1. Opening Scaling Tension
Every owner who has added AI to the business eventually asks whether the support roles around them still make sense. The fear is loud, but Adrienne Green keeps running into something more useful in her own companies: the tools are capable, and they also break. The connectors that link an AI agent to the systems a team works in need to be redone almost every week. A tool reports that it did not load, and the fix is to try again.
That is a small problem until it lands on one person. Green’s point is that when no one else on the team is accountable for the technology, the owner becomes the tech support. Growth adds systems, connections, and points of failure, and each of them routes back to the founder. That is not leverage. It is a new form of operational drag.
2. The Hidden Constraint
The constraint is not what AI can do. It is who owns the work when the tool stops doing it. Green is specific about where agents are strong: routine, recurring tasks inside systems with a reliable connector or API, compiling data, running on a schedule without forgetting or getting distracted, producing first drafts, and creating tasks for a person when a piece needs one. She is equally specific about where they are weak: tools with no native connector or an unreliable one, work that sends the agent to browse the web, which is slower and uses far more credits, and judgment calls that depend on context in the company or in a person’s life.
She also points to something experienced operators take for granted. Turning a tool off and on again is basic problem-solving, and an agent does not reliably do it yet. When the constraint is ownership and judgment, the scarce resource is leadership bandwidth, and it gets spent by whoever has to catch what the agent misses.
3. The Operating Shift
Green’s operating principle is to give the agent the routine system work and keep a named human on every step that needs judgment or depends on an unreliable connection.
In her business, routine tasks live in one task management system. The agent runs every morning, finds the tasks assigned to it that are due that day, and executes its part. If a task needs a person for the next step, the agent posts an update that tags them or creates a new task with everything they need. The person then reviews, decides, and executes. Where the connection is reliable, the person can simply approve and let the agent carry out the action. The agent compiles. The human decides.
4. Execution in Practice
Four insights from the conversation turn the principle into decision-making frameworks and execution systems.
Audit by role, not by tool. Green’s team has always kept a roles and responsibilities list as a board, with every recurring responsibility in each department listed by frequency, from as needed to annual. That made it simple to go through finance, HR, client care, and recruiting and ask which parts of each task an agent could front-load, so a person can take the work from there.
Keep a human where judgment or the connection is missing. Her example is the birthday and work anniversary graphics for her virtual team. At the end of each month the agent checks the team board, builds the graphics from templates using each person’s photo and details, and creates a task for the HR head on each date. The agent does 80 to 90 percent of the work on this task. A person still posts it, for two reasons. The message needs discretion, such as a day when a storm has hit where the team lives and celebrating would be tone-deaf. And there is no native connector between the agent and the chat tool, so a third-party workaround would add one more thing that can break.
Assign accountability to a named person. If the connector to the design tool fails, the HR head owns the fix, and the agent tags them. It does not fall to the founder. Connectors do disconnect, and the question that matters is who is responsible for each one.
Run every task through four questions. Is it routine and recurring? Does it live in a tool with a reliable connector or API? Does it need judgment, tone, or a relationship? Is the connection to the agent missing or unreliable? The first two send work to the agent. The last two keep a person in place. Green also suggests loading your procedures into an AI tool and having it audit what can go where.
5. Leverage Outcome
The result is not fewer hours spent on the same list of tasks. It is a different use of the team’s attention. People start at the decision rather than at data gathering, which leaves more time for evaluation and analysis. Green’s HR lead has more time for the people on the team because routine administration shrinks, and the same applies to the routine parts of payroll.
The staffing implication is plain. A business may need fewer assistants for the same amount of work, or it can cover more with the same team. The assistants become more valuable because their core skill shifts from knowing every platform to directing an agent, reviewing what it produces, and stepping in when something is flagged. A support person no longer has to know a specific tool inside out if the agent can work within it.
For the founder, leverage is capacity: the agent absorbs routine work, the team absorbs judgment and accountability, and leadership bandwidth stops going to keeping tools alive.
Connect With Adrienne Green
To learn more about Adrienne Green and her work:
Website: workergenix.com/
LinkedIn: linkedin.com/in/arealgreen
The Immediate Move
The constraint in an AI-enabled business is ownership, not capability. Structure matters more than effort here. Name the person accountable for every connection and every judgment call, and let the agent do the rest.
Start with one task. Pick recurring work you still touch, ask whether it is routine and connected, and decide which step needs a person. Each handoff done this way transfers ownership, reduces cognitive load, and protects the time you need for decisions only you can make.
Watch this before you hire your next support role.
Book a discovery call to see how the right executive support helps you scale with clarity, alignment, and control without burnout or chaos. Click here to subscribe.
Like what you read? Get weekly insights on scaling, efficiency, and profitability—straight to your inbox. Click here to subscribe.
Full Podcast Transcript
Does AI mean the death of the virtual assistant industry? That is something that has been debated with the virtual assistants, with people like me who own virtual assistant businesses, and with the entrepreneurs and business owners that we help who are using virtual assistants. In this brave new world of AI and AI agents and more and more autonomous higher-end functionality, is there a place for virtual assistants? And if so, where is it?
Now, hey guys, welcome to another episode of Scale Smart Grow Fast. I’m Adrienne Green, and today I want to talk about a conversation that I’ve been having a lot and something I’ve been figuring out from my own business, which is with AI agents with everything they can do these days, is there a place for a virtual assistant within a business? And if so, what is it? Now, spoiler alert, guys, I’m gonna tell you what I have seen in my own business, which is
AI agents are allowing my virtual team to do more and to take more off my plate because it’s enabling them to do the higher level work and it’s doing a lot of the basic time-consuming grunt work. So to for me and in my business, both in worker genics, my virtual assistant business, as well as in my property management for my properties and in Investway, our private money lending business, we have found that the
Best answer is AI agents and virtual assistants working together. The and that’s that’s what this whole episode is going to be about. So if you’re interested in leveraging AI agents, if you have a virtual assistant or not, you’re thinking about one and you’re wondering, do I still need one with AI agents? Can AI agents do everything? As somebody who’s a business owner myself and who’s in this same place of wanting to be efficient, wanting to use my time well, wanting to use my money well, here’s what I’m finding.
and how I’m doing it and I hope it helps you. So
First.
First, guys, if you don’t know me, hey, hi, I’m Adrian and I’ve been using virtual assistants since 2021. I started as a real estate agent, brought one executive assistant who was virtual onto my business, and they handled so much that instead of closing like 12 deals a year, I closed like 50 something in six months. My business blew up. I started a team that I later sold. And
It all started with this virtual executive assistant. And that’s why I’m so passionate about helping people with virtual executive assistants. Now, fast forward from 2021 to today, and the world is a different place. And so if I was in the same place now that I was in, I still would have brought in a virtual executive assistant and they would have been able to help me even more because of the AI tools that they could use and that we could use now.
Now, if you’re thinking this doesn’t make sense, AI agents, virtual assistants are just gonna replace them, that’s because a lot of the fear factor and the stories in the media are talking about AI is taking jobs. That sells. It’s like if it bleeds, it leads. If we threaten people’s livelihoods, that’s gonna get them to freak out and watch our news. And that’s what’s happening. But the truth is that AI is not there yet. If you are an avid AI user like I am.
You’ll know that Claude has connectors, right? Where it connects to different systems, right? We use it to connect to Google Workspace, Monday.com, Gmail. And I can tell ya that every week one of those connectors needs to be redone. So it says things didn’t boot or didn’t load or didn’t work that if you say, try again, miraculously it works. So AI is not to the point of fully replacing a human. It cannot do
That ultimate tech tool of turning it off and turning it back on again and restarting, and suddenly things work, that anybody who’s worked in business and has to use technology knows is is a basic problem solving skill, right? So there’s things like that that AI just isn’t doing yet. And so it it’s not, it’s not coming after our jobs right now unless you are purely doing copywriting or something like that.
For those of us who are in skill in jobs that do require a lot of different skills, a lot of connecting things, which is a lot of our computer based jobs right now, AI might take over some of that work, but not everything. And that is really how I am leveraging AI and my various businesses well. And that is how I am leveraging AI in my various businesses right now. I’m saying, what can AI do well? And we’re gonna give that piece to an AI agent.
So then my human being on my team is freed up to do more of what humans do well. So let’s say what does A what do AI agents actually do well? Where do they shine and where do they not? Where do we want to leverage them? Well, first I have found that it’s great to bring in AI agents on routine, recurring tasks inside systems that have a connector or an API so that your AI can play nice with it.
So for me and for my businesses, we do a lot in Monday.com, Google Workspace, Gmail, the Google ecosystem. And we use Claude a lot. And Claude has some natural connectors to those. So I can tell Claude, hey, in the XYZ folder on Google Workspace or on Google Drive, let’s redo that. So I can tell Claude, hey, I want you to evaluate.
the documents that are in folder XYZ on my Google Drive. I don’t need to give it a link. I don’t need to give it those documents. I just need to tell it where to look within my Google Drive. And because it’s connected, it can do that. It’s also great at pulling and compiling data or executing tasks on a schedule. It doesn’t overlook things. It doesn’t forget. It doesn’t get distracted. And it’s also great at maybe doing a first draft of something like an agenda, an outline, things like that, or a graphic.
but it’s not going to be perfect on the first draft without a human involved in my situation, in my experience. And it’s also very strong at creating tasks. So, like I said, we use Monday.com project or task management system. It is very good at both executing tasks around Monday.com and also creating tasks in Monday.com for the pieces that are better done by a human. It’s weaker at
tools with no native connector or an unreliable one that tends to disconnect or break frequently. It’s also weaker, I have found it things that require it to go look at things on Chrome or to open up its own native browser. That’s gonna be a lot slower. It’s gonna take a lot more credits. And so I don’t love it having to go do things on the internet because it’s really just like taking a screenshot and then evaluating it.
if I don’t need it to, right? It can do back-end research, but I don’t want it to read website XYZ multiple times a day, every day. That kind of thing. it also is weaker at judgment that calls that depend on context, what’s happening in the world, in the company, in a person’s life, unless you’re going to give it a lot of parameters of all the what ifs and that information that it wants to check is somewhere that’s easy for it to check, that can be a little difficult.
So there is a lot that it’s great at. And here’s how we split the work between AI agents and human beings with my virtual assistant team. The AI agent does the routine system work and compiles the data or the information that a human virtual assistant then evaluates and executes if need be. So we have our routine tasks that are all in our core task system of Monday.com.
And the AI agent is able to run every morning and see which of those tasks it’s assigned to that it can do and it can execute on those tasks that are all using systems that are connected. So data that’s on our other monday.com boards, data that’s in our Google Drive, data that is in other places that is connected, or using tools in other places that are connected. Canva has a great connector to Claude. So our Claude can use Canva, and I’ll give you a specific example of that.
later on. So the AI agent runs, it sees what tasks it’s assigned to in Monday.com that have a due date of today, and it is able to execute its part of that task. Then if the a if that task requires a human to review to do the next step or to do a piece, it can either post an update and tag that human. It can also create another task in Monday.com for that human with everything they need to do the next piece.
The human then is able to review, decide, and execute. And one of the reasons that a human may execute instead of the AI agent is when the tech connection is missing, unreliable, or limited. So if the execution is going to be in something that Claude and the AI agent can easily reach, then we may just have a human being say, Yep, do it, and Claude executes. Or if it’s somewhere where Claude can’t easily execute, then the human’s going to do the execution.
And again, I’ll give you examples of all of this.
And
I’m gonna give you one example here. I want you to know that we are following this same system across the business, not just in one department. So in finance, in HR, in client care, in recruiting, we are looking at everything and saying, are there pieces of this that are routine, that are in systems that are connected to Claude where an AI agent could automatically do this? And one of the things that makes this easier for us is that we’ve always had a roles and responsibilities.
list, which we actually keep as a board at Monday.com. And in each department area, every task that the person who’s in charge of that department would be doing is on that roles and responsibilities sheet. If it’s as needed, daily, weekly, you know, bi-weekly, monthly, quarterly, semi-annually, annually, whatever it is, every responsibility that falls into that department is listed there. So it’s been really easy for us then.
To go through, let’s say, in this case, we’ll talk about an example from HR, to go through the HR tasks and say, is this something where an AI agent would help take some of the workload off the human being? And if so, then we work with Claude, with the AI agent, to audit the task and say, hey, what parts could you easily do? Let’s get that front, like front loaded and then have the human being take the work you did and run with it from there.
That’s what I’ve been doing. And here’s one real life example. One of the things we do at Workogenics is when a virtual assistant on our team either has a birthday or a work anniversary, we have a graphic that we post in our internal chat with a message kind of celebrating and taking a moment to, you know, honor that person’s special moment. In the past, how this was done is the HR person on the admin team at the beginning of every month.
Looked through our Monday board of our virtual assistants and said, Who has a birthday? Who has a work anniversary? They went into Canva, they made those graphics, and then on the correct day in our internal chat, they posted that graphic along with a message. What now happens, now that we’ve incorporated an AI agent as well, is the agent runs at the end of every month, it checks that same virtual assistant board and it sees.
Who has a birthday? Who has a work anniversary? It goes into Canva and it takes the templates we have in Canva for these, and it’s able to take the person’s headshot from Google Drive and put it into the Canva template with their name and details from the Monday board that has everything and create these graphics. Then in Monday.com on our admin task board, it creates a task for the HR head to post these tasks.
In Google Chat, redo that. It creates a task on every day that there is one of these special events for our HR head to post this message and this graphic in our internal chat to celebrate that person. So now an AI agent is doing 80, 90% of the work for this task, and yet we still have that human being who’s posting it in Google Chat. And there’s a few reasons why that kind of highlight where it’s great for the person to be involved.
We want a person to be involved where we want their judgment or discretion. Or two, we want a human to be involved where there’s not a native AI connection or it’s unreliable. And while there are like third-party workarounds or things you could do around it, to me, that’s just asking for more things to break, right? Like Zapier, I love Zapier, and yet I get countless emails from Zapier about this Zap needs to be addressed. And if you’ve used Zapier, you know what I’m talking about.
So sometimes it’s just easier to keep a human being in that loop where the tech isn’t a native integration. And then, like I said, the other reason is where you want their judgment or their discretion. So in something like this, these you know, birthday and work anniversary graphics, yes, 99% of the time they’re gonna go great. But you’re gonna have, you know, the times that happen that maybe it’s been a crazy day, maybe there’s really bad storms in the Philippines where you know our virtual assistants are located, or there’s some other extenuating circumstance.
Or does it make sense or it’s tone-deaf to celebrate John Doe’s birthday today, right? And in those situations, I like that we have a human being involved in the loop to make that call. And then the other piece is that there isn’t a native connector between Google Chat and Claude, funnily enough, although it can use so many other aspects of the Google workspace. And so that would be a lot more work to actually have Claude.
automatically post those in Google chat. We could have said, change our system and email them out or do something like that. But chat is where we want it to be. And that’s intentional there. And so we want to keep it there and that requires a human being to go in so that we don’t have to use a third party tool. Hey kids, you need to go back upstairs. Okay. Obviously cut that. So that’s a specific example of one procedure
where we’ve had the AI agent take work off my HR heads plate and so now my HR person can do more other things and higher level things and that’s less that I have to do because now I can give them more of these higher level tasks.
And I want to again emphasize the nice thing about having a person in the loop here is first efficiency. It’s often a lot easier considering that not everything has a connection to the AI tool like Claude and that these things can break or be unreliable. It just tends to be nice to use what Claude does well and leave the rest with a human judgment. Again, that other piece we want in the loop, and then accountability. When I have somebody like my HR head who’s still in charge of this task.
If the Canva connector to Claude doesn’t work, which happens, you know, it that person is now in charge of fixing it. That person is the one who’s on this task. That person is the one that the AI agent is gonna tag to fix it, and it doesn’t fall on my plate. Let’s be honest, no matter what AI programs you are using, these connectors, these things they do disconnect, they do break, they do need to be reconnected.
And if you have no humans on your team, it’s just you, that all falls on you to be the tech support. And so when instead I have these humans on my team who are still in charge of these pieces, they’re in charge of making sure it works and they have the accountability to make sure it’s done and it doesn’t fall on me. And now my goal is not to like win some award for having the most AI in my business, although
If you’re in the business world right now, sometimes it seems like that’s some people’s goal. My goal is for us to have the best result in a very efficient manner. And for right now, I think that happens when we put AI agents and humans together. And now let’s talk about what humans get to do now in this world that all of this busy work and more administrative things are taken off their plate. So first
My virtual assistants are seeing they spend less time on routine. Things like this birthday and anniversary post, right? Things that are important and that do add to your business, but are routine and aren’t necessarily very thought-provoking. We also have a pretty decent payroll that we have to do at worker genics as a staffing agency. And AI, the AI agent is able to come in and handle a lot of those routine aspects of payroll as well. And that frees the human up.
to do the higher level time, which really means they’re given more time on evaluation and analysis. They’re working from the data that the agent compiled. So they’re starting at the decision, not at the data gathering gathering step. It also gives the the virtual assistant more time on people. So my HR person now is gonna have more time to actually connect with the virtual assistants on our team because they’re not gonna be so much spending so much time on administrative work.
Also, that allows every virtual assistant to have more coverage. So I do think that businesses are going to need less virtual assistants to do the same amount of work. So your choice is either you have less or you get more work done. You have a business that grows, you are able to cover more. And part of that could be that your virtual assistant team takes off more from your plate than they were before. And now virtual assistants also have.
Better skills to offer. A virtual assistant who can work alongside AI agents is more hireable, they can earn more over time, and they can also do more things because their ultimate core skill is knowing AI, not knowing every system. And this is one of the biggest shifts that I see for hiring in this AI agent world. A VA no longer has to be the expert in these very niche tools or disciplines.
Right? We don’t need a virtual assistant who knows Go High Level or who knows Kajabi like the back of their hand. Because if they can work with AI and especially if this tool has AI or API connection, which those do, then they can work through AI and through an AI agent to get into the nuances of these different tech systems without having to know those tech systems in and out themselves.
They just have to know how to tell AI, hey, we want to update this web page or we want to change these offerings. And AI is going to work within that system to execute and make it happen. The other new VA skill is that they really are going to need to know how to do oversight. The agent is now going to be doing a lot of the work in the system, and the virtual assistant is going to be reviewing what it produces and acting on it. So they’re able to have more of that higher level executive.
thinking that we want to be looking for. And the other thing is, and that the agent redo that. And finally, one of the other big shifts that I see here is that the AI agent is going to handle the routine clean cases, and your human being, like a virtual assistant, is going to stop in when something is flagged, it doesn’t look right, and they’re only going to be handling the issues. And this is somewhat of a different skill set, right? For a virtual assistant
Being able to use an AI agent to cover a multiple of different tech tools and things like that, to be able to oversee instead of really just be in the weeds themselves, and to touch the issues instead of spending a lot of time on routine data entry minutia. That’s a different person and a different skill set. And so that’s something we’re looking at work with worker genetics and making sure that we are hiring and bringing people on who can.
Do well with AI agents and this brave new world and the different way that the virtual assistant role has shifted or evolved with all of this.
So in your business, how are you I’m gonna
So let’s give you a quick and easy checklist to determine if a task or role in your business should go to an AI agent or a human. And it’s really
Okay, redo this again. So as you in your business work to get things off of your plate or maybe off of your virtual assistant’s plate and you’re determining what could an AI agent do or human do, I’m gonna give you a quick list and I recommend that you put your procedures, you put your SOPs into your AI tool and have it audit and help you determine what it can do and what should stay human. But for me, my questions
or my platform for determining this is first, is it routine and recurring? If yes, then I’m gonna want to get an AI agent to be the tri to start triggering that and doing its pieces every time it needs to happen. Two, does it live in a tool with a reliable connector or API? If yes, again we’re wanting that AI agent to do the pieces it can. Three, does it need judgment, tone, or a relationship? If so, I want a human to still own that step.
And is the connection to Claude missing or unreliable? If so, I’m gonna have a human do that step rather than get fancy with third party connectors. And does it end in a decision or an action by a person? Skip that. Delete that does it in a decision one.
So do AI agents mean the death of the virtual assistant industry? Not in my business, not by any stretch. In fact, it’s making the virtual assistants on my team more valuable because they’re now able to do more tasks because they’re not having to do all the steps of them, only the steps that require humans. And that means two that they can take more work off of my plate because now they have more time. And because they’re working with their AI agent buddies, they can do
More complex stuff over a wider area of skill sets.
And if you’ve listened or watched this far, you’re obviously invested in this whole idea of AI agents or VAs. And I would love your thoughts in the comments below. How are you using AI agents in your business? And how are you having the AI agent person combination work for you?
Thank you for joining me this week. Join me again next week for another episode of Scale Smart Grow Fast.
And then join me again next week for another episode. Okay.
I think I wanna redo the intro.
Alright, let’s try a new intro. You can see what you like better.
Does AI mean the death of the virtual assistant industry? This is a hot topic I have seen across the across the spectrum from virtual assistants to
other people with staffing agencies to the entrepreneurs, business owners, and executives that I interact with on a daily basis. We are all figuring out where AI agents come into play, where do we keep human beings in the loop? And so today I’m gonna walk you through as a business owner.
How I am partnering or leveraging AI agents plus a human combination to get more done, take more off my plate, and be more efficient with the humans that are on my team. And at the end, I’m going to give you my quick and easy checklist to determine what pieces go to the AI agents and what stay with humans. Join me here. Stay tuned as we go through this episode of Scale Smart Grow Fast.
Stay tuned as we dive in to AI agents versus virtual assistants.