You’re sitting in the monthly board prep call that runs fifty minutes. You scheduled the meeting, sent the prep materials, and would normally be taking notes as the conversation unfolds. But now, it looks a little different. The AI transcription tool catches all of it, produces fourteen action items, flags three deadlines, and formats the whole thing before anyone has closed their laptop. All of that is useful. But capturing what was said is only part of what happens in a meeting.
What it can’t tell you is that the head of sales stopped talking after the revenue slide and never spoke up again. You caught that. You were watching and knew that the silence meant something, and that your executive might need to know before the next meeting.
AI can capture the conversation and summarize the decisions. It might even note the silence. But knowing why that silence matters, and what to do about it, is a different kind of intelligence. That distinction is becoming one of the most valuable parts of executive support.
If you’ve spent the past two years reading headlines that put your job on the endangered list, this is the part they leave out.
AI has gotten very good at capturing the details, but it hasn’t caught up to knowing what matters in the room. Experienced executive assistants figured out that distinction early, and they’re the ones being handed more responsibility, not less.
At Boldly, EAs know when and how to use AI tools – and how to pair them with the problem-solving skills and judgement that make great support exceptional. This approach allows our EAs to stay ahead of whatever changes – not keep them stuck on tools that become obsolete or practices that no longer benefit the clients they serve.
Take Jenn, an EA on Boldly’s team. Before AI, she would spend hours writing summaries, prepping notes, and pulling together reports. Now, she can use AI to accelerate drafting and refining the report while using her expertise and knowledge of the client to refine the final product.
“I now have a report that’s complete and in the voice of my client, but I didn’t spend an entire day to write it. Instead, I only might have spent an hour or two.”
That’s where the gold is at!
The Work An AI Tool Should Be Taking Off Your Plate
Most of the work worth handing over to AI shares a pattern. It has one right answer and no relationships attached to it; it’s repeatable and structured.
- Transcripts and recaps from recurring internal meetings
- First-pass research like vendor comparisons, competitor snapshots, or background on someone before an intro call
- Expense categorization and receipt matching
- Turning a rambling voice memo into a structured brief
- First drafts of anything repeatable, like scheduling requests, event logistics, and vendor follow-ups
The point isn’t to make your role smaller. It’s to stop spending your most valuable hours on work that a tool can accelerate.
Those tasks are eating the hours where you should be spending your time doing what you’re best in the world at. If your Thursday afternoons went to reconciling receipts, that’s a missed opportunity for building relationships, thinking ahead, or putting your energy into something else.
AI should give you back time—not take away your opportunity to use it.
The Second Draft Is Where The Job Actually Lives
Ask an AI tool to draft an email to decline a speaking invitation, and you’ll get back something clean, warm, and completely generic. It reads fine—but send it, and you’ve made a mistake.
That’s because the person who sent the invitation introduced your CEO to the company’s largest client four years ago. The right answer isn’t an email at all. It’s fifteen minutes on the phone and a specific alternative date in the fall. AI can’t know that – only you can.
Here’s another example of when human skills matter most. A language learning model is asked to draft a note to a direct report about a missed deadline. It’s programmed to cushion it, soften the ask, and add a compliment at the front. If your executive is blunt by nature, that cushioning is a direct giveaway to anyone who knows him that he didn’t write it, so his EA uses the draft and strips it back to sound like him again.
Nobody prompts their way to good judgement calls. You get there by sitting in the meetings for two years and paying attention to who goes quiet, who never gets asked, and who’s about to be blindsided. That isn’t a soft skill; it’s the job.
Knowing Where To Stop: The Limits Of AI For Executive Assistant Work
Some information should never go into a tool. Your company most likely has a policy in place, but there are certain things that shouldn’t be used with AI. Things like:
- Board materials before an announcement
- Personnel matters
- Compensation or confidential financial information
- A deal that hasn’t closed
- Anything touching an executive’s health or family
Then there’s accuracy. A model will hand you a confident, wrong flight number. A misspelled middle name on a visa application. A statistic with nothing behind it. Verification isn’t a nice extra step anymore; it’s part of the task, because the assistant owns the output either way.
The fastest way to lose an executive’s trust is to send them something you didn’t read. But none of this is new to you. Discretion and execution was always the part of the job that couldn’t be handed off, and AI hasn’t changed that. It’s just raised the stakes.
Boldly EAs work inside client systems rather than their own, and that extends to AI. When a company has approved a specific tool with a specific data policy, that’s the tool, regardless of what an assistant prefers on their own machine. Where no policy exists yet, the move is to ask before pasting, not after.
What Experienced EAs Still Do By Hand
Ask around and most experienced EAs will tell you they’ve kept a few things deliberately manual. Use a model for the first draft often enough and the blank page starts to feel harder than it used to.
Small judgment calls erode the same way. Accept enough auto-suggested replies and scheduling recommendations without really looking, and you stop catching the vendor who was difficult in 2023 and shouldn’t be first in the rotation, or the Friday slot that technically fits but lands ten minutes after a brutal quarterly review.
Sitting with an ambiguous request instead of resolving it with a tool counts as a skill too. Sometimes an ask doesn’t make sense because the executive hasn’t decided something yet, and the useful response is a question rather than an answer.
The EAs who stay valuable are the ones who keep doing things that require a human touch. Not because AI can’t do them, but because judgment, context, and knowing what matters are still human skills.
The Time You Save Comes Back As Better Work
In practice, when an EA starts moving faster, executives rarely ask for fewer hours. They ask for more scope.
Say six hours a week of your time goes to expense reports and travel booking. Hand that to a tool, and what happens next usually isn’t a smaller invoice. It’s your executive asking whether you’d start running the speaking calendar… Then vendor onboarding… Then the first draft of the board deck narrative, before the CFO ever sees it… Your weekly hours probably don’t change, but what you’re entrusted with will.
Job listings have started asking for an AI-powered executive assistant, as though prompt fluency were the qualification that mattered. Prompt fluency is the entry fee. What executives are actually paying for is the judgment applied after the tool finishes running.
What You’re Actually Being Hired For
AI made one category of assistant work inexpensive; anyone can generate a meeting summary now. What that did was raise the value of everything the tool can’t reach: the person who remembers the history, holds the confidences, and will say the uncomfortable thing out loud before it turns into a problem.
That person is you, and the market for you is getting better, not worse.
Ready For The Dream Job You Never Thought Existed?
Boldly hires ridiculously talented remote executive assistants, the ones with years of senior-level experience who already know where the tools stop and the judgment starts.
Our team members are US-based W2 employees with paid time off, paid holidays, and a matched 401K, whether they work full-time or part-time hours.
You’ll support one to three executives long term, usually two to three years each, and you’ll meet them before either side commits. There’s no night or weekend work. We’ve been 100% remote since 2012, so remote isn’t a perk here, it’s how the whole company runs!
If you’ve been reading the AI headlines and wondering what’s left for you, we think there’s more than there used to be.
Ready to learn more? See open executive assistant roles at Boldly!
