AI for Virtual Assistants: Skills, Automations, and the Human-in-the-Loop

The Pragmatic AI Podcast is sponsored by Tighten.  T-I-G-H-T-E-N. We will take your AI ideas, prototypes or even vibe-coded apps, and we'll take them to production. Scalable and secure. Check us out at tighten.com.

In this episode, Matt Stauffer sits down with Kayla Helmick to talk about her journey from traditional virtual assistant work to leveraging AI for automations, podcast editing, and app development — and how she turned an early existential threat into the biggest boost her business has ever had.

They get specific about the practical stuff: building reusable Claude skills to process podcasts, fixing stuck Zapier and Tally form automations, and going from playing around in Lovable to building and deploying a full Laravel app with Claude Code — her first being a memorial website for a friend.

The conversation also widens out to the harder questions: keeping a human in the loop for quality, how AI is reshaping the future of virtual assistance and the work itself, and what it even means to call yourself a "virtual assistant" when you're shipping internal tools.

• Matt Stauffer on X - https://x.com/mattstauffer
• Tighten Website - https://tighten.com/
• Kayla’s Website - http://kaylahelmick.com
• Kayla on Instagram - https://www.instagram.com/kayla.helmick/
• Kayla on LinkedIn - https://www.linkedin.com/in/kaylahelmick/
• Kayla Helmick on X - https://x.com/kayla_helmick

-----

Editing and transcription sponsored by Tighten - https://tighten.com/

Big thanks to the companies that support the show:

Mailtrap - https://l.rw.rw/pragmaticai_4

Creators and Guests

Host

Matt Stauffer

CEO of Tighten, where we write Laravel and more w/some of the best devs alive. "Worst twerker ever, best Dad ever" –My daughter

Guest

Kayla Helmick

Virtual Assistant | AI Automations and Workflows

What is Pragmatic AI with Matt Stauffer?

Pragmatic AI is a podcast about practical AI for business leaders, developers, and thinkers—the stuff that actually works, not the hate or the hype. We cover using AI tools day-to-day across the spectrum of human experience.

The Pragmatic AI Podcast is sponsored by Tighten.

T-I-G-H-T-E-N.

We will take your AI ideas, prototypes, or even vibe-coded apps,

and we'll take them to production.

Scalable and secure. Check us out at tighten.com.

Yes.

welcome back to Pragmatic AI, where we talk about using AI in the real world, what works,
how to use it well, and when it causes more harm than good.

Practical tools and real trade-offs for builders and business leaders.

And my guest today is Kayla Helmack.

It says Kayla Helmick virtual assistant, but I feel like just saying the word virtual
assistant does not consider the fact that it is you are a virtual assistant, not to the

stars, because we're not famous people.

You are like a when people hear a virtual assistant, they usually hear

I've got somebody who's overseas who can do really rudimentary tasks for me if I tell them
exactly how to do it.

I have to tell them exactly everything I need with a bullet point list.

And if it's not in the checklist, it doesn't get done.

And that's a different thing.

I feel like you're almost virtual executive assistant.

We should talk about that.

But anyway, it's more than just virtual assistant.

So anyway, now that I've said all that, Kayla, would you say hi to the people and tell
them who are you and what do you do?

Yes.

So like I said, my name is Kayla Helmick and I am a virtual assistant.

That's what I call myself.

I've been doing this for about three years now.

Before that, I worked in a toxicology lab for over 10 years and then I was a quality
manager of a cannabis testing lab.

So it was a big switch.

But I wanted to work from home and be able to homeschool my son and have a flexible
schedule.

So that's what being a virtual assistant allows me to do.

Yes.

So now I work with a broad range of clients and I get to learn all kinds of new things
from each one and then apply it to the next.

So super fun.

So if somebody is not from familiar with what a virtual assistant does, because I said,
you know, people have assumptions, but when I talk to people about virtual assistants, the

vast majority of them either have heard that they should use one and don't know how, or
have tried and been burned.

And very few people have like this good I know what it looks like to work with a virtual
assistant well.

So give me a pitch for what you're doing for your clients on a day to day.

And I know it's varied but kind of what are some of the kind of things that you do?

Um, oh my goodness, a lot.

Uh, so editing a podcast, um, I do a little bit of inbox management and calendar
scheduling.

Um, but now we've got tools for all of that.

So I don't do a whole lot of that, especially in the last year.

Um, with AI, it has changed quite a bit in the last six months, I would say.

So I'm building landing pages and internal tools and automations and just making sure that
their backend operations are flowing smoothly.

Okay.

So historically I would think people people think of a certain kind of suite of jobs that
are negatively affected by AI.

And it's often first of all, knowledge workers, like if you sit in front of a computer all
day, you're you're at risk.

But second of all, people who are doing like nitty gritty work, right?

And historically a virtual assistant does nitty gritty work that the people they're
assisting is not capable of doing it.

Has the advent of AI over the last, especially the last year

felt like an existential threat to your three year old business?

Is it felt like a boost?

Is it felt somewhere in between?

Like what's your initial response been?

A lot of both.

I think, you you guys talk about a lot on the show, just like everyone came back from
Christmas break and AI was suddenly good.

And at that same time, I was like, oh my gosh, it's good.

Why do they need me anymore?

They're going to build all these agents and it's going to do all this stuff.

And so why, why would someone ever use a human again like me?

So I dug down and I was like, I'm going to learn everything I can about how to use AI.

And so.

At that point, now that I've learned a lot, I'm like, oh, I'm okay.

Like it's good.

But I still need to guide it.

So now I'm able to do a lot more.

I'm able to do a lot more work than I was, you know, than I could, I could have never done
a lot of things I'm doing now, six months ago.

So yeah, there's a huge boost.

but then are you like, I can do more than I could have before.

Do you feel like your clients are understanding the the increase in value that they're
getting from you as you're aided and assisted by AI?

Okay, cool.

I have one client and he's like, uh I think AI could help us with this.

Like, I feel like you're kind of the AI guru.

Can you just like go figure it out?

So that's super fun.

And I, you know, I think I learned something new every day almost about it.

So

to fight with um any kind of imposter syndrome being like you're looking at me as the AI
person, I still feel like I'm still learning, or are you at a point where you're like, you

know what?

Like I understand that my level of knowledge is at a place where I know some things other
people don't, or is it again, it's a little bit of both.

Okay.

Yeah, okay.

uh

I don't know anything.

ah But then, you know, when I'm talking to my friends or my husband, I'm like, my gosh,
look what I learned.

Look at it.

You know, look, I can do now.

And they're like, I have no idea what you're talking about.

So yeah.

Yeah, for sure.

talk about um like some like if you could just pick one thing that you did before for
clients and you're still doing now.

So not things that have yet you added new yet, but things you were doing for for clients
before that you're still doing now, but you're doing now with the aid of AI.

Could you give us an example of one of those?

uh One is this podcast.

We've been editing podcasts for three years now.

And it started with manual transcription, and we are no longer doing that at all.

It started out with sending the audio off for editing, and now we have AI that helps us a
lot with that.

uh Riverside has improved a lot over the years.

uh So that's something that I think, one, used to cost you a lot more money.

just in like outsourcing outside of a VA.

um But two, used to take us a really long time too.

um And so now we're able to produce a lot more content in a shorter period of time.

Yeah.

one of the things that we said when we when I was first starting this podcast is it's an
AI podcast.

Let's test the realm.

Like, what's the edges of how much we can lean on AI and A, get good quality and B feel
ethical about it, right?

Because there's always the questions of like and one of the number one things that we
found that was important was the trying to do a weekly podcast with our current systems

was really, really not super attenable.

Um and using automated systems, even if they're imperfect.

Automated systems plus Kayla's review is a very different outcome than waiting four days
for that.

And then it goes to this other person, you wait two days for that, and then Kayla reviews
it afterwards.

So there's still humans in the system.

There's still, you know, a personality here, but we just couldn't we couldn't do the
timelines we were trying to do beforehand.

Um

even, you know, with show notes, like we have skills now.

So I'm like, hey, here's the transcript.

yeah.

So so we get an automated transcript.

Cause I think it's one of the things that's fun is actually talked about the practical
ways we're using AI in a day to day.

It's pragmatic AI.

Because a lot of times people talk about you could, you could, you could, but the ones
that actually give me a vision of what I can use is people are saying, Here's what I am

doing.

And I'm like, I might not want to do that exact same thing, but it gives me a vision.

So talk to me a little bit about like what do we set up with the skills.

So I've got three different skills because we do two different podcasts and the two
different podcasts talk to another skill.

so a skill is I'm not having to go in and give it instructions every single time.

So when I built out these podcast skills, em I now just go in and I say, hey, I need you
to process using the pragmatic AI podcast skill, em process this episode.

And so it's just asking me for an episode number and a guest name, and it knows exactly
what to do based on instructions that I've previously given it one time.

So I'm not having to go in every single time and say, hey, this is what we have to do,
yes.

So.

a technical perspective, it's the same thing.

A technical from a technical perspective, you define a prompt maybe with parameters.

So you're you know, you the prompt has a name, the prompt has maybe parameters like, you
know, like like you said, episode number, speaker name, and then it's a set of

descriptions, uh in in plain text in Markdown that the AI can use to say, um, to do
something in a more kind of consistent and repeatable way.

Um and so for example, from a tactical perspective, if I'm

as a programmer consistently pulling down a piece of code, running the same steps on it,
having AI run the same steps on it.

And then I want the AI to give me the output in a particular shape.

I don't want to every time say, please pull this down and run these steps and give me the
output in this particular shape.

I say, hey, here's a skill and it's called review shift pull requests.

It's a type of pull request review.

And then it says hey, when you want to run this particular skill, the user, me, is going
to tell you what number of pull requests it is and then take these steps.

And so instead of having to basically copy and paste prompts here and there, it's like a
it's like a prompt library, right?

With with a little bit of customization.

So you have a prompt, a pre made prompt called a skill that is for the pragmatic AI
podcast and then interrupted you.

So tell us what it does.

So it looks at the transcript.

We're now doing uh SRT transcripts so that on Apple Podcasts before they released video uh
It would you know follow along with the with the script and you could just click on the

words.

It was really cool um But it's giving us show notes titles em social media clips keywords
em and then it's

this skill for the transcript.

So this first specific skill, do you feed a transcript that you downloaded from Riverside
into Claude and then say process it to give me?

So what are you asking it to give you?

Like what's the skill actually doing out of the transcript specifically?

as doing all of those things in one batch.

giving clips out of that too?

Yeah, so it's just telling me like what time period to go look at for the clip.

And so then I compare it to Riverside.

what are some potential interesting moments basically?

Wow.

I didn't even realize that.

Okay, cool.

Yeah.

And that's all just a single run with Claude with that skill with the transcript.

And you get all those and then you take them and you generate clips from it or you tweak
it and make the show notes from that or whatever else.

Right, yep.

And it'll even grab some like links.

It'll generate a blog post.

It'll generate, we do a little bit of a different YouTube transcript versus a regular
transcript.

So I get it all in one pass and then I just download the files and I can take it from
there.

And so then there comes the human review.

But I've got a starting point and I'm not staring at a blank screen.

Like what are we gonna title this podcast?

Yeah.

And when you're when I'm putting you through at least two podcast episodes a week between
different podcasts, that's a lot of time savings.

That's a lot of effort savings.

And and it's still human in the end, because, you know, for example, you'll if you get a
clip, and for those who don't know, we we take little, you know, thirty to ninety second

clips of each episode and we put it up on social media.

Um, but AI is gonna tell you look at these areas.

And then you're gonna go Riverside, you're gonna pick which of those you actually think
are worth doing.

You pull them out as clips, and then I look at them and I say, I like these two.

And then you go back and then you write a tweet promoting that clip, and then I edit the
tweet.

So there's still so many human aspects, but instead of you and I having to do all that
from scratch, we give like we get like a like a playground to work with, right?

Like we get we get the ability to say, I'm gonna respond and and critique and evaluate and
edit rather than I have to start from from

Plain, you know, plain screen.

So okay.

And sometimes I'll use a couple different tools.

So Riverside also selects social media clips that they think would be interesting.

So I might compare that to what Claude said might be interesting.

um And then, yeah, it's fun.

So podcast editing.

I keep hearing people talk about like kind of all in one platforms.

And my biggest concern with these all in one platforms is they do usually completely take
the human out.

Um, is there a part of you that would be happy if you just plugged a podcast in and it did
all the work for you?

Or is there a part of you you're like, nah, the output of this AI is not good enough.

I it's it's good for me to have a human in the middle of it.

Like, where are you on the quality of output you're getting?

It's still not good enough to just completely not have a human in it.

ah There are times like, and I'm using two tools usually to process one podcast episode.

And I'm taking maybe the writing style of this one versus this one and combining it as a
human.

So it doesn't sound like a complete robot wrote it, at least I hope.

Yes.

No, yeah.

And and that's a big part of what we do is is if you just took the output of these tools,
it would be the it's not this, it's that, or excessive, you know, uh not hyphens, what are

they called?

excessive em dashes.

And I love I'm an em dash like l fan.

And I can still tell when something was written by, you know, by AI.

Okay.

So yeah, it's excessive.

Okay.

So that's one of the examples that you do.

I know that world because I know podcasts.

One of the things that I don't do personally is use a lot of AI and automated business
processes in the back end.

Have you been doing much of that for anybody?

Yes.

So, know, part of that, I've always been building automations.

Actually, I kind of just started a blog and I wrote about one that I made in 2018 and I
handed it off to Claude and chat.gpt just to see if it would come up with the same

solution.

And it eventually got there.

But it, you know, what took me weeks back in 2018 to build took it a couple of minutes.

So it's, you know, a really, really cool thing.

Yeah.

So now.

You know, I've got like tally forms for one client.

We use a lot of tally forms and we connect it to Zapier and then we send it to Monday.

And so we've got a lot of information that's transferring, but sometimes I would always
get stuck in this automation, you know, before AI.

And I'm like, well, this one field just won't go over because of, you know, things just
aren't talking to each other.

It's not a simple field.

And so now with AI, I can ask it and I can describe my problem and it'll give me like a
script to run as a step.

Uh huh.

Yeah.

So as far as like automating back end business processes and even like connecting Stripe
checkouts into forms so that I can see what they've paid for.

Like I never would have known that before or I would have had to go dig into, you know,
forums and threads and, you know, Googling a bunch or the docs for Stripe to figure all of

that out.

So there's a lot of back end business stuff that I've been able to automate that I
definitely wouldn't have been able to before.

what environments are you working in?

Are you, you know, as you're doing these automations, you mentioned Zapier, I don't know
tally forms, but I assume it's just kind of a form software as a service, you can make

your own forms.

Okay.

are you working any of the more like uh AI focused ones?

Cause I know a lot of people are trying to be like the AI version of Zapier or like
Notion, we were just looking at yesterday, all of a sudden it's Notion AI or whatever.

Are you using a lot of business automation tools that are very AI forward, or are you
using more traditional automation tools, but AI is helping you use them?

I think AI is just helping me use them better.

ah know, Monday rolled out like a lot of AI in their platform and I haven't found it super
useful.

Yeah.

d cause you use a lot of AI?

You you are the target market, I think, for a lot of these companies wanting AI because
you want intelligence and you want, you know, a holistic view on these.

And yet most people I know are just sort of like, Yeah, they add these AI features and we
don't actually want them.

So I was curious, are there any are there any tools you use where you're like they added
AI and I think it's a really valuable, you know, add to the tool?

In Zapier it's actually super helpful.

It's great to just be like, hey, here's what I want to do.

Can you build this zap for me?

And then I go in and edit it.

um So that's super helpful.

of the zap, it's AI as helping you build the zap.

Okay.

Okay.

And it may just be like, haven't come across the need for it yet.

So, you know, and I think, you know, like with customer service, yeah, you know, I think
with customer service, you know, it's, super helpful, but none of my clients have needed

that so far.

So.

Interesting.

Um, okay, so you mentioned that you are also homeschooling your son.

what has touching on the conversations of AI been like as a a teacher, um, as a parent,
and as also like trying to make decisions for a kid?

I hope you don't mind me asking this, but like what where's AI at play there?

I think about it almost every day.

um There are times where it's super helpful, especially when I don't understand a math
problem.

Instead of like going to YouTube and trying to figure it out, I'm like, hey Claude, can
you show me how should work out this problem?

Because I forgot how to.

um So there are times where it's, yeah, and I use a calculator for this now and I have to
teach someone how to do it manually.

So there are times where it's super useful.

I had it like,

We had some test scores come back and there was a section where he just didn't do that
great.

so I was like, how do we fill in these gaps?

Cause I had holes, you know, in a lot of my education.

Um, and there's a lot of education tools out there where like, as they're struggling with
a problem, they've got AI built into it and it'll take them back to the basics that have

caused that hole in their education.

Um, we don't use anything like that.

But it was interesting.

So all summer I've been like, hey, can you generate a worksheet to fill in these math gaps
that we have?

And it's tailored to his grade and what he missed on the Stanford test.

uh So that's super fun.

But as far as like him using it, um I let him watch me use it sometimes because I'll be
like, hey, look, let's kind of use it as a thought partner.

Or we used it on planning out one of his projects.

So I'm like, you're going to write the plan.

You do it all.

And then, you know, we can see if we come up with any more ideas with AI just to add to
it.

um So, but I don't, I don't want him to ever just go to it.

um And I use it.

Yeah.

Yeah.

And I use it a lot just to learn.

Like I'm constantly watching it.

Like, okay, what's your process here?

You know, cause I'm just a curious person, but I don't want him to lose his curiosity.

And I feel like if, you know, that's a big thing that we have to worry about just handing
over these.

Yeah.

Do you have any anxieties that like, if I don't get him ready for the AI world he won't be
ready?

You're like, nah, it's fine, he'll he's he's capable of learning.

There are a lot of students in his class that their parents don't let them have any
exposure to it.

And I have to disagree with that because I don't think it's ever going away.

And I think it can be really useful.

And, you know, if you look at history, there are jobs that have been replaced that you
even look back and you're like, wow, that was a whole job, wasn't it?

That was a whole job.

Like people used to have, you know, a lady sit in the corner and just type all day because
that was their job.

Yeah.

And that's not a job anymore.

Yeah.

So I want him to be prepared.

I don't know how to prepare him 100%, but we are talking about it.

We're also always talking about, you know, not leaning into it 100 % and always fact
checking and, you know, but use your brain first, be curious first.

So.

love that.

I mean the you I know that you edited the podcast where I had an educator on who's talking
about we want to make sure that they know how to be able to think on their own and

everything.

So I'm just like, that's what we want them more than anything else.

Okay, I want to talk more, but I'm gonna pause and thank our sponsor, Mailtrap.

so today's sponsor is Mailtrap Enterprise Grade Email Delivery Infrastructure for Modern
Software Teams.

Mailtrap helps teams deliver an email reliably at scale.

The security, compliance, and infrastructure control modern organizations require, built
for enterprise environments and ready for AI era communication workloads.

Mailtrap integrates easily into existing infrastructure without slowing teams down.

So see how it works at mailtrap.io.

Thank you so much, Mailtrap, for sponsoring.

And for those who don't know, normally what I do is I say where I'm about to read it, and
then we just take a little break and like sip our waters, and then I I record that later,

and then Kayla has to go back in and edit it in.

And I'm like, you know what?

I'm just gonna read the ad right now because then Kayla doesn't have to deal with editing
it in later.

So

Yes.

Looks great.

Good.

Thanks.

Um, okay.

So I want to talk about how your process of learning AI went for you.

Um, so you are a I I think I know you well enough to know that you're probably someone who
doesn't describe yourself to the surrounding world as being technical.

However, you are a VA for a CEO of a tech firm, and several other people you're working
with also work in technology, and you are actively writing code.

vibe coded with AI, but still you're actively writing and deploying code on the internet
today, which makes you a highly technical person relative to normal society.

Were you there when you first started getting exposed to AI?

Were you at all technical?

And I know that you'd probably say you're not technical now, but like compared to normal
society, were you always the person who was like on the latest software and learning

things, or is that all growing significantly since adopting AI?

It has definitely grown significantly since I adopted AI.

Before I would, I mean, I was just like a pro office user maybe, you know?

In the lab I worked at, we had like an HL7 interface and I was the fix-it girl.

It was old school, it's very old school.

got it.

had a server in a big closet.

And so if something wouldn't go up to the magical web portal, you'd have to go in and find
these files um and you'd have to find what was broken in them.

And so I would just like compare patterns, you know, but like, so I was, I was technical,
but not technical in building.

but pre AI, you were that person who was able to be curious and dive into the things and
learn how the systems worked.

Even if you didn't have the training of the background, you were that person who's like,
Yeah, I can figure it out.

I can understand this, right?

I've always been a figure it out-er.

And it makes sense that that transitions into figure it outing, you know, with AI stuff.

So what was your first exposure to AI?

Were you just a casual chat GPT user and it's slowly ramped up?

Yes, was definitely, can you please rewrite this email?

Especially as I started working as a virtual assistant, know, AI, like Chat GPT came out
shortly after that.

And so I was like, oh cool, I can sound so much smarter, because I wasn't used to working
online.

You know, I worked in a lab doing testing, you know, and I would write the occasional
email, but I wasn't like communicating all the time online.

So it was great.

But then it progressed.

Um, and to now that I, I constantly have like a Claude code project going on in the
background.

Um, yeah.

I went to an event in Atlanta called AI Edge that my wife invited me to.

Uh I was with this group called Blavity, and she's a fellow with Blavity, so she kind of
knows about these events.

And they sat everybody down.

They said, If you've never done any AI code building before, we're gonna sit you down for
three hours and we're gonna teach you everything you need to know to learn how to build

your app.

And I was like, This is like what Kayla did.

And

With with love and respect, their their demo didn't work out well.

And so they really kind of taught you like two minutes of it.

And then they just circled around saying, if you get stuck, help us ask us for help.

It's it didn't the the teaching process didn't work the way they expected it to.

It wasn't what it could have been.

And yet they brought people together in a place and said, What here's what you need to do.

You need to install cloud code and you need to first design the system, then build the
system, and then review the system.

And it's just they they basically just taught that and then they just said ask if you have
questions.

And that and alone

was enough for several people who had an idea and had a little bit of technical bent to
have a proof of concept by the end of three hours.

And that was mind-blowing to me because I was just like, I always would have thought it
would have taken so much longer for people to get to the point where they're writing code.

So you have deployed I and for those not familiar, I would say, uh you can tell me if I'm
wrong here.

I would say there's two worlds of learning how to code with something like Claude Code
when you don't have coding experience.

One of them is the

It just sits on top of a website and you click on it and nothing happens.

Nothing's persisted.

There's no databases.

There's no users.

There's no logins.

There's auth.

It's just a tool.

And that's step one.

And that's difficult.

And but then step two is building fully full stack applications where, yeah, there's an
interface, but there's also databases and there's cache and there's memory and there's

back end frameworks, stuff like that.

So you do both.

Uh everybody at this thing I was at yesterday just did that first piece and it was
impressive.

But you do both.

So what was the process of going from I assume never having written any code in your life
prior to writing and deploying your first piece of code?

Um, so I did start in loveable.

I played around with it because I was like, cool.

Um, and I kind of built.

can you tell them what it does?

And I can fill in any gaps if you need.

Yeah, so Lovable just lets you, I mean, you're pretty much just describing what you want
and it just goes and builds whatever app that you can dream up.

it's, yes, web app.

Yes, or at least the last time I used it, I haven't been in there in a couple of months.

But I built like a whole fitness web app that I can access from my phone and it does all
the things that I want it to do.

I was able to like hook up.

some APIs to it so that it could talk to a chat GPT.

But it was just, you know, it's really easy.

It's just like, you're just talking to a human.

Like I want it to look like this.

I want, you know, can you move this button to the left a little bit?

And so that's what Loable is.

So so the first thing you use.

So the benefit of lovable is they aren't just building code for you.

They're hosting the code.

They've got kind of design systems built in.

So you describe it, and as long as you're willing to keep paying lovable forever, you
really get the thing kind of like like I would say if somebody wants to build and I'm glad

you mentioned that because I didn't even thought about it, if somebody wants to build an
app or a website with AI, lovable is just the easiest freaking way to do it, right?

Like and

And you could build something that's just an informational website.

You can build something that's actually interactive.

you're not gonna launch a software as a service off of Lovable, but if it's just something
you're using or just something you're putting in the internet for people to visit and

rudimentally interact with, that's the place to get started.

So that makes sense.

But you built things before you actually got to building full stack level applications.

I feel like you've built some things like that PDF tool.

What did you build that in?

Because that's more than just lovable.

yeah, so I used Claude code for that.

So lovable was just, kind of played around with it.

I was like, oh cool, like you can make things by describing like with just normal words.

I don't have to know anything about coding.

And then I moved over to Claude code and I started using Laravel and I was able to build a
memorial website that I built.

Yeah, so the memorial website was the first

thing I've ever built with Claude Code.

And I had started like playing around in website builders like I had in the past.

I was like, this is gonna take one forever.

And two, my friend who lost her son in a sledding accident, she wanted a couple extra
features.

And there was no way that I was gonna be able to do that in just a regular, you know, like
show it or Squarespace, anything like that, website builder.

So I was talking to a friend and he was like, just download Claude Code and tell it to go
get these Laravel instructions and just start building.

And so I was able to build it a lot like I did with Loveable, just giving it, you know,
just normal words.

Like, this is what I want.

And so I put it in plan mode and I kind of reviewed it.

And I was like, yep, looks good to me.

I mean, sounds good.

mode for anybody who hasn't done it before.

So with cloud well actually let's talk about cloud code.

Did you download Claude code and install it like in a terminal green text on black
background eighties looking style?

I had to ask Claude in the web version how to do that.

But I did it.

like how to get a terminal on your local machine and everything.

Okay.

So if somebody wanted to follow in your footsteps, they start with Lovable, but if they
see need something that's a little bit more complicated than what Lovable can do, they go

to Claude.ai, they sign up, they say, Hey, I want to use Claude Code, teach me how.

And then they install Claude Code in their local and then they open up Claude Code.

Did you make like a new Laravel app and then open up Claude Code in that directory?

I think Claude Code made it for me on the first one.

Yeah.

it up in like in a new folder and you're Hey, I want a Laravel app in this folder, make it
for me, and here's what I want it to do.

Yeah, well now I'm using the desktop app.

So I prefer to use like the desktop code part of the app.

desktop code part edit code on your local machine or does it edit code directly up in
GitHub?

okay.

you can point it at the folder now that I understand that.

a new person should download the cloud cloud desktop app.

Okay.

and then just point out a folder or it'll create it for you.

But so for those not familiar with Laravel, Laravel is a full stack web application
framework that is as, you know, complex in how you host it and configure it as it possibly

can get on the internet.

This is not something made for people who don't know how to code like Loveable Loveable
was made for people who don't know how to code.

Laravel's not made for people who don't know how to code.

So what was your experience of building, but then also getting it on the internet?

Like what was that, what was it like for you?

So I just asked Claude Code what to do next.

um And so because of you guys, I knew about Laravel Cloud.

So was like, well, that's where I will push it to the internet.

But I don't know how to get it up there.

So I just asked Claude Code to walk me through it.

um And so, yeah, um it didn't know Laravel Cloud um as far as like the user interface.

So, you know, like Cloud for like...

Cloudflare, it knows quite a bit about the user interface and it'll tell you exactly where
to go.

But for Laravel Cloud, I had to like screenshot and I'd be like, is this where you want me
to go?

Like, is this where I'm supposed to go?

And so I just feed it screenshots.

And it'll be like, yeah, it should be in there.

Like you're looking for, you know, some type of this word and that's where, you know, like
your environment variables go or you need to make like a new object storage bucket or

yeah.

So I just walks you through it with screenshots.

Thought about this, but Laravel Cloud, Laravel just released a tool.

And again, for those who don't know, I work day-to-day in a framework and ecosystem world
called Laravel.

And it's a chunk of code that we use to build apps, but they also have uh tools we use to
host our apps called Laravel Cloud is the main one.

Um and Laravel just officially released uh maybe a couple of months ago, uh a Laravel
Cloud skill that teaches your local agent how to interact with Laravel Cloud.

And I hadn't thought about this.

That they probably have a series of tools so that the agent can directly interact with
your cloud so that you don't have to be doing exactly what you're doing there.

Cause like they have a CLI.

Again, I'm I'm trying not to be too nerdy here, but Laravel Cloud has a CLI, which is like
claud code on the terminal, but now it's for managing your Laravel Cloud.

So if you connect your Laravel Cloud CLI, you can make all these changes that Kayla was
having to figure out in the user interface and you can make them all through the terminal.

And the benefit of stuff being in the terminal is

An agent can use it easily.

So rather than an agent being like, give me a screenshot and I'll tell you what buttons to
click as I try as an agent to parse what the page says, the the it says, you know, that

the the CLI tells the agent you can do these things, and the agent says, I want to do that
thing, and then it does that thing.

So it's I'm fascinated by the idea of uh using AI-based tooling to enable

non-programmers to do things they couldn't do before.

And of course, there's a lot of people in my world who are terrified about this idea
because they're like, we're not gonna have jobs.

I I don't have that concern.

Just like you that moment of like, am I losing no, no, no, no.

But I do think that there are some people who will be able to use AI for things that they
only could have gotten from VAs in the past.

I think there's people that are going to use AI for things they only could have gotten for
programmers of the past.

But I also think that as a VA, you can do more than you could before.

As a programmer, I can do more and offer more of my clients.

So I'm like I'm hoping that there's a world where there's space for each of these.

Um

So as you think about building apps for your clients, uh that's a whole different
offering.

Are you are is is your job title gonna change at some point, or do you think this is still
all a part of what a VA offers to their to their clients?

I am processing that.

So I was working on redoing my website and it's a lot about AI and automations and
workflows.

And I came across that point where I was like, don't know what to call myself anymore.

Cause I'm not AI, which I feel like a lot of AI is a virtual assistant.

Like I'm not a robot.

not.

uh And I want to distinguish myself because I'm not just managing inboxes and following up
on things.

um

I'm building a lot of internal tools and so I don't know.

I'm still processing that.

I don't know what it'll look like in the future because it's changed a lot.

Yeah.

I I'm fascinated by that idea because, you know, one of the things we're actively working
on Tighten at Tighten right now is, you know, I took over as CEO in twenty twenty two, I

think, or twenty twenty three.

Um, and my first thing to do was just to learn all the business operations of the company.

But now that I'm a couple of years in, once I know it, now I'm like, great, now my
systemization process happens.

Now I want to not be the one doing all the things, but I want to build systems and
processes to ensure that other people do do them and report to me.

So I'm actively in this place.

I'm just like, let's automate everything.

Let's systematize everything.

You know, like not to get people out of their jobs so that people aren't doing the same
rote crap all the time.

Cause often what a VA does is do the rote crap so other people don't have to do the rote
crap.

And so I love the idea of a VA being able to say, let's set up a system so nobody has to
do the rote crap.

And now we're all doing more interesting things and more creative things.

Um, is there a power that you look six months out and you're like, if I could do that

that would really change things.

Is there some level of building or integrations or tooling or something like that?

You're like, if I could offer this to my clients or if I could do this myself, that would
change things.

Or is it this point you're like, th I I can do everything.

I just need to I need to do it.

Um, I feel like right now I can do just about everything, but I would want someone with
more technical knowledge than I have to review what Claude and I have created before I

could offer, you know, something else.

So internal tools, I'm fine offering, you know, as long as like I'm not dealing with
private information.

I'm just making things work easier and better.

But as far as like offering things that I could probably charge for more, I would want
someone, you know, with more technical knowledge than I have to know if this is going to

break in six months.

So.

the it's the Kayla and Tighten show where you build something and then we we audit it and
then you deliver that to your client.

Okay, great.

We'll talk about this after the show.

I love this.

Very cool.

Um, okay, so I wanna kind of wrap up with some of the broader questions that I don't
always get a chance to ask everybody, and then I gotta get to the community story.

But some of the broader questions I don't always get a chance to ask are what do you think
the future of the world is given the you said it's here to stay, right?

Like you're you're in that place of like it's here, whether I like it or not.

Do you think that's a positive?

Do you think and and you've talked about like from a perspective of you as a VA, but let's
talk about the world.

Uh is this a positive?

Is this a negative?

Is it in between?

Are there parts you like and parts you don't like?

Are there movements of AI that you're not happy about, but other movements like what's
your overall perspective as you look at the world ten years from now versus the world ten

years ago?

can't imagine what it's going to look like in 10 years because things have been moving so
fast.

um Yeah, but I think, you know, because I am able to do so much more than I was even six
months ago, you know, it's fairly exciting.

uh You know, what does my business look like in a year?

I don't know, but I think it's a lot better than it is now.

um But I just worry about society and not actually having the knowledge.

um And that's just, yeah.

Yeah.

you know, I'm even, I've done it.

I'm like, I don't really need to know this.

Like just give me the answer.

And I'll catch myself and I'll be like, no, you probably should go back and read like, or
ask it to explain it to you.

eh Because it doesn't help me to just get answers.

eh So it goes back to that curiosity.

Being, always being curious, but always learning.

Yeah.

I don't know if you've had this experience, but I moved to Atlanta where everybody drives
with GPSs, even if they know where to go because traffic is so variable.

Um, and I hate it because I've have not learned the layout of the town like I have every
other town I've lived in, even when GPS existed, because I will choose not to use GPS so I

can learn things.

But in Atlanta, I can go I can know where to go and go there and get stuck in 45 minutes
of traffic.

So you have to use GPS every time.

And I really feel like I just don't have a connection with the layout of my city like I
would otherwise.

And so I I wonder how much chat is going to be doing that, you know, and access to these
LLMs overall is going to be doing that where when you can rely on the thing every time, it

just what even develops that curiosity in the first place?

And I of course both of us as parents were trying to figure out how do we keep that
curiosity in them, which is why I I am delighted when I hear that, you know, people said

the Gen Z is a rebel against AI.

I'm like, well, I hope you guys can be realistic.

But also good because I'd much rather try to talk you guys into using 5% of AI so you can
get good jobs versus talking you out of being complete AI zombies, you know?

Yeah.

Um, okay.

Let's see.

Let's see if there's anything else I wanted to ask you before we wrap it up.

if somebody is in your situation where they are a curious person with no full, you know,
fully technical background, but certainly technically capable.

And they were like, you know what, listening to Kayla on this podcast has been enough for
me to decide that I want to at least learn some of this.

Where is the first place you'd have them start today in using AI?

Claude code.

Just go download the desktop app and start playing with it.

ah You know, if you have an idea that you've always thought like, man, if I could just
build that, if I, you know, could develop that, that would be so cool.

Like I think back 10 years ago, I had so many ideas in the lab, if I could just build
them, but I couldn't, and they were always very expensive to have someone else build.

ah You know, but I could have built so many internal tools for us.

Hm, that makes sense.

And so now when I have an idea, I'm like, oh, let's go try it.

ah And so, you know, I might keep iterating on it.

But yeah, I think you just go download it and go play with it.

Yeah.

Okay, very cool.

Um, is there anything that you hope we were gonna talk about today when you knew we were
coming on that we have not gotten a chance to get to yet?

ah I can't think of anything.

Okay.

Well, my last thing is as you know, I like to share ways people from the community are
using AI.

And uh my wife and I just this morning looked at a home in our neighborhood that we're
considering um trying to to get in and and fix up uh because we love the idea of fixing up

homes and because we also love to have some family in the neighborhood.

And so I saw this one and I had to look at it.

So Cam Kemshaw said, no, Cam Cam Kemshaw Bell said,

In the first half of 2025, before my daughter was born and starting a new job, I took six
months off to renovate my house.

I used ChatGPT to learn the skills I lacked and ended up doing a full house renovation,
which had previously been quoted $130,000, but taking the time off and doing it myself

came around $20K for materials.

I worked right up until the week before my daughter was born, got finished just in time.

It was a really rewarding experience, but could not have been done without using AI.

I could tell you that story in a much smaller scale using YouTube.

Right.

Like the number of times I and people I know have said, I taught myself how to install a
ceiling fan using YouTube But I've never in my life imagined doing something that large

using AI.

So I have to ask you, Kayla, are you using AI in your day to day life to change your
relationship to household chores or repairs or anything like that?

Does this resonate with you in any way?

Um, we haven't used it.

I think I used it to like help us design our bonus room a little bit and just kind of
visualize.

I'm a very visual person, so my husband will come up with this like great idea.

And I'm like, I'm not seeing it.

Yeah.

So I like went and picked out chairs and I was like, can you put these chairs by the
window?

And so it generated an image.

Um, but I haven't asked it for like specific instructions.

I'm still going to YouTube for that.

Maybe because I'm a visual person.

Yeah, I mean I'm nervous because I don't know about you, but I've definitely asked it for
recipes at times.

And they're not recipes, they're things that look like recipes.

Right?

Like the same thing as when I ask it for code and I'm like, it's it's it looks like the
right code is not the right.

So I'm like, I don't want you telling me to do something that's gonna give me an electric
shock.

Yeah, I will say when, so I was baking last week and my icing did not do what it was
supposed to.

And so I asked AI, was like, can I do anything to save this?

And so it gave me a full science lesson and then told me like some things I could try.

It's still flopped, but I was like, it was a nice thought, I tried.

God.

Yeah.

Okay.

Well, I'm fascinated.

I have to probably go find this tweet from Cam because I need to learn more about this
idea.

Because I do I I what if we end up getting this house, I'm gonna have to do all the work.

You know, that's yet other than maybe the things that I can't safely do.

And also, I don't know if you've ever done drywall.

I'm not touching drywall with a ten foot pole.

But uh you know, other than drywall and like things that will kill me, like electricity,
I'm gonna try and do everything.

So I'm just like I definitely assumed it was gonna be YouTube university the whole way
through.

So I don't know.

We'll see.

I might try it try chat and see where it takes us.

Okay.

Kayla, if people think you are fun and fascinating and or they want to hire you to be a
virtual I even know if you're taking on clients right now, but if they did, how would they

follow up with you and learn more about you?

You can just go to Kayla Helmic, H-E-L-M-I-C-K dot com and everything is linked there.

Super easy.

Yeah.

ah

Kayla, you did great.

I know you're normally behind the scenes, but this was a ton of fun.

You uh fascinating stuff.

I learned stuff, which is great because I talk to you every week and I still learn things.

So thank you so much for joining us.

Awesome.

And for the rest of you, we will see all next time.

You know, right.