrastructure that used to make independent work inaccessible — the overhead, the production costs, the technical barriers — has been compressed dramatically. A woman with your background, your analytical skills, and your domain knowledge can now build and deliver services that would have required a small team five years ago. That is not hype. That is the actual state of the tools right now.

The wealth gap is not going to fix itself. And it is not going to be fixed by the next raise, the next title, or the next performance review. It is a structural problem that requires a structural response — and one of the most powerful structural responses available to you right now is building income that you own, that compounds, and that isn’t entirely contingent on someone else’s decision about your value.

You are not starting from zero. You are starting from ten years of expertise, professional credibility, and domain knowledge that most people building AI-powered businesses would kill for. The people making noise in this space often have the tools and not the knowledge. You have the knowledge. The tools are learnable.

The window is open right now in a way it may not be in five years. AI is early enough that people who figure it out now will have a real head start. The audience for women in finance learning to use AI is not yet crowded. The service models that haven’t been built yet are waiting for someone with exactly your background to build them.

And finally, the thing I most want her to hear:

You were taught to be excellent at someone else’s business. Nobody taught you to build your own. That wasn’t an accident — and it’s not a permanent condition.

The tools exist. The moment is right. The expertise was always yours.

What comes next is up to you.

Subscribe now

There’s a story we’ve been told about AI that I want to push back on.

It goes something like this: AI is a technology skill. To use it well, you need a technical background. You need to think like a developer, understand how the models work, maybe even know how to code. And if you don’t have those things, you’re starting from behind.

I believed a version of that story for longer than I’d like to admit. And I think a lot of women in this profession do too — which is part of why we’re underrepresented in AI adoption, and part of why the tools can feel like they were built for someone else.

But here’s what I’ve come to understand after spending a lot of time learning, testing, and teaching this stuff: the most important skill in AI isn’t technical. It’s communicative. And if you’ve spent a career in accounting or finance, you’ve been building it for years without realizing it.

What prompting actually is

A prompt is just an instruction you give an AI. That’s it. You type something, the AI responds. Simple in concept, but the quality of what you get back depends almost entirely on the quality of what you put in.

Research consistently shows that clarity, context, and specificity are the most predictive factors for high-quality results when working with AI. Not technical knowledge. Not programming experience. The ability to communicate what you need, clearly and completely.

Users who write clearer, more structured, and context-specific prompts experience measurably higher productivity benefits and lower misinterpretation by the AI. In other words, the people getting the best results from these tools aren’t necessarily the most technical people in the room. They’re the clearest communicators.

Think about what a good prompt actually requires:

• Context — giving the AI enough background to understand your situation

• Specificity — telling it exactly what you need, not a vague version of it

• Audience awareness — knowing who the output is for and communicating that

• Iteration — reading the response critically and refining your ask

If you’ve ever written a client memo, briefed a partner, explained a complex reconciliation to a non-accountant, or coached a junior staff member through a process — you have been doing all of these things. You just haven’t been calling it prompt engineering.

The communication advantage women bring

I want to be careful here, because I’m not interested in flattening a whole group of people into a stereotype. Individual variation is real, and not every woman is a strong communicator and not every man is a weak one.

But research does consistently point to patterns worth naming. Studies indicate that women tend to foster collaboration and alignment, focusing on group consensus in their communication — while men more often dominate and compete for airtime. Women also tend to score higher on active listening, reading context, and adjusting communication style for their audience.

Those are exactly the skills that make someone good at prompting.

A good prompt isn’t a command. It’s a conversation. You’re not just issuing an instruction — you’re giving the AI enough context to understand your world, your constraints, and what “good” looks like for your specific situation. The more precisely you can articulate what you need and why, the better the output. The more you can read a response, identify where it missed the mark, and explain the gap — the faster you get to something useful.

Women in accounting have spent their careers being precise under pressure. Translating complexity for different audiences. Asking clarifying questions before signing off on something. Catching the thing that doesn’t quite add up.

That’s not a soft skill. That’s the skill.

What this looks like in practice: the CRAFT framework

I built a framework I use for almost every substantive AI prompt I write. I call it CRAFT:

C — Context. Who are you, what’s the situation, what does the AI need to know to help you well? Don’t make it guess.

R — Range. What are the boundaries? How long, how formal, how detailed? What’s in scope and what isn’t?

A — Action. What exactly do you want it to do? Be specific. “Write a summary” is vague. “Write a three-paragraph executive summary for a non-financial audience” is actionable.

F — Format. How should the output look? Bullet points, prose, a table, a draft email? The AI will default to something — tell it what you actually need.

T — Test. Read the output critically. Does it answer what you asked? Where did it go off track? Refine and try again.

This isn’t a technical process. It’s a communication process. And if you read through those five steps and thought “that sounds like how I already approach a lot of things at work” — you’re right. It does.

The reframe I want you to carry with you

The women I know in this profession are meticulous, articulate, and trained to think carefully before they commit to an answer. They give feedback precisely. They write clearly. They know how to take a complex situation and distill it into something another person can act on.

These are not incidental qualities. They are the foundation of effective AI use.

The most important principle in prompt engineering is clarity — AI models excel at following clear, specific instructions and struggle with ambiguous or incomplete ones. Every accountant reading this has spent years being trained to be clear and specific. That training transfers directly.

So the next time you feel like AI is someone else’s territory — like the tools were built for the developers and the tech bros and the people who were always more comfortable with technology than you — I want you to remember this:

The skill they’re all trying to learn? You’ve been practicing it your whole career.

You’re not behind. You’re already there. You just need the framework to see it.

That’s what we’re building here, together.

— Kelsey

Stay sharp. Stay current. Carry forward.

There’s a number that stopped me cold when I first came across it.

In a Federal Reserve Bank of New York survey, 50% of men reported using generative AI tools in the previous year. For women, that number was 37%. That’s not a rounding error. That’s a gap that has real consequences for careers, income, and opportunity — and it’s happening right now, mostly in silence.

Because here’s the thing: no one is really talking to us about it. The AI conversation online is loud and constant, but it’s largely happening in spaces that weren’t built with women in mind. The content skews technical. The framing skews bro-ish. And a lot of us — smart, capable, analytically sharp women who have plenty to gain from these tools — are on the outside looking in.

I want to change that. But first I think it’s worth understanding exactly what’s going on.

Subscribe now

The data is clear, and it’s bigger than one survey

A Harvard Business School research team, led by Associate Professor Rembrand Koning, examined 18 studies involving more than 140,000 college students and workers across multiple countries. Across most of those studies, the share of women adopting AI tools was 10 to 40 percent smaller than the share of men.

A sweeping analysis from Berkeley Haas and Harvard found that women made up 42% of ChatGPT web users, 31% of Claude users, and just 27% of ChatGPT app downloads. The researchers flagged something worth sitting with: if this disparity persists, it could create a self-reinforcing cycle where AI systems are trained on data skewed toward men, which could widen the gaps further.

In other words, the less women use these tools, the less the tools are shaped by women’s needs. And the worse the tools serve women, the less reason women have to use them. That’s a loop we need to break early.

Lean In’s research adds another dimension: women are 23% less likely than men to receive manager support to use AI, and men are 27% more likely to be praised for using it at work. This isn’t just about personal motivation. There are structural headwinds here — and for women in accounting and finance, who already navigate a profession where the old ways run deep, those headwinds are real.

Here’s where it gets interesting: the ethics factor

When researchers dig into why women are slower to adopt AI, one reason keeps surfacing that I think deserves a much more nuanced conversation than it usually gets.

Women are 38% more likely than men to have ethical reservations about AI — 22% of women compared to 16% of men.

The way this finding usually gets reported, it sounds like a problem. Women are hesitating. Women are holding back. Women need to get over their concerns and just start using the tools.

But I want to offer a different read.

The women in this profession have spent careers being trained to think about accuracy, accountability, and risk. We ask hard questions about data. We notice when something doesn’t add up. We understand that signing your name to something means something. Of course we’re asking hard questions about AI — we were built to ask hard questions about everything that touches our work.

Deloitte’s research found that only 18% of women using generative AI trust that providers will keep their data secure, versus 31% of men. That’s not paranoia. That’s a reasonable response to real uncertainty, and to an industry that hasn’t done a great job of earning trust from the people it’s asking to adopt its products.

The ethical hesitation women bring to AI isn’t a flaw to be fixed. It’s actually the exact disposition we need more of in how these tools get used and governed. The question isn’t how we get women to stop asking those questions — it’s how we help women stay engaged with AI while asking those questions, instead of opting out entirely.

Because opting out isn’t safe either. If women step back from AI because we’re uncomfortable with the ethics of the technology, we lose our seat at the table entirely. And the people shaping how these tools are built, deployed, and regulated will be even less likely to look like us.

What this means for us specifically (why I started this Substack)

For women in data and finance, the stakes are particular. Research on accountants and financial advisors found they had the lowest share of AI adoption among the professions studied, at just 18% — in part because they “handle sensitive information” in their jobs. Our caution is profession-wide and it makes sense. But caution and avoidance are different things.

The good news is that AI is the number one topic women want to learn about — but 63% report a lack of adequate skills and access to training on the job. The desire is there. The access isn’t. That’s a solvable problem, and it’s exactly the gap this newsletter exists to close.

So what do we do?

A few things I believe, having spent a lot of time in this space:

We don’t have to abandon our skepticism to use these tools. Critical thinking about AI outputs is not optional — it’s actually the skill that makes you valuable. The person who can use AI and catch when it’s wrong is worth more than the person who uses AI and takes it at face value.

We don’t have to learn everything at once. Pick one tool, one use case, one part of your workflow. Try it there. See what it does and doesn’t do well. Build from that.

We don’t have to do it alone. That’s the whole reason I built this space. The women in this community are figuring it out together — sharing what works, asking questions without judgment, and staying sharp as the landscape shifts under our feet.

The gap is real. But it’s not permanent. And the women who close it — who bring their rigor, their ethics, and their expertise into the AI era — are going to be the ones who define what it looks like for this profession.

That’s you. That’s us. Let’s keep going.

— Kelsey

Leave a comment