Ep 136 – JP Holecka, Power Shifter Digital – Building the AI-Driven Agency: Lessons from Power Shifter’s Evolution
Agency Bytes is proud to partner with Ignition.
START YOUR FREE 14 DAY TRIAL
ignitionapp.info/agencybytes-trial
Use Code OUTSIGHT25 to save 50% off!
Listen & subscribe on the platform of your choice
Featuring: JP Holecka, Power Shifter Digital
In episode 136, I sit down with JP Holecka, founder and CEO of Power Shifter Digital, a Vancouver-based agency leading the shift toward AI-driven digital products and content creation. With over 30 years of experience in design, film, and technology, JP has guided multiple teams through successful AI rollouts—transforming workflows, scaling creativity, and redefining how digital agencies deliver value.
We talk about what it really takes to evolve your agency for the AI era, how to navigate the culture shift that comes with automation, and why embracing AI is less about replacing people and more about amplifying what they’re capable of.
KEY BYTES
• AI isn’t replacing creativity—it’s amplifying it
• True transformation starts with changing workflows, not job titles
• The most successful AI rollouts start with internal adoption before client delivery
• Leadership has to model curiosity and experimentation
• Agencies that treat AI as a tool, not a threat, are finding their competitive edgeSHOW REFERENCES
Access JP's AI Miro Board referenced in the episode here. PW: cleardigital
CHAPTERS
00:00 Introduction
02:01 JP’s background and the evolution of Power Shifter
06:32 The first AI experiments that changed everything
10:45 Getting team buy-in and overcoming initial skepticism
14:58 Building processes around AI rather than forcing it in
|20:10 Human creativity in the age of automation
25:36 How AI has changed client expectations
31:12 Leadership lessons from scaling an AI-driven agency
36:45 The next frontier of digital work
40:30 JP’s advice for agency founders starting their AI journey
43:00 Rapid Fire Questions
JP Holecka is the founder and CEO of Power Shifter Digital, a Vancouver-based agency leading the shift toward AI-driven digital products and content creation. With over 30 years of experience in design, film, and technology, JP has guided multiple agencies through successful AI rollouts—transforming workflows, scaling creativity, and redefining how teams collaborate with generative tools.
-
Steve / Agency Outsight (00:01.156)
Welcome to Agency Bites. I'm your host, Steve Guberman, from Agency Outside, where I coach agency owners to build the business of their dreams. This way, this, shit, see, take two. I also didn't ask, how do I say your name? Is it just Holecka?
JP Holecka (00:10.903)
No worries.
Yep, JP Holaco is good.
Steve / Agency Outsight (00:15.728)
Cool. All right, take two. Welcome to Agency Bites. I'm your host, Steve Guberman from Agency Outsite, where I coach agency owners to build the business of their dreams. This week, I'm joined by JP Halecka, founder and CEO of PowerShifter Digital. They're a Vancouver-based agency leading the charge in AI-driven digital products and content creation. He's got more than 30 years in design, film, and tech, and he's helped agencies transform how they work, scaling creativity, streamlining workflows,
and rethinking how teams collaborate with generative tools. Thanks for joining me. How are you?
JP Holecka (00:50.966)
I'm good. Thanks. Pleasure to be here.
Steve / Agency Outsight (00:52.8)
Awesome. Yeah, man, I love what you've been doing, what you've been sharing. We're friends through the Bureau of Digital and it's a great community and a lot of sharing and inspiration goes on there, but you kind of lead the charge with what you're doing at PowerShifter in the AI space in that community. And so I want to dig into that because, shit, who else is talking about AI today? Nobody. So let's be the ones, right?
JP Holecka (01:17.07)
Well, you know, maybe we can find some signal in the noise from the quality of the content. How's that?
Steve / Agency Outsight (01:22.874)
That's the goal. That's the goal. so give kind of, mean, listen, you've, you've been doing what you're doing for ages. When did it shift over from we do content, we do design to let's build things using these new tools. What was that kind of process like for you?
JP Holecka (01:39.501)
Well, the, chat GPT epiphany was very soon after it was launched, to the public for me. and what I quickly realized as a person who's dyslexic and has ADHD was that even in the form of the, when it launched, it was a complete, epiphany.
as to the things that I could now do. My vision and experience with the LLM GPT was very different, I think, than most gravitated to we can make content with a prompt. And I experienced it very differently and saw me, or what I was using it for was data analysis for plain text data.
you know, natural language data analysis for our research, for our user research interviews, all kinds of things in that regard. And then I realized that I've always wanted to be able to write ephemeral content, news type content, industry type content, but with the combination of my ADHD and dyslexia, I would start and end by the time I was finished, it was literally yesterday's news or maybe five days old.
I never had a huge desire to write a lot of evergreen content on my own. have over the years done a lot of it with writers, back and forth, take several weeks and, know, get successful content up and running. But now I realized that I could record my articles out loud, transcribe them. I trained the GPT very early on, on the actual writing style that I did have that I had done with many years, evergreen content. And I was able to get content out.
very quickly my content, not the prompt and hey, make me something. But an editor that edited my words, which was me, the editor. So it was this weird mirror with that change E P T. And I just kind of went from there. I went really all in on it and started to generate my own content. And it was the same time I was working to create my own personal brand, I guess you could say on LinkedIn. And I had
Steve / Agency Outsight (03:38.8)
Mm-hmm.
JP Holecka (04:04.174)
had just, taken some training for several months of training on what I should create. And then GPT and other LLMs felt meant that I could actually execute on this campaign. And so that was really for me, the aha moment. then I just, there's some scant research, I think that shows this dyslexia, people with dyslexia and prompts. There's a core, some kind of correlation that we seem to be able to achieve things, and prompt very well.
I got, I got a lot of great results early on and, we just started to apply it internally, through our, know, through our, well, my shared excitement as the leader of the agency. And then we turned it into a bit more formal, where we were tasking people within the agency to start to try test and explore. And then we had show and tell days early in the.
So every Friday we'd have a rotation. Someone would have to show something they did with some form of generative AI. It could be super simple. One fact, one of our PMs first used it to do a bunch of comparison on refrigerators where they were doing a renovation and she showed how that worked. But that was her aha moment because she had no contextual relevant way to explore AI or she couldn't figure out how to make it, how to make it work with the things that were
work related, when it came to something personal, she made that connection and then from there, that's how she ran with it. after that, we realized that finding ways in which to use the generative AI, you know, it doesn't always have to be work related. If someone's got something else, like how do you unlock the way in which they can connect its power with the things they need to do? We just kept going further and further on that front when it comes to the operationalizing the agency.
Steve / Agency Outsight (05:55.257)
Interesting. Yeah. It's interesting because you're talking about like this content plan that you were trained on and then here's here the tools that are going to help you execute it. And so kind of, you know, if you take that into just a everyday agency rollout, let's let the human build a strategy and then let's use the tools for some of the execution work. Not all of it, right? To your point, it's you recording your voice because it's easier to do that than to go typing it.
JP Holecka (06:17.811)
Mm-hmm.
Steve / Agency Outsight (06:23.284)
I've built prompts that's that interview me. And so I fire up, Riverside and I have GPT running and it asks me questions and I'm repeating it back and then it converts it into content. Probably very similar to what you're talking about.
JP Holecka (06:37.317)
excellent. See, that's a great tip. I go unprompted. And so that's, that's great. Like, and so using the voice and the actual audible.
Steve / Agency Outsight (06:46.882)
It knows what I talk about. knows, you know, all of my phone calls are dumped into it from my recorder, all of my client contracts and client notes. And so it knows the regular challenges that my clients face. And so it prompts me like, Hey, what's the challenge that one of your clients, faced this week? And I'll say, you know, finding profitability in an organic SEO client. And then the GPT will, you know, turn on it's Diane Sawyer and start interviewing me. then.
I've got all this video on here so it's really strong.
JP Holecka (07:16.087)
That's I'm making a note anyways.
Steve / Agency Outsight (07:20.132)
Go, you can listen back to this. all the notes you need in the audio when this comes out. But to your point, so that was kind of your first foray into it. Did you guys do any sort of like formal road mapping of like, here's what I see the agency starting to do, or are you just kind of like, as things pop up?
JP Holecka (07:22.462)
Yeah, yeah.
JP Holecka (07:37.742)
I think like many, like many, you know, there's the, that, tried and true expression building the plane while you're trying to fly it. Well, that was like exponentially more relevant when it came to adopting AI internally. But I've led the charge for the longest time. would say for the first year of the agency that was really me doing the exploring and coming back and saying, we want to do this. We want to do that. And then as I started to do public speaking.
for how other agencies could do it. The roadmap came out of creating those decks, honestly, for how to formalize it, to present it back to others as in how I've formalized it. I have a Miro board that I have shared with the Bureau and I'm happy to share it after when you can put it in a show note links. It has about 65 % of the ways from marketing all the way through to client support.
of a post product launch or project launch, how we're using generative AI in the mix. So from content generation to contract analysis to RFP analysis to generating SOWs now and generating proposals now. We've gone proposals for things like web flow, which are pretty known quantities when doing a web flow project. There's really not a lot of surprises in a web flow project. It's just really the scale.
ultimately and the amount of content, but it's pretty straightforward and through feeding it previous proposals and also the LLMs are getting pretty smart as they continue to scrape. They're pulling in data, they're pulling in budgets and estimates and proposals and retraining. So we've found that the Webflow estimations are 90, 85, 85 % accurate on average.
and because we've got a trained GPT, the proposals have a repeated format. we're applying it to not just the content, but also analysis and the generation of, of, of documents. And, it's sped up the process, hugely. Yeah.
Steve / Agency Outsight (09:47.601)
I'm sure, yeah. mean, some shops are leaning into full platforms to manage their agencies. This episode, for example, is sponsored by Ignition and not just throw a plugin for them, but also they have SOWs built into their platform. so that's one step in optimizing process, going from a Google Doc or an Excel sheet or a PowerPoint deck or some...
I've got to recreate the wheel. mean, shit, I still know designers are using InDesign documents for their proposals. And so.
JP Holecka (10:18.867)
Yes, I just ran into that last year and I was, it was a design centric agency and branding agency. And so I, I understood that to a degree, but in order to survive in this economy, the deal flow has to be high. And in order to, address and get those, it's a numbers game. And so if you can increase the input, you're going to need to move faster on the proposals.
And I've been doing this for, I've been doing this for 30 years in various forms or another. And one thing I have learned is that we spend an agonizing amount of time in the early days, bespokely making all of the proposals and anguish agonizing over them. But when I go and part of the exercise was when we started to do the training and going back over the proposals, I realized they're really not very different. And, and, and the effort.
Steve / Agency Outsight (10:48.538)
Agreed.
JP Holecka (11:18.093)
And then what we do in software and websites and whatnot, there's not a lot of subjectivity of what the outcomes are. The outcomes are going to be what they are and we can, you know, control that. But the subjectivity in it's not like a brand, I'm not doing campaigns that are very specific and subject. So I realized that we need to just more proposals out the better. And that's how you're to win.
If you've, that's what I mean. Yeah. Quicker and the more of them out, um, and, and, and, an economy that's suffering and has been for the last several years, and it's been devastating for a lot of agencies. The it's, it's a, it's purely numbers game. Um, and you got to remove the subjectivity and the agonizing over this and that. Cause ultimately I find most clients are, they come back and whether you want it or not, it's, um, you're qualified and you're the right price. And that has been ultimately.
Steve / Agency Outsight (11:47.256)
and quicker. Yeah.
JP Holecka (12:15.18)
You're qualified, you're trusted, we vetted you, and you're the right price. And a lot of times you're not the right price. So if that's the case, we're not gonna race to the bottom, but we're gonna keep the numbers going and increase the proposals that could need to go out the door. Have I resolved that riddle on the inbound? Not yet, but we're getting there. We're working on some things and our outbound has now starting to generate meetings as of this week from a campaign that started two weeks ago. So we'll see.
Steve / Agency Outsight (12:40.89)
Very cool. I want to dig into that, but I'm curious as a follow up on AI helping with pricing on like Webflow. It's very black and white and formulaic. Are they able to sense buyer's intent on like additional value you can add or, you know, we formalized it to be a $10,000 project. I think you can ask 15 K or like anything like that built into it.
JP Holecka (13:02.412)
Not yet, but what we've done in the last few weeks is, well, a few things, and we're going to automate all of this together. So when we take a first discovery call, you know, it's pretty standard now, you send them a note of a summary. It's quite often it's a generic summary, but you know, it means that you heard them and you sent them something back. Now that's just table stakes for any agency. So I thought, how can we change that? So I redid it. So I did a prompt that said what we heard.
So I created a what we heard on the call. It's a lot softer. It's specific. It's suggestive. If there's it makes no it's not a sumped of in any way, right? You know, I'm still trying to build trust. Did we hear it correctly? Basically did we hear it correctly? And this response every single time we send that email out is oh god, that's amazing. Yes, you totally heard me just reframing the actual summary to be like that. So then I thought how can I
Steve / Agency Outsight (13:36.016)
Mm-hmm.
JP Holecka (13:59.137)
get one transcription from that discovery call to do multiple things. So we have a what we heard that comes from the lead on the call. And then we created a new one, which is what they said and what they meant. So it's listening. It's listening to the call. it's essentially looking at it from a client services perspective. And what are their real needs? What are they asking for? What is their sentiment? What are their potential challenges? And it kind of builds a little profile there.
It's for those that were not on the call. One of the things I find the most difficult to do is communicate what is the prospective client really like in between the lines and trying to find the time to do that. This now goes with to the team to give it more of an analysis. And then we do a third breakdown is it either writes it as an RFP style, not a brief, but more writes it as an RFP.
Steve / Agency Outsight (14:28.208)
Mm-hmm.
JP Holecka (14:54.452)
Or if it's smaller, can write it as a brief so that I've got three or four things all coming out of the first transcript. So yeah, but the analysis of the client's intent, and I actually shared it with the agency, I shared it with the Bureau, the prompt, and I made a GPT, that custom prompt as well, that does it. And I should ask them see how it's working for others, but it's been pretty cool for us so far.
Steve / Agency Outsight (14:59.994)
Love that, yeah. I think these.
Steve / Agency Outsight (15:18.276)
Yeah, yeah, I love that. think the core ability to record calls and then get some, just break down for people that weren't on the call or feed it into a CRM or like all these different things that can be done with it is great for account management, for biz dev, for just interpreting like what the intent is and direction you can take things and you can train them to understand like is there room for upsell, know, was there indication that there's other challenges?
that we can answer for down the road and things like that. And I just think it's brilliant from an account management standpoint.
JP Holecka (15:50.103)
Yeah. And that's our next step to, for account management. We're to start to take every single call, that we have with them, whether it's projects or otherwise related. And those will go into a central, at the moment it'll be in a G drive for, for, for, you know, and we can have Gemini actually use that every time something's added to a G drive folder, it'll actually go through and do some stuff with some of the automation that we've got, to start to look at account analysis. We're a small agency.
We've gotten smaller in the last few years, just from the economy being what it is. And so we're working with fewer folks and we're also looking to not grow immensely, not from a revenue, but from a people perspective. We're going to see how lean we can keep things so that we're more nimble and more able to adjust to the economy.
We had to do several rounds of layoffs and it's never easy for those being laid off. Of course, it's the worst, but also the impact to the agency from institutional knowledge to the emotional challenges that go with, you So we were going to stay smaller, more of a Hollywood model, but so we're going to have some of these tools that will assist the teams to be able to punch a little higher above their weight until things settle in. Ultimately it is to grow, but while things are still pretty
undetermined trajectory for the next 18 months at least we want to stay nimble.
Steve / Agency Outsight (17:21.786)
Yeah, I think it's the right model. Yeah, laying off sucks for everybody, especially the families with those impacted. But if you can grow from a top line and additional value to existing clients without adding more headcount, pulling in resources, freelancers, and building more robots as needed, like it's a beautiful model and why not keep going in that direction.
JP Holecka (17:41.61)
Yeah, we know we've, and we're going to rely on our trusted network of freelancers. We have been in this business for a long time and we've had a lot of freelancers that have been working with us for 10 plus years and always answer the call when we call them. And the nice thing about having a great contractor network is that you can rely on repeatedly is that when they step into a project, they already know your culture. They know how you work.
And your trust, the trust in them is huge and they're not, they don't feel just like a hired gun. And that's the other thing, which you don't want.
Steve / Agency Outsight (18:15.652)
Yeah. Right. So you mentioned, kind of reverted back off track, you mentioned kind of an automated AI driven outbound campaign that's starting to produce results. Can you talk about kind of some of the, I guess, tools of trade that you've been using and what that's been looking like?
JP Holecka (18:33.467)
We actually hired an agency, so it's not AI. They are probably using AI. In fact, interestingly enough, the agency did use a custom AI chatbot for onboarding, not a person. So I was given, it was custom about all the things that we were trying to achieve. It was an interesting take. Is it better than a human?
Steve / Agency Outsight (18:38.063)
Okay.
Steve / Agency Outsight (18:50.0)
Interesting.
JP Holecka (19:02.539)
I think a human, if I would have changed one thing, I would have probably had a human introduce it on a call in that as opposed to getting a first thing that I got. But I will say the summary and everything that happened right after I finished hitting that forms, like minutes later, stuff started to come in that was ready to be actioned and it was parsed off to the rest of their team to action. So I do think that the bot is an interesting approach. We'll see the campaign.
I've been, I was working for them for the month and a half building up all of the, you know, ideal customer profiles and, and our offering and all that good stuff. And then they started the outbound emails last week. I have had success in outbound previously with another agency in LinkedIn outbound, which had a really high, an extremely high, uptick when it came to those accepting invites.
This is a teach to fish. So, we're, this is a one time shot with this agency. So they're teaching us the process of, of doing out bounds with not just the emails, but everything that you need to, to, and now I'm a calendar. I've got, went out, started last Monday. We've got three meetings this week. So that is a good sign.
Steve / Agency Outsight (20:17.935)
Awesome. I hope they're the right meetings. A lot of times there's a meeting for the sake of meetings, but I love that you're approaching it from what can we learn and then adopt and bring in-house and kind of automate and run off on our own. So the fact that you're going into that and the other agency is accepting that that's how we're approaching the engagement is beautiful.
JP Holecka (20:20.105)
Yes.
JP Holecka (20:30.347)
Yeah
JP Holecka (20:35.787)
Well, that's how they positioned it. They positioned it too, which is a teach to fish, which is an interesting model for an agency. We teach our clients to fish, but our projects are substantially bigger than an outbound campaign one-off would be. So I'm, don't know the longevity of something like that. mean, you, if you've got to do a lot of, I mean, they're in the outbound business, but there's ultimately you do run out at some point and it gets thin. If you're, know,
Every more, every agency strategist I've ever met has said that real growth from comes from account growth. So, you know, and so it'd be interesting model for them. We have, you know, we've grown big, large accounts, but the challenge we have quite often being in the software product space is once the project's done, they can be six or seven figures. There's really no work for you.
in the software space. And so it is, I do find those same consultants quite often don't totally understand the software space and that is predominantly the way it goes. So you make a lot of money upfront in the first 18 months, it can be, but you are quite often not the ones to continually build that. When you're building software, their ultimate goal for most...
enterprises to bring it in-house and run it themselves. So that's challenge we faced for nearly 20 years.
Steve / Agency Outsight (22:03.761)
So here's a question about that. I just literally an hour ago had a conversation with another AI integrator agency in Texas where what we were talking about is what is that full lifecycle of the software? And is it that much of a cliff at the end of delivery where AI is changing so quickly now that iterations would be integrating new models, new platforms, new systems, things like that. So the roadmap and that lifecycle
is getting extended out even further than a traditional, here's your software, go run with it, put it on your internal platform, right? I mean, isn't there more to it with such rapidly changing software?
JP Holecka (22:42.603)
I think hype, like the hypothesis is good. have not experienced that firsthand just yet. Um, a lot of the, um, AI work that we've done have been proof of concepts. Um, cause the enterprises are still trying to figure out where it fits in or can we do this and hasn't been a huge commitment. It'd be interesting to see where we go. I feel like the, with any hype cycle, you know, you've got your, you know, we're, we're, we are.
Steve / Agency Outsight (22:55.579)
Okay.
JP Holecka (23:11.647)
We are in the trough of disillusionment for sure. Like not just toe in, we're pretty well into it. Where the enlightenment is going to hit, I'm not sure, but I do think we're gonna see a few things with every cycle that we get. Would you like a side of AI with that? We'll be going away soon. Just people are think gonna be really asking for specifics now that they know what generative AI can really do and what it can't do.
So I think we're going to see a contraction. I think we're to see a contraction in the AI space to a degree. I think that all the models are hitting limits. in 2024, with exception to this year, big jump with Google's Nano Banana, and that was a big jump. But ultimately, all the image models had really topped out in 2024. They cannot get past a certain amount of slop.
in the images or things they just can't render. And we're seeing the release numbers now stagnate for like nine months, 10 months between releases of new models. We've seen that with GPTs are still 20 % error rate or imagination rate or whatever. That number is not changing. It's like fixed. And so when you're running software, it has to have a 99.9 %
Steve / Agency Outsight (24:23.249)
Interesting.
JP Holecka (24:40.616)
you know, accuracy rate and all the software we've done in the past has been that way. So I think we're going to see a bit of this elastic kind of retracting a bit and realizing it's not the Holy Grail and AGI is a fantasy how it's being defined by the CEOs today. But it's going to find its place. And I feel this is the year where it is finding its place. It's finding its groove as a tool. There is definitely attrition that's going to happen. But I think as my son
says and his friends have told me it's a very small sample size. We are not going to be spending 40 or 50 years prompting. We are going to live a full life like you did. We are going to live a life in which we learn and explore firsthand. We are going to live a life where we make mistakes and we're not going to live a life where we're sitting in front of a thing telling it what to do. And that's small sample size and my kids, my friends kids are creative, but
when you look at things at a macro level, what you know, the CEOs and that can't predict in the future is can't predict is that they can't really predict much at a macro level, honestly, because there's so many factors that we just can't take in. then once the momentum hits, it's too late to stop it. that gets, could see if there's a, you you keep unemployment long enough and enough blame goes to the fact that it's for it from AI.
You know, they are ground swells and then employers have to adjust. It's all of a sudden you've got seniors retiring and there's no juniors in them, because AI has replaced them. If there's going to be a continued 20 plus percent error rate in the outputs of AI and there's no solve on the horizon for it, those seniors are gone. Who's going to be overlooking.
I had, you know, we're doing a lot of AI. I'm not trying to come down on it. I'm just trying to say like, there's all these, there's a lot of magic here, but there's also, let's like, where do we draw the line? I had my very first proposal go out the other day, which had an imagination, which I had an hallucination in it. I was at a conference and didn't have enough time to spend looking through the proposal. And it took a new, I took a relatively new acronym that is for, I can't recall, but it's for large language models, SEO.
JP Holecka (26:52.17)
And it made three different versions up of what the acronym was. And I missed it in here and my team, because it's such a new acronym, my team didn't really even think to look for it and thought that's what it meant. And then the client sent it back and said, I think there's an error here. Luckily, good relationship. They know we're doing AI and knew I was at a conference and didn't get to check it, but that could have been devastating for a prospect. And, um, if we had full automation and without QA, human QA is that know the subject matter.
Steve / Agency Outsight (27:14.512)
Wow, yeah.
JP Holecka (27:20.938)
You know, we could have done real damage to say you're using AI and it's I just got some slop in the very first proposal that you put forward for us. Um, so that's a no-go. So I'm not sure where we're all headed with it. Um, it's definitely, um, um, making things, um, uh, faster, better and more cost effective, but it's also being heavily subsidized right now by tons of investment. So may go up. mean, I'm Mike, Mike Gemini ultra is 200 some odd dollars. Um,
I'm spending more on tools for myself than I have spent on software ever. And that also, we have to justify those costs as well.
Steve / Agency Outsight (28:00.134)
Yeah, it's an interesting intrusive. Excuse me. It's an interesting perspective that you have or I guess even prediction of like what the next year could look like. It's probably the first real seemingly realistic perspective take I've I've heard of like maybe we have kind of leveled off on some sense. There will be some contraction. I love the sound of that. It's not something I hear very often, but it's interesting you address kind of the junior bench side of things.
JP Holecka (28:28.254)
Mm-hmm.
Steve / Agency Outsight (28:28.497)
I'm speaking to a college tomorrow and it's one of the things I'm talking about is, kind of you guys are facing like a serious issue of not a lot of shops are hiring for the junior bench right now, but there's going to be a massive deficit in the next three to five years as the senior team, you know, attrition's out. I'm curious if you've got, you you've got younger kids, if you have any, I guess, sage words of wisdom of like, how can they differentiate? How can they show up differently in the marketplace?
show their value, things like that.
JP Holecka (28:59.71)
I think they have to understand prompting. It's going to, think like anything, you know, when you see a CV for the, you know, for the last 30 years, it said it knows how to do, you know, you can do use Microsoft Word. You can use Photoshop. You know, I can, I, my prompting level is that, know, whatever, whatever, however we want to measure it, it's going to be a part of the mix and you have to understand how LLMs work and, and, and how diffusion models work. But I think at the end of the day, you still need to be the subject matter expert.
at whatever level you are. you're a junior, you have to be the best junior and know the best stuff that you can at that level to be able to and not reliant on. And I think we're going to start to see that and feel that of juniors without any oversight, working the models, they still need to have mentorship. And I think it's going to take some bigger thinking and, and not the obvious bottom line, right?
So what is the real bottom line? We get false economies sold into us all the time because it's moving so fast and quarter to quarter thinking. is the, is the, is the potential downside two years for those that don't think bigger and just look for ways to increase margin through just a fast adoption of AI? they getting themselves into a financial problem downstream? think those that hedge it.
Steve / Agency Outsight (30:04.017)
Mm-hmm.
Steve / Agency Outsight (30:26.065)
Mm-hmm.
JP Holecka (30:27.505)
and say it's a tool and it's in the mix. And we're going to continue to be, lately the charge with humans first, creativity first, those that are passionate, that that have those that have a point of view, those that can interact with clients, and do real, real thinking, real thinking. I think that those agencies ultimately will probably fare better than those that went all in and went with margins first and only that.
Steve / Agency Outsight (30:57.765)
Yeah, I love that and I appreciate that. On that note, because you said margins first, so I'm going to switch over to a couple of rapid fire questions for you. That's a great perspective that you just delivered though, because I absolutely love that human first. So first is coffee, tea or something stronger?
JP Holecka (31:07.763)
Sure.
JP Holecka (31:17.306)
Tea in my big boy cup
Steve / Agency Outsight (31:19.131)
There you go, Papa Bear, I love it. What's one thing people would be surprised to learn about you?
JP Holecka (31:24.701)
Surprised to learn about me. I was a competitive fencer and I also had a pole vaulter in grade eight because we had a my science teacher was an Olympic coach. So I did two real oddball sports in high school.
Steve / Agency Outsight (31:36.955)
Couché, I love it. What's one piece of advice you'd give your younger self before starting your agency?
JP Holecka (31:43.421)
Don't be so afraid.
Steve / Agency Outsight (31:46.033)
Say more about that.
JP Holecka (31:47.145)
Don't be so afraid of things or what people think about you or be, you know, I'm a pretty sensitive person overall, just the way my brain's wired, but I let a lot of things get under my skin for a long time. If it didn't work out or someone said negative things about me and it's taken, I'm 57, it's taken a long time to get to a place where I don't let that bother me as much. I'm not going to say it's gone zero, but I can move off of things a lot quicker. And I wish that that had been sooner. I spent a lot of time.
worrying about what other people thought. And then when I stopped worrying and started doing what I wanted, ultimately a lot of better things came out of that.
Steve / Agency Outsight (32:26.447)
Yeah, I love that man. Good stuff. JP, I am grateful for all that you shared with us today, your experience on all levels. I'll watch for the show notes for people to check out your Mirrorboard and PowerShifter. So thank you very much for joining us.
JP Holecka (32:40.104)
Yeah.
JP Holecka (32:44.552)
My pleasure.