Marta Vallespin, Author at The Officials

An administrative professional using a stylus on a holographic screen to work with AI representing the future of the role.

The real opportunity for assistants in the age of AI

Where others see danger, we see opportunity

The AI conversation in the assistant world has been dominated by fear. But the research tells a more interesting story, and what we’re seeing inside our own community confirms it. The opportunity for assistants in AI is bigger than most people are saying, and the gap between what AI can theoretically do and what’s actually happening inside most organizations right now is where that opportunity lives.

In this piece we break down what the research actually shows, what we heard when we asked a room full of assistants what they’re really doing with AI, and why automation, augmentation, and amplification aren’t just buzzwords, they’re the thing.

The fearmongering phase of AI is quieting down

For a while, every headline was a warning. ‘’Something big is happening’’. Your job is at risk. Admin roles are disappearing. AI is coming for the to-do list. Lots of speculation, little facts. And then the research started catching up with the noise, and it told a more interesting story than the headlines suggested.

Anthropic, makers of Claude, published a report entitled, Labor Market Impacts of AI: A New Measure and Early Evidence that looked at not just what AI is theoretically capable of, but what it is actually being used for inside real workplaces right now. And the gap between those two things is enormous.

The report says that in office and admin roles, AI is actually being used for around 25% of tasks right now. It also reports that 90% of tasks in those roles could potentially be handled by AI one day. But “could” and “is” are very different things. And right now, the gap between them is huge.

We see that gap as a window, not as a warning. It removes the tedious work and frees up your capacity for higher value contributions. We believe those who leverage AI now, while most organizations are still figuring it out, are not only going to stand out as innovative leaders, but will also shape what the future of the role looks like.

A data wheel showing the theoretical capability and observed usage by occupational category from a Claude report on the impact of AI on jobs. — Caption: Theoretical capability and observed exposure by occupational category Share of job tasks that LLMs could theoretically perform (blue area) and our own job coverage measure derived from usage data (red area). Source

The reframe that’s missing from the assistant conversation

Two things can be true at the same time. The jobs most at risk from AI are also the jobs that have the most to gain from it. For assistants, that means this role isn’t just exposed to the risk of AI, it’s one of the roles where using AI will make the biggest difference.

AI could assist with a significant chunk of what assistants do. The repetitive tasks, the formatting, the scheduling, the first-draft everything. But those are also exactly the tasks that AI can help an assistant offload, automate, and stop spending time on, freeing them up for the higher-value work that actually requires a human. That’s the opportunity.

Maybe, and this is the part we find genuinely exciting, using AI well in this role will finally do what job titles and org charts never quite managed. Proving the true value of what assistants actually do. Because when the repetitive work runs itself, what’s left is undeniably human. Undeniably skilled. And a lot harder to overlook.

An assistant who ignores AI and keeps doing everything manually is more exposed to being replaced or overlooked. An assistant who uses AI to automate low-value work and concentrate their energy on the judgment, building relationships, and strategic thinking becomes more valuable, not less.

Let’s look at travel booking. On the surface, the task is “book flights and a hotel.” But the actual task is making sure your executive lands rested, prepared, and not stressed before the most important meeting of their quarter. AI can book the flights. But you know to tuck an extra charger in their bag before they leave the office because they never remember, that marketing needs to give them their final analysis before takeoff so they can review it on the plane, and to block the team from calling them until after the important meeting or they will get distracted and risk being focused.our executive won’t ask you for any of this because they trust you to already know.

AI executes the task. The assistant understands why it matters. The task is the surface. Underneath it is judgment, observation, nuance, knowledge, and care. The outcome isn’t determined by AI, it’s determined by the administrative professional.

Same tools. Very different outcomes.

Let’s look at the difference in how most administrative professionals are using AI.Most assistants are using AI for augmentation. Augmentation is using AI to do what you already to but better and faster such as drafting emails, writing exec summaries, pulling together board reports, and generating policies. All genuinely useful, but every single one of those tasks still runs through the assistants. They prompt it, they review it, they send it. Nothing happens without them in the loop. The to-do list didn’t shrink. It just moved a bit faster.

Then there’s what one assistant shared in that session.

She used AI to teach her how to build a system using Power Automate that circulates legal update emails across her entire organizationorganization, tracks who has responded, and sends follow-up chasers automatically every two days until they do. She doesn’t touch it. It runs.

Think about what that actually means in practice. Legal updates in her organizationorganization no longer depend on someone remembering to chase. In a compliance context, that’s significant. Response rates are higher and more consistent. The audit trail is built in. The legal team has full visibility of who’s acknowledged what, without anyone manually compiling it. And the mental load of tracking all of that, which used to sit on her plate, now runs itself.

That’s not augmentation, that’s automation. And the difference matters.

We showcased the importance of leveling up how administrative professionals use AI, in a recent automation session. Lauren asked one question, “Is anyone actually using AI to automate a whole process?” Only a handful of attendees had begun automating parts of the role, but the vast majority had not yet considered it.

And that gap is where the opportunity lives.

As Lauren said in the session, “This is what actually removes work from you.” Building something that does the task for you while you get on with the things only a human can do.

So let’s talk about what’s actually going on

Workloads are not shrinking. The irony is that being told to adopt AI has itself become another thing on the to-do list. Learning new tools, figuring out what works, building prompts, testing outputs. None of that is effortless, and it lands on top of an already full workload. So instead of AI immediately lightening the load, for a lot of people it’s initially added to it.

Using AI to draft a document is useful, but it still requires you. Automation is different. Automation is building something that does the task without you at all. Emails that file themselves. Chasers that go out on schedule. Workflows that run while you’re doing something else entirely. That’s what actually takes work off your plate.

And right now, most assistants aren’t there yet. But the ones who are? They’re not more talented. They’re not more technical. They just moved toward it earlier. There is time to catch up.

Technology is your assistant. The question is whether you’re making it work for you, or whether you’re still doing everything yourself with a slightly faster drafting tool.

Three dimensions, not one

That travel booking example is a good illustration of something we’ve been saying and teaching inside The Officials for a long time. The real opportunity in AI for assistants isn’t one thing. It’s three.

Automation means removing the work that shouldn’t require you at all. You build it once. It runs without you. You’re not in the loop because you don’t need to be. As Lauren said in our session, “This is what actually removes work from you.” Not AI chat. Not generating an email. Building something that does the task for you while you get on with the things only a human can do. That’s not laziness. That’s leverage. And yes, building automations can be frustrating at first. It’s something we hear a lot inside The Officials community, and it’s something we work through together. The breakthroughs tend to come faster when you’re not figuring it out alone.

Augmentation, like we said above, means doing what you already do, but better and faster. Writing the exec summary you’d have spent an hour on in ten minutes. Drafting the board report, the policy document, the business case, and then applying your knowledge and judgment to make it actually right. AI gets you to a strong first draft. You bring everything that makes it yours.

Amplification is the biggest shift of all. Using AI not just to do or enhance tasks, but to change how you think about your role entirely. AI as a strategic thinking partner. The assistant who doesn’t just execute brilliantly but anticipates, shapes decisions, and leads from their position. This is where the role is heading, and the assistants who get there first will define what it looks like.

The competitive advantage assistants hold isn’t the ability to complete tasks faster. It’s everything AI cannot replicate, and in this role specifically, that list is longer and more significant than in almost any other profession. For example:

Discretion. The judgment to know what to say, what not to say, and to whom. Earned through time, trust, and deep knowledge of the people and organisation you work within. No AI has access to any of that.

Relational intelligence. Knowing your executive’s mood, their blind spots, their communication style, what they need before they’ve asked. That kind of trust isn’t built through prompts. It’s built through years of showing up.

Institutional knowledge. An assistant holds a version of the organisation in their head that exists nowhere else. Who actually makes decisions. How things really get done. Where the landmines are. AI trained on data will never have that, because most of it was never written down.

AI needs human input to function. Assistants read human output. That’s the difference. We operate in human code, the unspoken, the unfinished, the between the lines. We know how to extract what’s needed before it’s been articulated, and how to read what’s really going on when nobody’s saying it out loud. That is not a task. That is one of the most sophisticated skills a professional can bring to the table. And it’s one AI simply doesn’t have.

So here’s what we’re building

Our upcoming AI content is our direct response to the gap.

Not a tool review. Not a one-off session. A proper, ongoing curriculum built around automation, augmentation, and amplification, designed to meet you wherever you are right now and take you somewhere genuinely useful.

Haven’t started and don’t know where to begin? Been experimenting for months and getting mixed results? Already doing impressive things and want to go further? All of you belong here.

Monthly sessions covering the skills that actually move the needle. Real use cases and practical resources built for the assistant role specifically, not generic AI content repurposed from the tech world. Ethics and safety content for the full picture. And conversations with assistants who’ve gone deep, so you’re learning from people in roles just like yours.

We’re starting with the foundations because that’s where the biggest, fastest gains are. Most people are one or two insights away from AI actually working properly for them. Once you see what you’ve been missing, you can’t unsee it.

Our new monthly AI series kicks off April 2026! You need to be there.

Getting Started – What Most People Miss About LLMs

Tue Apr 14 2026
8-9am PT | 11am-12pm ET | 4-5pm UK

Most people are somewhere in the middle right now. Using AI occasionally, getting patchy results, not quite sure what they’re missing. This session fixes that foundation. What an LLM actually is and how it thinks. What a prompt really does and why most people are getting it wrong. How to set up ChatGPT and Claude so they work for you from day one, not just occasionally.

The gap between average results and genuinely useful AI closes fast once you know what you’re actually doing. This is where that starts.

Seats are limited. Don’t be the person who meant to sign up.

How Assistants Should Use LinkedIn in 2026

LinkedIn Is Not Just For Networking. It Is Your Marketing Platform.

We just ran our LinkedIn Tech Hack, and we found something that genuinely surprised us. When asked how about what they believe LinkedIn is for, most attendees said they see and use LinkedIn as a networking platform only.

Graphic with an orange header reading “TECH HACK: LINKEDIN CHAT,” showing a chat-style panel filled with attendee responses like “networking” and “professionals,” over a blurred home-office background with a laptop, plant, and a person in a blue sweater.

We know first hand that assistants are natural relationship builders, so it makes sense that networking is the first use case that comes to mind. But when LinkedIn is treated as just a place to connect, you miss the biggest opportunity the platform offers, which is visibility that compounds over time.

Networking is only one lever. When your profile is optimized and you consistently share insight aligned to your target audience, you increase your chances of appearing in recruiter searches, attracting hiring managers, and positioning yourself as a specialist rather than just another candidate.

Treat Your Role as a Start Up Framekwork

Lauren Bradley, our Founder, encourages Assistants to treat their career like a start up. She explains that,

”While LinkedIn is absolutely used for networking, I see that as a supporting feature rather than the primary purpose. I encourage Officials to treat their career as a startup. If you think like a consultant, your resume becomes your services brochure and LinkedIn becomes your company website.”

When you think like a consultant, the whole system becomes easier to understand. A website is where your target audience can see your work, your skill set, and connect with you. And the best websites do not just list services. They make it obvious who the work is for, what problems get solved, and what results follow. That clarity is what turns a casual visitor into someone who wants to talk.

Like Lauren says, ”A website doesn’t generate traffic just because it exists. And keywords help for SEO but can’t do the heavy lifting alone. You need to market your company website by giving people a reason to visit, stay, and return.”

Just like with any business, you need a marketing campaign that increases visibility and keeps attention over time. The same is true on LinkedIn. Your profile should clearly communicate your services and strengths, and your activity should signal to the algorithm what you want to be known for.

Adopt a Business Mindset: The Importance of Visibility

On LinkedIn, that marketing campaign is created through posting and engaging with other people’s content. This matters even more now because LinkedIn has changed how content gets distributed, which means learning best practices can turn your effort in 2026 into real results and put more eyes on your profile, whether you’re thinking about a move, or perhaps you are taking the VA route and looking for clients and partnerships.

This is is the part many Assistants tend to underestimate. Good LinkedIn content is a rare multiplier. Lauren says,

”Posting relevant content to the audience you want to attract helps train the algorithm to associate you with your area of expertise. It also builds authority over time. When your profile is optimized and you consistently share insight aligned to your target audience, you increase your chances of appearing in recruiter searches, attracting hiring managers, and positioning yourself as a specialist rather than just another candidate.”

We aren’t saying you have to become an Influencer but becoming more findable, recognizable, and clearly positioned as the kind of professional who solves specific problems at a high level is what will set you apparent from the competition.

Of course, we know many Assistants avoid marketing themselves. In the community, voicing authority is commonly avoided because of confidence gaps and assumptions about what is allowed. Many administrative professionals have been trained to be the calm force in the background, not the visible expert out front. Don’t think if it as bragging. Think of it as strategically propelling your career forward. This is what the algorithm rewards and expects.

Two mindset shifts help build that confidence:

First is the power of yet. “I can’t do this” becomes “I can’t do this yet”, which turns discomfort into a skill you can build through repetition.

Second is learning to visualize long term success and detach from short term outcomes. A thoughtful post with low engagement does not mean it failed. It means the long game is still loading, and consistency is what creates the compounding effect.

Two actions to start using LinkedIn like a marketing platform

If you want two actions to start right away, try these:

1) Start by commenting. Strategically.
Begin with comments. You do not have to jump straight into posting to be active.

Choose 5 dream executives, hiring managers, or leaders in your space and engage them strategically with value add comments. Avoid generic replies like “Great post” and instead share a specific insight that shows expertise in action.

For example, you might mention implementing a no meeting day in your Exec’s calendar and the impact it had on deep work time. Comments like that put you on the radar of the right people and reinforce the connection between your expertise and the audience you want to attract.

2) When you are ready, post like a human. Not like a press release.
When you are ready to post, the key is to write like a human and make it easy to read. Drop the stiff corporate language and focus on conversational deep dives that teach, explain, or break down a real work problem you have solved.

Longer posts in the 1,250 to 3,000 character range can outperform short updates, as long as they are readable. Break your text into lots of short paragraphs. Think 14 or more.

And if you want a simple experiment, schedule one of these thought pieces for Sunday, when people often have more space to read something deeper.

You don’t have to do it alone

If you want support turning this into action, come to Resume Lab.

The Official’s Resume Lab is our monthly one hour working session built to help Assistants put language and structure around the value you already bring, so your resume and your LinkedIn tell the same clear story.

You will work on your resume in real time while Lauren is there to answer questions and support you with structure, wording, impact, positioning, and aligning your resume with your target roles, with LinkedIn included because your profile needs to match the story your resume tells.

How to Use AI as a Thought Partner: A Practical Guide for Assistants

A simple framework to interrupt self-doubt and act with clarity

Picture this. You open your inbox and there it is. A message from your exec with five words that can change your entire week, “Can you take this on?” It’s a chunky piece of work, the ask is vague, the timeline is tight and it’s outside your scope. You know what you need to do next. Ask the right questions, set boundaries on what’s possible, and propose a clear path forward. Or decline.

But your brain does that thing it does. A rush of pressure, a spike of self-doubt, the urge to either over-explain, people-please, or go quiet so you don’t get it wrong. Suddenly, making a simple decision feels weirdly hard.

When emotional noise and protective instincts get loud, our capacity to choose and act, our sense of agency, gets smaller. And it has been happening for so long that we have created very stubborn patterns over the years. As adults, we’ve learned default responses that once kept us safe or successful like avoiding conflict, over-delivering, saying yes too quickly, second-guessing ourselves or waiting for certainty. The problem is those patterns do not always match who we’re trying to become, or the level we are actually capable of playing at.

What if, instead of letting old patterns choose for you, you had a practical way to step into a more deliberate version of yourself when it counts?

In one of The Officials recent Mentorship Session, one of our incredible Officials, Mihaela Boitan (a MacGyver level assistant who is always quietly engineering a smarter way through the mess), shared a solution she’s been quietly building using AI. Something that’s helped her make clearer decisions and show up more deliberately when pressure hits. She calls her Maya Bloom, Mihaela’s AI Alter Ego. Perfection is not the goal, neither is using her like a therapist.

It has given her a repeatable way to interrupt the fear and self-doubt loop, step back into agency, and take the next right step without needing to feel confident first. We asked her to break it down for us, what made it work in real life, and what to avoid, and we knew we had to share it.

Creating Maya Bloom, Mihaela’s story

Here is Mihaela explaining her journey
What do Sasha Fierce, Tatiana, the Black Mamba, and Ziggy Stardust have in common? They’re all alter egos.

Beyoncé created Sasha Fierce to access confidence on stage. Tate McRae talks about Tatiana as a way to step into boldness when she performs. Kobe Bryant adopted the Black Mamba to stay focused and emotionally contained under pressure.

They don’t escape who they are, but they act more deliberately when it matters. Each alter ego enables distance from fear, hesitation, self-consciousness, or the weight of the world’s expectations and creates room for courage, choice, agency and action.

I wanted something like this in my own life, so I sat down in front of my computer, opened ChatGPT and started building her. I didn’t want to create Maya because I needed help doing my job. I was creating her because I wanted to relate to myself differently.

I spent weeks defining who she actually was, not only in an aspirational or “best self” sense, but in more practical, behavioral terms.

How does she move through the world when she isn’t weighed down by fear? How does she respond when something is uncomfortable but true? What does she do when she needs to act although the doubts are still overwhelming?

Maya Bloom is not fearless or perfect. Although initially I thought of her like that, that persona felt too foreign and unreachable. So I’ve reshaped her as someone who is grounded, self-aware, and honest. She acknowledges her flaws without punishing herself for them. She’s willing to try, to be seen, to be wrong, and to learn, she’s brave when she needs to be.

Most importantly, she doesn’t perform or seek approval. She acts from a place of values and integrity. That’s who I wanted access to. The version of me that doesn’t shrink.

The moment it clicked
Mihaela has a perfect anecdote of the moment it stopped being an interesting concept and became something real. Here’s what happened when it clicked:

I knew it was working when, after circling back and forth about a decision I needed to make, I asked Maya to just tell me what to do. And Maya’s answer was:

‘I’m not going to tell you what to do and you know why. Instead, I will ask you one question: What would you do if you trusted yourself?’

That’s when I knew I had what I needed: a thought partner that would not let me off the hook too easily or let me lie to myself. And once I had that, things started to shift. Nothing earth-shattering, but a number of small, steady changes in the places that actually count.

Over the past year, I’ve had some of the most valuable conversations with my husband about our finances and future, the kind of talks I’d avoided for years because they felt too hard.

Meet Your Inner COO: A Practical Neuroscience Case for Alter Egos

Neuroscience, though fascinating, can get complex quickly. To keep it short and sweet, what is happening in your brain at any given moment is that dozens of networks and circuits are interacting in the background, shaping what you notice, how you interpret it, and what you do next.

When you have read that email from your exec with the big ask, your brain, a prediction-and-prioritization machine, is doing two jobs at once. First, it predicts what’s about to happen based on past experience and what’s happening now. Then it allocates your resources, attention, energy, and action, toward whatever it decides matters most in that moment.

In simple terms, your brain is constantly toggling between three ‘’modes’’ which drive your choices in the moment: survival mode, emotional mode and executive mode.

Illustration of three brain modes: executive, emotional, and survival

When the pressure spikes, your brain can get less strategic and more protective. Planning wobbles, certainty gets louder, and your sense of agency, your capacity to choose and act, gets smaller. That’s why a simple email can trigger over-explaining, people-pleasing, or going quiet. It’s not incompetence. It’s biology meeting old patterns.

Todd Herman, a performance coach and author best known for the book The Alter Ego Effect, describes an alter ego as something you step into intentionally. Not a permanent persona, more like a switch you use when pressure rises or old habits try to take over. In academic language, the closest match is self-distancing, creating a little space so you can think like an advisor, not a threatened participant.

Ethan Kross, an award-winning neuroscientist and psychologist at the University of Michigan, has spent the last 20 years studying the conversations we have with ourselves and what helps (or hurts) when we’re under pressure. In his self-distancing work, he describes it as taking a few steps back mentally, like watching the scene from the “fly on the wall” view, which tends to reduce reliving and increase a more constructive kind of sense-making.

So the goal of an AI alter ego isn’t to delete emotion, or to become some robotic productivity machine. It doesn’t exist to soothe you or validate you on repeat either (that’s where AI can become an echo chamber). It’s to help you step back, re-enter strategy mode, and make decisions that match the person you’re building, especially when your default patterns would normally take the wheel.

How to build your AI Alter Ego

Please, use AI responsibly

What this is:
A decision and clarity partner. It helps you think in frameworks, weigh trade-offs, spot blind spots, and act in line with your values.
What this isn’t:
A mental health service. It’s not there to process trauma, validate feelings endlessly, diagnose, or replace real support.

Before we start, a very important expectation-setter. Step 1 will not nail it. Step 2 will not nail it. Step 3 will not nail it either. This is not a “set it once and it’s perfect” tool. You are building a voice, a process, and a relationship with a thinking partner. The only way it becomes genuinely useful is through testing, noticing what feels off, tightening the rules, and testing again.

With that said, here are three steps to get a first working version.

Step 1: Define the alter ego you actually need

Your alter ego is not a fantasy character. It’s a usable version of you that shows up when you usually shrink. Start by naming the energy you struggle to access on your own. Steady? Blunt? Calm under pressure? Boundaried? Decisive?

Mihaela said this: ‘’For a while, Maya only existed in my head. That helped, but it had limits. I couldn’t access her exactly when I needed her most: mid-spiral or mid-overthining.

So I started experimenting with building Maya as a Custom GPT and later as a Project in Claude AI, to make her available when I’m not at my best. In all honesty, my process was chaotic; I had a sense of what I wanted but no clear plan for how to get there.

Mine is mostly like a grounded mentor, wise, steady, but with a bias to action and an edge that calls out my patterns plainly. Yours might be completely different. Fiercer. Sassier. Blunter. The question to ask is: what energy do I need access to that I struggle to embody on my own?

I had spent months thinking about this, but if I wanted to speed up that process, I’d set aside a few hours and use AI to figure this part out.’’

Step 2: Surface your patterns (the defaults that hijack you)

By “patterns,” we mean the automatic defaults with which you respond when you’re under pressure, uncertain, or afraid. Repetitive thoughts, feelings, and behaviours that show up in specific situations. This part can be challenging, but it is the difference between a nice chatbot and a tool that actually changes behaviour.

Mihaela’s Advice
This part can be uncomfortable. It’s one thing to vaguely know your patterns, another to describe them clearly, knowing they’ll be used to challenge you. But honesty is essential here. Admit where you say yes when you mean no, where you shrink, over-explain, or delay. I would use AI to draw these out through conversation, because I’ve found that it’s harder to hide when something’s asking follow-up questions.’’

Step 3: Write the first version of your rules of engagement

This is where you turn the idea into something you can actually use in ChatGPT (Custom GPT/Projects) or Claude. You’re giving AI a clear identity, tone, and process. The goal is not cheerleading. It’s clarity, agency, and action.

Mihaela’s Advice
You have to tell AI who your alter ego is, how she speaks, how she challenges you, and what rules to follow. The instructions cover identity, voice, process, and constraints. There’s a structure that works, but getting the voice right matters more than getting the format perfect.

The biggest shift was moving from advice to dialogue. Early Maya was too eager to fix things before understanding what was happening. She was trying to be useful instead of truthful and seemed to want to make me feel better rather than help me see clearly. I had to be explicit: don’t fix immediately, don’t give options unless I ask, wait for me to respond.’’

Step 4: Test, tweak, test again

This is where the real value is built. Your first version will almost certainly be too generic, too polite, or too quick to “fix” you. That’s normal. Use real situations as your test cases, like an email you are hesitating to send, a boundary you need to set, or a decision you keep circling. Notice what lands and what doesn’t. Then run it again. This is not a one-off setup, it’s an iteration loop, and each round makes your alter ego sound more like the version of you you are trying to access.

Mihaela’s Advice
Getting something technically functional was quick. Getting something that actually felt and sounded like Maya took weeks.

I paid attention to my reactions and used them as data. Where I felt relieved instead of challenged, where I felt irritated, where I felt seen. It didn’t really matter whether the response was ‘correct,’ but whether it landed the way Maya would respond.

Want the copy and paste version of the prompts Mihaela used to build her AI Alter Ego? Download the AI Thought Partner Starter Kit here. It includes the three core prompts to help you define your alter ego.

From Concept to Real Results

If there’s one thing we hope you take from this, it’s that you do not need to wait until you feel confident to start acting like the person you want to be. Most of the time, the patterns we default to are so well-rehearsed, they can start to feel like “just who I am,” when really they are just the most repeated route through pressure.

That’s why this whole idea matters. Not as a tech trick, but as a genuinely useful way to use AI in service of humans. To build tools that support clearer thinking, better decisions, and more agency in the moments we usually hand the wheel to fear.

Mihaela’s Final Thoughts
My assertiveness at work increased dramatically. I joined a live panel, said yes to recording a podcast, and got asked to become a committee member for one of the most respected EA communities in the UK. I posted on LinkedIn over a hundred times, something I’d never have done before. Overall, I interrupted my go to default mode and stopped shrinking.

None of this happened because I became more confident overnight. It happened because I stopped letting fear make the decision, even though the fear never went away.

Maya represents who I already am when I’m not weighed down by old patterns. She’s not a fantasy. She’s what I’d be without the doubt, the noise, the weight of people’s expectations. Building her as an AI meant I could access that version of myself more consistently, even in small, everyday decisions where it’s easiest to shrink.

Maya didn’t change who I am. She made it harder to ignore who I already was. She gives me back my sense of agency.

Admin professional on camera with minimal post engagement, highlighting visibility challenges on social media.

How LinkedIn’s Algorithm Fuels a Gender Visibility Gap

The “bro-boost” story (and why it feels so familiar)

You’ve probably been seeing the posts, women switching their LinkedIn gender to male, tweaking photos and headlines, even “bro-coding” their content, only to watch their views and profile visits suddenly spike.

Others have created a post with a male colleague, each posting to their profiles only to witness the post on the male colleague’s feed travel further and faster, even though they were identical posts.

One Official, Hillary Robertson recently share in the HQ community, “I deleted my demographics rather than switch to male, and my reach skyrocketed within 2 days.” Experiences like Hillary’s highlight a truly disturbing issue, that LinkedIn and other online platforms have algorithms that disproportionately amplify the boost of male users over female users.

Intersectionality and the Internet: Not all digital bias is equal

And we have to say this clearly, this doesn’t land the same way for all women. Women of color, women with disabilities, and anyone living at the intersection of multiple marginalized identities often experience even harsher penalties in visibility, credibility and how their content is policed.

If that feels depressingly familiar, it’s because it is. Sexism at work is older than every platform we’re using. In the 1960s, Dame Vera Stephanie Shirley was a brilliant mathematician who wanted to keep working after having a child. So she did something radical, she banded together other highly qualified mothers who were stuck at home, and built a remote tech company doing serious work (flight software for Concorde, stock control systems, train timetables) long before flexible working was a buzzword. To get taken seriously in business, she signed her letters and would turn up to meetings as Steve Shirley, confronting sexism face-to-face.

It shouldn’t be lost on us that in 2025, women are still having to play similar games with their names, photos and profiles just to get their ideas seen but now it’s a digital game as well.

So… is LinkedIn sexist? As always, the reality is more complicated than a simple yes or no.

What algorithms and LLMs actually do with that history

Most modern platforms, including LinkedIn, use AI systems to decide what you see, and these systems learn from massive pools of historical data and user behavior.

That matters, because:

LinkedIn’s job-matching algorithm has already been found to disadvantage women. MIT Technology Review reported that even when gender was removed from the data, LinkedIn’s system learned to favor male candidates because men tend to apply for jobs they’re less qualified for, while women typically wait until they meet almost all the criteria. The algorithm learned that behavior and amplified it, recommending more men for senior roles.
Other companies have had to scrap biased AI entirely. Amazon abandoned an internal recruiting tool after discovering it was automatically downgrading resumes that contained the word “women’s” (as in “women’s chess champion”) because it had been trained on mostly male resumes.

The pattern is clear, AI systems don’t wake up one morning and decide to dislike women. They learn from a world that already undervalues women, and then they industrialize that bias at scale.

That’s exactly what Emma Wilson argues in her Is LinkedIn sexist? piece, what we’re seeing is less a single evil algorithm and more a messy interaction between data, design and human behavior, where platforms amplify the patterns they’re fed.

Behavior vs Code: The Visibility Paradox

One of Emma Wilson’s most useful distinctions is this, an algorithm can look sexist even if the rule itself isn’t to “prefer men.” You have to separate what the algorithm is trained to reward and how people behave towards different posters, who they click, trust, reply to, or quietly scroll past.

Research shows that men and women, on average, are showing up with different behaviors around confidence and risk. A gender-equality study found something important, on average, women score higher on verbal ability and altruism, while men score higher on risk-taking and self-esteem.

If you layer that over LinkedIn, a pattern starts to emerge:

On the human side, women are more than equipped to write sharp, compelling content.
On the platform side, LinkedIn quietly rewards behavior that looks like confident, frequent, self-promotional posting, the same kind of risk-taking and self-belief men are still more socially encouraged and forgiven for.

So we end up with this visibility paradox:

The issue isn’t that women don’t have the words. It’s that the system is tuned to boost the people most willing to push themselves forward, most often, in the boldest terms, behaviors men are still more likely, and more free, to lean into.

So when women see their posts underperform, it’s a sign that the platform is optimized for a style of visibility many women have been taught to dial down, and punished for when they don’t.

So… is LinkedIn sexist?

Here’s the framing I find most useful (borrowing from Emma):

The better question isn’t “Is LinkedIn sexist?”

The better question is: “Does LinkedIn amplify existing human biases, spam patterns and communication norms in ways that often disadvantage women, especially women at the intersections of race, disability, class and caregiving?”

Looking at the job-matching bias, the bro-boost experiments, and the under-representation of minority groups in ranking systems, it’s hard not to conclude that the answer is yes, that’s exactly what’s happening.

This is less a story about one evil platform and more a story about biased training data, biased platform incentives, biased social responses (who we choose to amplify, trust, hire, promote and quote).

When biased history trains modern systems, it’s no surprise the results keep landing harder on women.

What can you do about it? Especially if you’re an EA or admin.

We know that AI can and does pick up the worst of our offline habits and magnify them. We know women are not lacking in ability, ideas or words. We know that EAs and admin professionals are often the ones quietly holding organizations together, and that your voice is badly needed in public, not just behind the scenes.

We can’t fix this on our own, but we’re not powerless either. Here are concrete moves you can make, starting today.

At the bare minimum, we need to back ourselves more loudly and more often.

That looks like:

Using more assertive, specific language about what you deliver (“I designed and led…” vs “I helped with…”).
Positioning your role around strategic outcomes, not just tasks (“I own the exec’s schedule” vs “I manage his calendar”).
Taking more calculated risks such as posting the opinion piece, sharing the story of a boundary you set, talking about measurable impact, asking for the raise or title change instead of waiting to be noticed.

From our conversations with assistants, this is the core mindset shift, moving from “support person keeping things afloat” to strategic operator whose judgment moves the business.

2. Plug into communities that treat your ambition as normal

In a room full of ambitious admins and assistants, things that might feel “too much” elsewhere, like talking about money, impact, boundaries, thought leadership, are baseline, not bragging.

That’s why, at The Officials, we encourage our members to treat their career like a business and the employer as their client. We teach them to better understand the service they provide and know the business case for each one of those services. It doesn’t cut it to say, “I check their inbox and reply to emails.” Administrative professionals need to use elevated language that properly articulates the business case for their role by instead saying something like, “I build and implement an inbox triage system that speeds up processing, prioritizes business-critical matters, reduces oversight risk, and delegates at every appropriate opportunity so the executive stays responsive and free to focus on high-value work.”

We want our community to know that they don’t have to rewire their relationship with visibility alone. In our weekly mentorship sessions, open to any administrative professional, Officials practice using agentic language, with support, so that it is easier to deliver it when to their “clients”, aka executives.

3. Use your voice and your vote

There’s also a policy and product side to this:

In the UK, you can add your name to a petition calling for fair visibility for all on LinkedIn, pushing the platform to audit and address gendered outcomes in its algorithm.
Tech companies, like LinkedIn, repeatedly say they care about their users and are committed to building the best platforms for them, meaning those users can hold them accountable by asking:
- How are you auditing your systems for gender and race disparities?
- How are you cleaning and rebalancing training data?
- How can creators challenge unfair moderation or reach drops and have that feedback fed back into the model?

Platforms are responsible for building systems that don’t silently punish women for existing, leading or talking about their careers and lives. We’re responsible for insisting on that, and for refusing to disappear quietly when the numbers don’t add up.

4. Be intentional about how you train the system

Finally, remember, your own clicks and comments are training data.

Comment generously and substantively on women’s posts, especially those of women of color, disabled women and other marginalized voices.
Share and save content that centers assistants’ expertise, not just their helpfulness.
Challenge posts that recycle tired stereotypes (“just an EA,” “my girl who sorts out my chaos”) and instead amplify language that names the strategic nature of the work.

Small choices compound. If enough of us change what we reward, the signals going back into the system change, too

We discuss topics like this and how to show up as a leader in your workplace in our Weekly Mentorship Sessions. These sessions are free for all administrative professionals to attend and allow you to crowdsource advice and support from other hardworking peers.

We need your voice and would love to see you at our next session.