How SEO Grew Up With Cameron LiButti

We get real about what AI search changes and what it does not, then map SEO back to fundamentals like attribution, buyer intent, and revenue. We also share how agency teams can use agent workflows and governance to move faster without turning the business into chaos.
• Matt’s return, agency ownership changes, and why the timing matters
• Cameron’s path from engineering to SEO through referrals and Google Business Profile wins
• AI search as a conversation starter with CEOs while fundamentals stay critical
• Why brand traffic and last-click attribution mislead decision makers
• Cutting low-value traffic through pruning to drive more calls and leads
• Agent harness basics using folders, instructions, APIs, and automation
• Getting teams into IDE workflows, avoiding chat-only memory limits, and protecting .env keys
• Data privacy for regulated industries, self-hosted models, and AI governance policies

Guest Contact Information: 

Website: www.bidviewmarketing.com
LinkedIn: www.linkedin.com/cameron-libutti

More from EWR and Matthew:

Leave us a review wherever you listen: Spotify, Apple Podcasts, or Amazon Podcast

Free SEO Consultation: www.ewrdigital.com/discovery-call

With over 5 million downloads, The Best SEO Podcast has been the go-to show for digital marketers, business owners, and entrepreneurs wanting real-world strategies to grow online. 

Now, host Matthew Bertram — creator of the LLM Visibility Stack™, and Lead Strategist at EWR Digital — takes the conversation beyond traditional SEO into the AI era of discoverability. 

Each week, Matthew dives into the tactics, frameworks, and insights that matter most in a world where search engines, large language models, and answer engines are reshaping how people find, trust, and choose businesses. From SEO and AI-driven marketing to executive-level growth strategy, you’ll hear expert interviews, deep-dive discussions, and actionable strategies to help you stay ahead of the curve. 

Find more episodes here: 

youtube.com/@BestSEOPodcast

bestseopodcast.com

bestseopodcast.buzzsprout.com

Follow us on:

Facebook: @bestseopodcast
Instagram: @thebestseopodcast
Tiktok: @bestseopodcast
LinkedIn: @bestseopodcast

Connect With Matthew Bertram: 

Website: www.matthewbertram.com

Instagram: @matt_bertram_live

LinkedIn: @mattbertramlive

Powered by: ewrdigital.com

Transcript

Cold Open And Big Promise

SPEAKER_00 0:03

This is the unknown secrets of internet marketing. Your insider guide to the strategies top marketers use to crush the competition. Ready to unlock your business full potential. Let's get started.

Matt Returns And Owns The Agency

SPEAKER_01 0:17

Howdy, welcome back to another Funfold episode of The Unknown Secrets of Internet Marketing. I am your host, Matt Bertram. I am really back. I actually had some backlogs for everybody that was listening podcasts, and I was going through an accelerated MBA program through Goldman Sachs. And uh, you know, that towards the end there was eating up a lot of my time. But uh I am back in the saddle, if you will. I also, for anybody listening and been kind of following what I'm doing, I did buy out my other partners. So I now fully own EWR Digital, and there's a lot of things that we're gonna be uh talking about and doing. But uh thank you all so much for the support. Uh Chris and I are actually having lunch after this podcast. So we're gonna go meet up. We're still on good terms, still a client. Um, but uh just kind of want to let everybody know the big step change happening. And and I thought it'd be a great time. I know I've done this periodically, bring on other agency owners to talk about um kind of where they're at and kind of benchmark what's going on and how things are changing. There's actually a big agency conference coming up that I'm gonna be going to. There's SEO week uh next week that I just don't have time for, uh, but went last year, uh, a lot of good guys there. Um, Cameron, welcome to the show, buddy.

SPEAKER_03 1:33

Thanks, Matt.

Cameron’s Accidental Agency Origin

SPEAKER_01 1:34

Hey, like, you know, I I love having these conversations, kind of like a fireside chat, letting everybody kind of see what's going on. There's just so much noise out there. There's so much that's happening. Um, it kind of makes your head spin, like new workflows, new tools. It's constantly pivoting, uh, new models coming out from an AI standpoint of like which ones are the best, what to use, like uh, you know, like cloud design just came out. Like, there's like it's very hard to build a business with everything constantly changing. Um, but it's fun. It's it's fun. So uh why don't you tell the audience a little bit, just kind of about your background as we kind of get into conversation, just kind of where you're coming from.

SPEAKER_03 2:17

Uh yeah, quick, quick background on me. I was actually an engineer. The marketing digital stuff was accidental. Uh, just like starting the agency was accidental. Um, I engineering consultant for right out of college, mechanical engineering. And then I started helping a buddy back in 2013. And we were doing that, you know, the thing back then that worked was we were selling supplements like crazy online. We were having fun. Uh we got our hands slapped around 2016 over the remarketing lists, which if you have PPC people listening, they probably remember those days. Um 2019, I was talking to a doctor who was a friend, and he kind of jokingly said, You talk about marketing differently. And I said, I well, I have nothing to sell you. And I'm just like a purely like a data guy who does this for myself. And he said, Well, what if you helped me? And I was like, I don't know. So, anyway, I helped him, and my name became well known in that medical industry, and we've been partners with the uh it's a national group of doctors ever since. And honestly, we've never done our own marketing. Uh, the podcast circuit for this year is new for us because we've just gotten so many referrals for the last seven years that it's just been all we can handle. So um, that's the the short of it. The SEO piece was probably the most natural piece for me, being that I was an engineer, it just made sense back in that 2015-16 when we were transitioning from you know manual algorithm updates by humans into that machine. I refer to AI, I'm still it's a bad habit, but as machine learning, because that's how I learned it, right? Uh TensorFlow on Google was my first introduction. I don't know if you ever played with that, but um, that's how I got into AI was teaching myself about you know all the probabilities inside of TensorFlow. Um, and so yeah, it was the most natural thing. And then I helped a doctor, and it was just purely SEO. I remember he uh he sent me a picture. We started working with him in July. He sent me a picture and he sells hearing aids, sent me a picture in December. They had this pegboard, and he said it was the first time in 30 years that the pegboard was full of devices waiting to be picked up. And I'm this was 2019, so I mean, spoiler, it was 98% Google business optimization. He had competitors like dropping fake profiles on his on his location to filter him out. Uh it was just a lot of like cleanup stuff, and then obviously I've evolved from there. Um, we do well web dev. Uh, I'm not I always joke, like I'm an you're hiring the engineer. I have people who are good designers and stuff, but we couldn't early in the early days. I was always trying to partner with developers, and it was never going well. It's like this isn't how you build a website, this doesn't, you know, and they just didn't understand the the marin of SEO and good web dev, good storytelling, um, good branding. So we brought that all in house, and that's what we've been mainly doing those two pieces for the last six and a half, almost seven years. Uh obviously we have to do PPC now, which we are getting really great results the last two years. I'd say there was a period in like that 20 to 22 where PPC was kind of in Google seems to have really dialed that in. So we're into that. Um, and then now we're we have to be in AI now, right? We don't really have a choice at this point. We have to know about it and understand it and be ready. Yeah. That's the quick view of who I am.

AI Search Hype Versus SEO Basics

SPEAKER_01 5:46

Yeah, no, I I think um coming at it from an engineering standpoint or even like a system thinking standpoint is really how how to approach it. Um, you know, quick background for anybody that's just coming on and listening. Uh we we had kind of the opposite. So we were doing uh web development, and then we and then people quickly said, well, how do we get people to come here and buy from us? And so so that that led into SEO. I actually got introduced to the agency. Um, I had uh sold a couple businesses and the podcast was around the corner from my house. And so uh I I wanted to go meet Chris and and check it out. And and I was really like a paid ads guy. Uh and the infrastructure of understanding paid ads helped me kind of understand what was going on with SEO. And that was really before they kind of lifted the curtain up and uh showed you everything that's happening, and now everything is out there to learn it in, I guess, the world of AI. And and now we're really deep into the engineering and the math. And like I feel like SEO has kind of grown up, if you will, from what it used to be from kind of like content, uh like keyword optimization, like there's a lot more that that goes into it uh these days. And uh it's really exciting.

SPEAKER_03 7:06

I see a lot of SEOs complaining, I don't know, where Facebook groups and stuff about how they're attracting all this attention because of AI. I as somebody who owns a business, I'm curious if you think this too, but as somebody who owns a business, I like that these CEOs and these business owners are aware of it now because now it gives me like a starting point to have a conversation with them and explain like, yes, AI search is real, but that doesn't mean the fundamentals got thrown out the window in the last 12 months. They're I would argue more important than ever. And so I've I've really leaned into that conversation. But you are, you know, I've got I can't tell you how many times I get emailed forward from one of our clients like, hey, they said our client across the street is showing up in AI search more than us. Like, it's just a really good, it's a really good cold email. Like, I I I'm not gonna send it, but if I was doing cold email, I'd probably send it too. We we had one yesterday, she pulled it up. She's like, they're saying the the business across the street has more. And I was took all of 10 seconds to just compare her on AHRAFs. Like, this is your data, this is her data. Please tell me where you're failing. And she goes, Oh, oh, yeah, but completely understand that that's a lead gen email.

SPEAKER_01 8:18

Yeah, no, there, there, there's definitely uh a lot of that uh happening. Um, I I've actually seen a lot more negative SEO uh happen, even though Google said that's not a thing. Uh, I believe it's a thing. Uh I know it's a thing. I have some clients, it's a thing. Um, you know, I I think that you're right. The conversation's been elevated. I think a lot of times that conversation with the C-suite is because paid ads can show a clear ROI. Or, you know, uh now attribution, I think it's certainly debatable on, you know, if you're running uh a person's name, right, or the brand name, especially when you talk about the affiliates. Like, so so quick, quick story, like where we can maybe uh really connect on it. Chris, Chris has actually a big supplement company that we helped grow over the last couple of years. And and essentially we started getting into the affiliate world, uh, and we were more on the agency side and we started getting into the affiliate world. And literally, like we hired four different companies, and it was all straight like scams, basically, uh, where they were just like um, you know, they're they're cannibalizing the search volume we already generated and then like attributing that to them. But I think that a lot of that conversation, you know, their last click attribution is really a thing, and and CEOs could understand that clear delineation that came from this platform, and hey, like this is what works. SEOs haven't done a good job explaining what they're doing, but now people are starting to ask questions. And so I think that there's gonna be a lot more attention. I think there's gonna be a lot more um money that moves towards SEO. I think that uh SEOs are gonna get paid more for what they have to be able to do and what it's encompassing. And I think that, you know, SEO is growing into marketing and marketing and SEO are almost kind of like uh the architecture's been woven across the board together, right? Because you talk about social, you talk about paid, like it's all affecting everything, right? And then, you know, you add this AI layer to what's happening. And I mean, you know, I've been saying this for over a year now, like it's coming. Like, whether you want to debate it or not, like whether you want to like put your head in the sand, it's coming and you got to understand it. And and the more you like treat it like a very, very intelligent person, it helps you do SEO better across the board, it helps you find things, it helps you see things. And and so we've really leaned into whatever you want to call it AI, SEO, GEO, AEO, what like everybody's debating over that. I'm just like, let's ship.

Attribution And The Brand Traffic Trap

SPEAKER_03 10:59

I don't care what you call it, we got a win. Um you talked about attribution. Um, it's uh so uh harder. So when I'm working, so we have a mix of clients with a lot of doctors, lawyers, engineer types, smart, very intelligent people. Not they're not like they're just not their specialty, is not SEO. Um, and then we work, we have a handful of nine-figure brands that we just do SEO for. Those conversations with the nine-figure brands are interesting because you're talking to a marketing director that does understand SEO. Um, last year we had a nine-figure brand contract with them, and he said we were the first agency to talk to about not caring about the brand. And he said for years, SEO companies would show up and say, Hey, well, look at all this traffic you were getting. And he was like, Well, I'm smart enough to know that 50% of that is brand traffic, whatever the number is exactly, a majority of it. And I was the first one to be like, Well, you're failing here, you're failing here, you're failing here. He's this is where we're gonna go win. He's like, You're that to me to show up and talk about attributions, like these are where the buying terms are. Um, that's a conversation I like to have a lot with owners. We've had a lot, and then on the local side, we've had a lot of clients come to us and say, I'm getting, you know, 4,000, 5,000 visitors a month. I'm like, your market can't support that much traffic. You're just ranking for blog stuff. You're getting, you're in Chicago and you're getting, you know, traffic from all over the United States. Not that that's bad, but that's not where your buyers are coming from. And there's been a handful of clients where we have cut their traffic in half by pruning and being more specific about their topical map, and their phones just take off, right? They're like, my phone is ringing all the time, and they'll see their traffic report. They'll be like, we're down 20%. What's going on? Like, because I don't it's not that I don't care, but I I'm not after that traffic. I want the I always talk about like the awareness level, right? I want the people that are most aware, ready to buy, that don't know your name yet. That's like an effective for me at the local level, that's the most effective SEO strategy. Again, not to say that there isn't informational search isn't important, important, but if we're gonna get cannibalized by AI search, which we can show up in those informational searches, get credit for them, but we know people aren't converting through.

SPEAKER_01 13:18

Um yeah, yeah, yeah. I I think that it's just forcing the conversation to be what needed to always happen, right? Like you're like if your KPIs are traffic and you want to grow traffic, and you're you're you're ranking for all these blog terms that you're ranking for 5,000 blog terms, and like you know, you're a local small business. Like, does that help you? Like, if that's the metric you're going, oh, we're our traffic's going down, right? Like, who cares about your traffic, right? Like, like I'm like, what are the events happening in GA4? Like, what are the events you have set up? Like, let's get to the buying signals, like you said. And and I think one of the best things that happened was when they started breaking apart content into different categories, and then they just started throwing informational search out there in Google and say, Hey, like you're gonna show up for this, it's gonna build your brand, that's great. But that's not the goal of a lot of businesses. And and there, there a lot of people have posted charts on social media that their brand traffic's going way down, but the revenue, which we should always be talking about, that's the conversation with the CEOs, is going up. And so it's kind of like a moot point, uh, you know, it's just like we're we're driving more revenue. Let's look at different numbers. Like you've been trained to look at things that are not useful. Um and so I think that that aligns the conversation better, and now people are starting to understand it more. And there's that kind of level of curiosity uh um uh to AI. And, you know, I mean, there's a lot of like very curious CEOs out there about AI. And the conversation for us is more moving into um, you know, how is it moving into the workflow? How is it moving outside of marketing? And and I think marketers in a really strong position because we were impacted first, you know, now legal review, you got HR, you you got all these other areas that AI is dipping into, but it it impacted marketing and knowledge work first. And so, I mean, I was even talking to some engineers and like they're like, oh yeah, we were we were doing this, we were doing that. And I was just like, I'm in this like all day long, every day, multiple windows open, like, you know, and you're you're like, oh, like I wanted to, you know, do something on a spreadsheet. And I was just like, okay, like that's cool. And I and I think that there's I think that there's gonna be a lot of people that like that if you're an engineer or you're a developer, that you you're gonna get some compound leverage in in what's happening, at least in the short term. I mean, I don't know where where where the future goes and and you see the numbers kind of straight up. Uh so I mean we'll we'll see. I I think it's you know, I think it's adoption faster than even the internet. And look what happened with that, you know.

Pruning Content To Boost Leads

SPEAKER_03 16:10

So so we're trying to build this, it might be a little outside the scope of really what you usually talk about. But if if the people listening are familiar with what I call like our agentic harnesses, but essentially it's a folder, right? This is people use this vague language. I'm coming from the engineering world, I'm sensitive to this because I think people try to do speak like this to talk over you. And when someone says like a harness, it's just a folder system with a set of instructions. And I just show all of the people this all like it's just a folder. There's an agent folder, there's a memory file, and you can make them more advanced, but at the core, that's what it is. And you set up some routines and cron jobs and some automations, but that is really what uh an agent is, and then you plug in different intelligence, and so uh I'm getting to it to my point, but I just like baseline understand what this is. So let's say you have a, I don't know, a keyword research agent that is you've given it its its AHREFs, maybe you've given it HRFs API, um, you've given it some other gets backlink data, search data. Um, you can build that agent. And when you run it, maybe you you you can interface it with it in many different ways. Let's say you're interfacing it uh through Claude. And when you interface it, it goes out and does those things. And and and so the way that my company is transitioning, as you talk about workflows, is we're trying to make a company-wide agent harness, meaning that every area of the business can plug and play with uh different intelligence. So we have like visual regression AIs where we're looking to see if a client change, right? Our clients like to have access to their site. And so we run that once a day. It just visually looks it over, compares it to the day before, it says, hey, something visually no longer makes sense. It pings a developer, right? It's just watching, it's not making changes. So we're trying to build these types of agent harnesses across the business that um, if as you have a lot of people listening here that do SEO, understand how buried in data you are all the time, how many spreadsheets and historic data, and you're trying to look at trends and how do you make better decisions. And so that's where we're implementing AI is not making the decisions, but surfacing some issues faster so that we as the experts can make those decisions quicker. So, yeah, maybe a little outside the scope, but that's a workflow that workflows it's kind of pretty much touching every area of our business at this point. Uh, the hardest one to be able to crack has been design. It just they do it, but they don't do it like if and this is dependent on your clients. Like if you want to order a Fiverr website, AIs can pretty much, they're all gonna do better than that now. But if you're I'm assuming with you, definitely with like my company, we if you're gonna order or you're gonna work with us for a website, you're not, you're gonna get a creative director, you're gonna, you know, you're going all in on what your hopes and dreams are. And so AI really is not quite there yet on the design side. It looks good to an untrained eye, but if you show that to a creative director, like, ah, that's not quite there. But other side, the technical, the data side, we are finding a lot of use cases right now.

Agent Harnesses And API Workflows

SPEAKER_01 19:11

Yeah, a lot of these AI generated websites, you can kind of tell they have a certain look to them, unless uh you you really feed in and it ingests a lot of the client data. And so like the ingesting of the models per per the client, and you know, it then the clients come in and say, Well, oh, I actually don't like the voice, I don't like uh the post on social, I don't like my website language, I need to redo it all. And I was like, Well, okay, well, AI is not gonna give us the leverage here that we were looking for. And they're like, No, I'm looking for a total revamp. And it's like, okay, well, yeah, you need a creative director, you need to do all these things. But if you're like trying to like keep going with what you got, like there, there, there's some some some really good opportunities. I I really liked actually where you were going. Like, so I mean, APIs, like they're so helpful. Like, there's you know, the secrets, there's the keys, like talk, talk for baseline for people. Basically, you're just giving it tools and access to do all this stuff. And then you can build, like you can build different skills or a way to utilize this data or orchestrate what's happening with the agents. I would love for you to build on what you said because I think that that's where the real conversation is going. Is like, okay, like if you want to build this layer that everybody's gonna talk to, you you probably need to start using GitHub, right? Like you need to start dropping some stuff in here so so people can talk. You might need to start using Claude Cowork or, you know, like Codex is really for like a power user if you're a big chat GPT person or Claude Code. Like, but the but how do you get the whole company elevated?

SPEAKER_03 20:51

I think is that next conversation where where I think it's gonna go into like so my I being an engineer, I love AI so much that on the side I do help and consult on AI. Um, doing actually a class next week with a local sales company about it. Um you brought up GitHub. So I the hardest part about AI right now is that it's you know, we all hear the 80, 20, you know, they're 20 just 80% of the work. Those people are off and running because they are creating their own ecosystem that just OPs them, right? Overpowers them. But what does it look like company wide? So we have 14 people here. How do you set standards so that everybody's not running in a million directions? I just read an article this week about somebody at Amazon. Their entire job is to find agents in Amazon. And he said they are finding every team. It's so easy to stand up an agent. I mean, I explained to you, it's literally a folder system. It's so easy to stand up an agent for a semi technical person that the company is just running wild and it's turning into just this mess of spaghetti. So to me, it's it's actually an operation, it's a comp, it's a weird personality that has to solve for this because not only do you have to be an operations person, which are people people, they're not technology people, you also need to have a technology person, an IT-like per mind. And those people usually don't cross over very well. And so the thing that I have focused on a lot in my business is what is the operational structure? What is the org structure? I think I think we're the businesses that get out ahead are gonna have org charts that mix both people and agents into them and have very clear, definitive, like this is where the agent stops, or this is where the human starts, or vice versa. Uh so you you know you just brought up GitHub. I that's the best way I know how to do it right now. You can put these folder, right? You just put a folder into GitHub and then a bit of a technical term, but I even my non-technical people have to have uh IDE. So the IDEs that we use are either anti-gravity from Google or VS Code, which anti-gravity was built on VS Code, if that matters. Um, but these are development um environments, and so people see this, non-technical people see these tools and they're like, I that is nope.

SPEAKER_01 23:14

Like inbuto, like let's spin up some Python, right? Like, there's there's all these layers from an engineering standpoint that you have to put together to utilize these things in these environments, and it's new for a lot of marketers, it's very new, and so we've pushed so my SEO, which was interesting.

SPEAKER_03 23:31

My SEO team was more excited about the IDEs than the development teams. We develop internally, we develop all in WordPress. Uh, if we build outside of WordPress, we'll partner with an agency. That's usually when we're getting into larger clients. But the WordPress developers did not do not want to touch the ID. They're like, I don't know what this is, I don't know how to use it, I know how to move block and code editors around inside of Elementor or whatever they're using. Where the SEO team um on the SEO team, we have non-technical SEOs, which means like the link builders and the content writers, those people end up on the SEO team. And then obviously we have the the back-end technical SEO people. Um, but the non-technical SEO people took to it right away because they recognized how easy it was to fix a lot of their um spreadsheets or connect to something and look, use an API to pull data faster. Like I was doing this all manually. Now I can just connect, pull it in, I can push it up into a spreadsheet, or I can write an app script with this, put that into a spreadsheet. So it's given them, they've just sprinted away with it. It's been incredible. I like super proud of those, but that team. Um disclaimer on the ID when you get into it. Uh, again, a little outside the scope of this, but learn what a.env file is really fast.

SPEAKER_01 24:49

Like, if I give you one security protocol, like understand, like you're we don't share it, don't share it in the in the field because then it'll go to whatever place is processing it. So I think the next layer after this, Cameron, yeah, is like, I mean, and this is where we're at, right? Like if people want to know like where we're at, we're exploring setting up some of the open source. Like, there's there's new stuff that's just come out that it like it doesn't take as much setup as it did in the past. And putting it on a server where all the data stays, and you're talking about like legal clients or you know, uh HIPAA with doctors and stuff like that. A lot of these clients, you've got to, you've got to use the enterprise tier to get the terms of you know, terms of service where they're not gonna do anything with the data. But keeping it on your own server, keeping all that information in-house is where kind of legal and medical, that that that's like in these regulated industries, that's really what their big concern is. Because I mean, think about these these major LLM companies or AI companies, whatever you want to call it, technology companies, have everyone's data. And I mean, you know, like, oops, oh, those keys are gone. Like now we got to go reset all the keys in the EMV file. And and also the weird thing is you even set these rules sometimes, okay? Yeah, and then it like you've got to reminder like, hey, remember not to do this, like, don't do this. Like, it's like a really intelligent individual, but also it's weird how they sometimes, even if you set rules, sometimes they forget them or make mistakes, or how it's I think memory, I think memory is the next big conversation as well. Like, so where stuff's housed and how like so. How are you? What is your data memory structure? Uh that's an interesting question.

SPEAKER_03 26:38

So that's why I don't like people working in their um, I don't like people working in the uh the chat windows because of that. So so anybody who's listening, if you worked in a chat, you'll be half an hour in and then it just seems to all forget. If anybody's played with open claw when it first came out, you that was when you felt it the worst. Like you'd be you'd be 45 hour hour in, and all of a sudden it just like it was like someone erased its memory. It's like what it comes it compacted, it just compressed it, yeah.

SPEAKER_01 27:10

It just compressed it all, and it just like you gotta be familiar with anything you talk about, like you should be closing out those sessions to keep it really tight. And that's what I've seen on like enterprise setups, like it's very, very tight on everything that's done. Yeah.

Company-Wide AI Standards And Org Design

SPEAKER_03 27:25

So we'll build the folder structure. So SEO, this SEO. So I'll build, I have an I have an SEO audit folder, right? And so that has inside if like it's a template. So you go in there, it has an agent file that tells the agent what it needs to do. Um, you have a you know, have a.env file that has the the API keys that we might connect to. It has a memory file um that's general memory. So when I when I fire that off, um, I'm in an IDE with the folder structure open. I use cloud code in my IDE. My team uses all of them, but I particularly use that one. I'll open a cloud code window, right? And so what I will let's say something happened with that audit the last time I didn't like. I can use that cloud code to edit the instructions of that folder, but I'm not gonna work in that that window when I go to actually run an audit. I'll open another window. So I do this all the time. I'll have a cloud uh code window open, I'll be running an audit, and something will happen that I didn't like. Obviously, my instructions were off or wrong, or it just wasn't there. I'll go back to the window that's coding the agent, I'll change it there because it has all that context, and then I will switch back to the audit window, and we so I'm not losing, I'm not overlapping, I'm not mixing uh it's its memory on what it's doing. But the folder system is all getting updated in real time. So the folder system stays tight. Um, so yesterday I built, I was trying to figure out prototyping. So one of the issues we run into is directionally, like, how do we get clients to agree to a web design faster? Um because it takes so much time. You know, you got to go through discovery, then you got the design, and like it's two to three weeks, they're gonna see design. How how cool would it be? They're excited, you know, 48, 72 hours later, you're like, hey, is this kind of the direction you were? Yeah, oh my gosh, I love that. Well, so I was building an agent that helped do that, and so you know, I was opening multiple windows. So the first agent I built it, the second agent I tried to run a first pass at it, it messed up. And then I was like, All right, this is where you messed up. So it was editing itself, and I opened a third window. Now it has none of the context of the previous two, it only has the context of the memory file, what it's written, and so you can keep it clean and you can test. And so these are like little things. If you don't spend time in AI, you don't find them. You like the sales team that I know, they just sit in their chat window, and he's like, I just forgot. I'm like, you gotta get out of the chat window. It's too memory intensive on the like anthropic or open it, they're not gonna, they're not gonna have that persistent memory the way you can build it yourself. So um, that's a little bit on memory how we do it, how I do it.

SPEAKER_01 30:08

And yeah, no, though I think that that's a good best practice. I would say also don't keep your uh IN uh in V file open because like it can sometimes copy and pull from that accidentally. Like there's some bad, like for me, I dove into AI right away, but as kind of I don't know, influence or whatever, I didn't talk about it. Like I didn't talk about it. I I was I I'm like, I need to understand this stuff before before I say anything. So I went back and took a lot of classes. I I did a lot of education to build like the transistor foundation, like how does it work? Like what goes into it? Like, you got to know a lot about programming. Uh, not not like you don't need to know how to code, but you need to know how programming works and how uh engineers think. And so being an engineer, like you you got this like you know, a head start for a lot of people, but I mean, I I think everybody felt it too. Like when you take a non-technical person and they're like, oh, here's how we have Facebook set up, and there's like three tiers. Like for almost everybody I've talked to, they're like, what? There's three different logins and three different permission sets that do like that, like you said, that that transition of thinking this way and then this, like wasn't happening. I think even early on now, I think it's getting better and better, it's more intuitive. But you like right now, like like you're talking about everybody, like individuals that are like power users building their own ecosystems, absolutely. And it's like, okay, so you have a couple people that are off to the races, and then you're like, okay, other people are getting left behind. How do we get on a system that like everybody can use it? And we were starting to do like custom GPTs or gems or whatever you want to call it to build those kind of little data lakes for clients and stuff like that early on. But then there's like the step change of, okay, like let's go fully agentic. Okay, like what do we want to give maxis to? And all the horror stories, which there's actually a paper I really want to read that just came out of like basically they they let the agents run loose. But I mean, that's basically what open claw was, right? Like a bunch of people that don't know how how these things work or what how to set up rules, just sort of like, here's my credit card, and like, you know, here here's access to like my email and like everything. And just like, let's see what they do, right? And then there's cron jobs running in the background, and like crazy things are happening. And I I don't know how much of that was sensationalized, but but you know, like what it was at that uh uh that platform where they all talked, uh what like the chip would you know what I mean?

SPEAKER_03 32:44

Like what they they saw the book or whatever. Yeah, they let their agents go wild and create like a social media, and it looked like it looked like they were plotting against us, and then it came out that people were like figured out how to make them post. Like it was why it was a wild 72 hours.

IDE Adoption And Security Basics

SPEAKER_01 32:59

Yeah, yeah. So that then I I think I think a lot of people are really unsure uh of AI um and they don't understand how it works. And I've heard some like different things, and you know, even in the past, we you know we we had a couple of people on our team when it came out that really weren't weren't comfortable with it and we had to part ways. And you know, I I think that uh what's funny is now some of those people are I can tell because they got the um dashes and and their posts, right? Like, but um I I don't know, I think it's a really crazy world right now. And I I I think I would go back to saying the marketers that can understand how to use this and use it early are are gonna have a huge advantage. And I what I'm seeing is I'm getting a lot of requests like you to move outside of marketing, to move into more operations, to move into applied AI in different areas. And I mean, it's kind of like SEO, right? Like you do it in this industry, you do it in this industry. It's still the basic foundation. And so now you have this kind of AI layer that you understand. And when you look through the the that that that lens or or those glasses, if you will, you can apply it to almost anything once you understand like how it works or what the workflow is. And I mean, one of the data points that was just shocking, and I and I believe I've shared it in the past, but 71% of businesses are still operating like the industrial area. And an example of that is copying and pasting spreadsheets. And I've I've had people show me their workflow of their job, and they're going into this database and they're copying this and they're pasting it here, and then they're grabbing that or something with a screenshot, and they're sending an email here, and I was like, okay. And I was like, what else are you doing? And I'm like, no, that's my whole job. And I was just like, okay. So then I'm like going back to their the owner, and I'm like, okay, so I know you really like this person, like, what else can you have them do? Because we can just automate this whole thing, you know.

SPEAKER_03 35:02

Yes, I talk about that a lot with friends uh when they're like, Well, I don't even know what to use AI for. I'm like, be be curious, like that to me is so I had a buddy, this was early on. He was the first person to build a company with AI just to see if he could he's a real psycho in a loving way, like just a real special human. And he used Replet. This was early, so 26. He built this in he started building it in late 24. Um, and so they did he make this really clear. He does his regular company that he owns does over 30 million dollars a year. He just did this just for fun to see if he could automated the whole thing with AI before AI could really do any of this, and he built uh a little business that did$500,000 almost completely automated. The only thing they couldn't solve for at the time was uh graphic design, and so they had they had a graphic designer in doing the work. But other than that, all the all the the cold email, cold everything. Um, it was a it was a uh uh a market that you know didn't have a there was no leader in it. So it was he just tripped tripped over it and it was a great but he he coded up an entire CRM with Replit. It was the first time I saw somebody like do really do business with AI, and I was like, okay, we are past the this is cool phase, this can help me write a blog phase, this is like real now. Um, and then obviously we all know what's December 5th with the Opus release. It was like the this is this is real now. Um, and all of all of us that had been playing with it who who were who were using it but still could see where there were holes, kind of started seeing those holes get closed. And um a friend got uh preview. He has a job that has government contracts. He got a preview of it. He got a preview of it this week, yeah, at work.

SPEAKER_01 37:01

Yeah, I've heard that.

SPEAKER_03 37:02

I there's a couple people I've talked to that are kind of yeah, and you don't know, like Anthropic is obviously very good at PR, so you don't really know. Like, is it really as great as they say they are? And he said it's pretty incredible. It is it is essentially for uh people who don't know. So there's they call it a relf loop. It was essentially a set of instructions.

SPEAKER_01 37:25

Yeah, yeah, that there's a harness, there's a harness out there, yeah. Uh-huh. It's swarm and everything.

SPEAKER_03 37:30

Yeah, this does it on its own. You don't need the relf loop anymore. It is it is that's how he described it. He's like, I don't have words exactly, but it it self-heals, meaning, like, it is yeah, you give it an objective and it knows to kind of keep circulating uh until it meets it and corrects its code base on its way. So we'll see. Uh, I don't think they have the compute capacity to re-release it. I don't, you know, I know they're claiming it's a security threat, which it probably is, but also they don't have the compute for it. So we'll see. Uh we'll see what happens with it. I know companies like his are they're getting it previewed right now.

SPEAKER_01 38:07

What where do you see the future is going? Like where, like with knowing all this, like the the question that I have started to ask, and you know, like people are sensationalizing stuff, but not really, because like I mean, some of the questions like that I've asked just to the chat bot, right? Like, not to quad code or anything like that. Like, like what is gonna be the uptake, right? Like, who's gonna actually really use this and how many people and how quickly is that compression gonna happen? Because I've I've really seen when people jump on it, boom, like they catch up really quick, right? And so I'm going, okay, how many people are gonna catch up on this? And then what does the world look like? And then, like, okay, the people that don't, what does that world look like? Right. And I'm going, like, the future, like from a workflow standpoint, is constantly changing. And I think marketers have an advantage because it's always been constantly changing. So when I tell my team, like, here's a new, here, here's the new workflow, like, they're not gonna like flip out, right? Because at other companies, like, you go, we're gonna change this. There's the whole like psychology and you know, like community culture that that like we're not changing sort of thing. But I feel like marketers are like, okay, like this with this is gonna change, like this is the new tactic, whatever. Um, I don't know what it what are you seeing, or like kind of what what would you predict in your crystal ball if you were to look forward?

Memory Limits And Token Cost Reality

SPEAKER_03 39:34

So I I I am very optimistic as an engineer. Uh, I have robots sitting on the other side of the wall because at some point I will go back into robotics because I love it. But if I am looking at so I my my biggest concern is that like the use and eyes of the world. So I'm older millennial, and we have the 20 years of experience, so AI is just like a superpower to us, but we've put in all of the reps already, so this just makes us that much better. So I have a lot of law, I'm in Chicago, so I have a lot of law firm friends, and so one of their concerns are how do we train the cup the kids coming right out of law school? We don't all that like busy work we just gave them to get reps is just done by AI, can be done by AI. So where I'm not optimistic is like, what do we do to train people to get to that point in their career where they are experts? If AI is capable of doing that dirty rep work cheaper, how do we build, how do we make the experts of the world in 10 years, 15 years, 20 years? I I don't I know AI is gonna continue to get better. I don't foresee, I don't think we land on the like the complete, you know, a utopian side of it. I think it still has it still has to be aimed directionally by experts in humans. Um, and and so it's for me the the the negative or the question I have is how do you get the young people up to where they need to be? Because once, like you said, once you get to like our stage, like the people at my company that are experts are flying. It's the people that quite weren't quite experts, they don't know enough to ask the right questions yet. They're the ones struggling with it. They're like, I don't know what to say. And so we're I'm trying to a lot of what I think about is how do I build education for my team around like how do you build a website with AI? So we're testing, we are a WordPress company, but we are testing other platforms because we know the AI stuff is coming. Um, and then obviously we see step functions, so maybe it just continues to get better and better where it gets you don't have to be in the ID.

SPEAKER_01 41:46

I I think people are gonna build stuff that you can put just off the shelf, plug it in, and so that that those things that waterfalls or or that orchestration's already kind of pre-built. Um there's uh what was it? Uh there are goji or something like that. Someone showed me that. It's it's interesting. There, there's a bunch of like little tools out there that are not like core, they're like certainly like satellites that people are playing with, but people are trying to build some of these things like off-shelf and you can like plug in and do different things. And it really comes back to like what is your mental model? And like, how do you look at these things and how do you understand? And like, I we spent a lot of time on prompt engineering, okay? And even before that, when you talk about the reps, it's like how articulate can you be, whether it's an SOP or just communication? Like, so like communication, how clear can you be? And if if you're putting in you know poorly designed prompts and you're not communicating well, the AI, as smart as it is, it can understand like misspellings and this, that, but it can't understand exactly what you mean. So I think the core skill set is almost like communication, right? And like even if you're like, okay, quite good plan mode, whatever, like you need to describe exactly what you want and those nuances of where you got the reps in of what good looks like are so important because if not, you're just like, oh, looks good building, right? And I've done like, oh, looks good building. I've burned a bunch of tokens, and then I'm like, ah, that's not what I wanted, right? Right. Like, so you gotta like slow down and and really like understand everything and read every line of code and like you know what I mean? Like, not every line of code, but like you need to know what what is happening. Um, and then like I've started to like, you know, I I work a little different than you, but like I'm color coding like the you know, the the the different sessions and like I have labels and I have like I've built little things on like where I'm at on like token usage because you know, like you don't want it because it every time. So for everybody listening, another pro tip out there is if you don't, if you want to like token management is important and so. So, like also good best practices. If you if you don't want that information to be compressed, which it will just happen, but it will forget some stuff, and you want to kind of orchestrate how how you want that to happen a little bit more. Like, know that when you're typing, it looks at everything that you put in that field. So, like you said, going to the other fields, making the change, and then it's instantaneous so you can see the change in the next window. That that does make a lot of sense. But like, I don't think people understand that. It's like you have this huge conversation and then you've changed topic, it has to read everything context-wise, yeah uh to before it gives you the answer. And it's burning tokens to do that. And so, like that's that's even like with this tool, like a lot of these other tools. That's why we're even going, okay, well, we're gonna need to build our own open source, which that's crazy. So, like, whether mythos is like like, who knows, right?

SPEAKER_02 44:57

Right.

SPEAKER_01 44:58

But today, the technology today, yeah, is mind blowing, right? And if you can say, okay, I'm gonna put that on our own server and I'm gonna have, I'm gonna, you know, sovereign, my my my information is gonna be sovereign. It's gonna be my information, it's not gonna be processed, like everything you put in Chat GPT, by the way, never goes away, everybody. Uh so you know, like what was you know, what was Facebook? A surveillance tool, right? I don't know. But like what I would tell you is um like once you have like this has been unleashed, like the like it's like Pandora's box, like it's out of the box, and now like all this kind of you know, inference and cross-training of these models, it's scary. Like, actually, I'm like legitimately like scared, like like there's so much good, and then there's like so much bad that could come of all this, and like I think the only way to be prepared is to like understand it and know and not like put your head in the sand, but it's so powerful and like it's democratized now, and everybody has access to it. You just gotta go learn it.

Self-Hosting Models And AI Governance

SPEAKER_03 46:06

So we're considering, I'm curious if you this has crossed your mind as an agency owner. We're considering building our own cluster so that we can manipulate model weights. Uh, and then right, we'll have our agent at Harness that is Bidview, right? Uh, but right now we we use we use Cloud, we use Gemini, we use we use Codex. Uh but instead of using those models, we use our open source model that we host on our own computer cluster that is the model weights are manipulated by us. In this inst instance, it would be for marketing. Uh we're considering doing that because I think we will get we will eventually get to clients that are concerned about their data. I mean, we don't, I mean, SEO data is they're not concerned about that, but we'll probably get into a situation where there is client data passing through a website that they're worried about, and we need to tell them or explain to them how we secure that or we keep it safe. And and so that's one of the things I'm considering this year. But the leaps that keep yeah, the leaps that just keep happening, it's hard to commit to like I'd have to have a person on staff that can continually update model weights of a model to keep it like it's a it's a question, is it's a hard thing to solve for. We're considering it, but I I think Bidview has to be a little bigger to hire the appropriate person to come in and help us manage a system like that.

SPEAKER_01 47:33

Yeah, uh, I mean, we've looked at like Kim Key 2.5 and like we we've we've looked at some things. It was we haven't looked at it for us as much. Um, I have like two things to say about what you just shared. It was mainly for clients that want to set up their their own setup, right? Like, and that's where a lot of the conversations have come in. Like, okay, you understand this. We need to we need to inject this layer into our team. How do we do that? But we're in a regulated industry, so so what do we do about that? There's also in Texas, there's some laws that are about to go live. Uh, like there's a lot of things that are gonna go live in September. So I think a lot of conversation, like even nationally, like so it's about governance, it's about AI policy, what's happening with this data. I think to your point, like a lot of clients, like, I'm like, no, you got to get give me your old client list and we'll upload it to Facebook and we'll make like a lookalike audience, right? And they're really leery of that, right? They're really leery of that. And I go, I understand why you're leery, but I'll tell you they have all the data and they are and they're only gonna they're they're only gonna hash match it if if they have it. And if they have it, they already have it. Like they like so if it, you know, if like 80% of the emails match, like, you know, okay, you gave them 20%, but like these big companies have so much data. I I'm actually more concerned with uh what it what it like quantum computing, like right, like at that point, like like you know, you that you don't you don't have any so so okay, so you just assume like it's all out there, but but I do think that like you know, you could put you can say don't train the model, like I would encourage everybody to go kind of look at that, look at the terms of service, put the terms of service in a uh LOM and then come up with your own policy for vendors of what you want them to do with it. Like, who knows what's happening now? What you're saying with Amazon makes perfect sense. Um, you know, and and I I think that uh a lot of these companies don't have policies and they're just spinning up all kinds of stuff, and then others are like, don't use it at all. And so, like, where's that happy medium? And I think that that's where the next conversation is gonna be is around governance. I think we hit on a lot of things that I think we'll see over the next you know, six months or year. But to your point, like, yeah, I haven't totally thought about that because like the overhead to do that, um it hasn't been an issue from from clients yet.

SPEAKER_02 50:08

Right.

SPEAKER_01 50:08

Um, and we're trying to be as you know, like like responsible with their data as as possible. Um you know, and I don't know. I I think it I think what are the relationships that Google has with Gemini, right? Like they already have all that data. Like, okay, Mantis has everything from Facebook and the whole, you know, uh like like we'll we'll use Mantis. Well, here's the interesting thing. They still on the engineering side on the silos, we're running like we were trying to like run some Mantis ads, and then it was like saying that these ads like are not gonna work. And or if you like even use like uh if you use uh whatever the I'm sure it's Mantis, whatever they're calling it, meta AI, like to generate images in Facebook, they're uh down promoting uh their own images. Like they haven't turned on, like so. I've I was I was like, oh, let me use their images because you know anything new that you come out with, right? It'll it'll it'll kind of give it an extra boost. So I started using it and I was like, oh, it's it's like depromoting like my post because I'm using their own video image uh generator. And I was like, okay, but if I if I use like a chat GPT or whatever, um, you know, like it it will do fine. And I'm like, yeah, so so so I mean, I think this stuff is really just moving so fast that like once that curve maybe starts to S-curve flatten out, we can kind of see see what's happening. But man, the world's gonna look really, really different at that time. And I mean, I don't know. Like, I is SEO gonna be a thing? What does that transform into? What does marketing look like? Like, I think that good understanding of branding and marketing and those fundamentals, like those reps, like I think we're gonna be in a really good position. But yeah, that's what's happening. Uh, sorry, I'm like going on a spiel here, but but um that's what I'm hearing from law associates, like you said, and I've read some articles that the new, like the people that are just coming out of school, like there's not really a need for that person to do that because AI can do it. So then you're like, well, do I hire that person? Okay, well, they gotta have this knowledge of AI. Oh, I don't see that, but they don't have the knowledge to be here. So there's gonna be a threshold of a lot of hires that are coming out of school that like, do you really want to hire that person? Like, no, you really need that five years, like when the recruiters come get you of whatever you learned. And how are those people gonna make the jump? Um, right, you know, I I I don't know. I I I think it's real world experience. Uh so and then everybody's getting fired. So then you got a bunch of people out there doing their own thing. Like, but that's where most of the small business ours are like the world, like from a system thinking standpoint, like all these things have like you know, second line effects and third line effects. Like, I don't know what's gonna happen, to be honest.

Training The Next Generation Of Experts

SPEAKER_03 53:06

And yeah, I mean, one people will have different reactions to me saying this. We live in the United States, so obviously we need infrastructure improvements, pressing and pushing. Obviously, the technology people are some of our brightest, they have been, they went there because that's where the largest economic gain was. And so if they get squeezed out, people are gonna go look for where the other large, those intelligent people, where are the other large economic gains gonna be? And in the United States, these are these are all these infrastructure projects that we don't ever get around to because we don't have the human capacity to do it. So now that's not to say that the that generation that does get pushed out, they're not gonna re rejob into that, but some of them might, but there might be, you know, over time it resorts itself out into ways that we it's hard for us to fathom because we're just I'll be 40 this year, right? So when I grew up, it's like go be an engineer, go learn how to work on computers, right? That was the whole mentality, and so that's where we all went. I mean, if I look at all of most of the people that I grew up with, it's engineers, doctors, and lawyers, right? Like those are the those were the what places we went. We're seeing AI actually increase the need for doctors because it's able to do and read results faster. So we right, the big argument five years ago was our uh radiology is dead. And then we saw the opposite of that. We saw we need more radiologists because the AIs are able to read uh more uh more of the test faster. And so now people are using radiology for tests that we never thought about. So it's increased that need. Um obviously we talked about the problems with the lawyers, and then on the engineering side, if you're thinking like physical engineering, we need so many of those people right now. Maybe not the computer engineers. That's that seems to be the direction of where the jobs are getting compressed, but uh I it'll be a reshifting of like how that um white collar, like that upper class white-collar careers get shifted. I think you know, I don't foresee AI even if it's as great as it is, it's gonna take a long time for a person to be comfortable walking into a a room with an AI that diagnoses you, right? Like it's gonna be a long time before we get to that. Then the nurses are obviously understaffed, and I don't I I don't care how good your humanoid robot is, I just don't see humans outsourcing that to uh right. So these are there the jobs are there, they're they're underserved now. It's gonna be just a rebalancing of there's only so many people, right? To be able to do these types of yeah.

SPEAKER_01 55:42

Wow, I think I think that's wow. I think that's a great place to kind of end. We'll have to have you back on. I think I'll see what the comments are from this uh podcast. Uh we went, I went off script, uh, you know. Uh but uh I would I like your point. Like that makes a lot of sense that a lot of the knowledge worker people are gonna grab gravitate to the where the biggest gap in the economic gain is, and they're gonna seep into those roles. Like I've seen that in oil and gas because oil and gas is so typical. Those engineers get laid off, and some of them come back next cycle, but a lot of them don't. They go into other things and they add a lot of value uh in these other things with the cross uh industry discipline. And so I can I can absolutely see like a distribution right happening from all these people that are getting laid off from Facebook. Like, where are they gonna go? Like there's gonna be a distribution across a lot of these companies, and then that's gonna bring them up to the next level, and it's kind of the world is flat and kind of get it gets you know it gets layered layered around. I love that. I love that. Okay, so Cameron, for for you know, I'm glad that you you teed up kind of what you did and the kind of clients you work with at the beginning of the call. Um tell t tell anybody that's still listening um like uh how to get in touch with you, how to like, you know, where where you kind of put some of these thoughts, um, you know, any anything you would like to share and also anything you would like to add uh to this conversation that you feel like we we haven't covered, but I'll turn it over to you.

Where To Find Cameron And Closing

SPEAKER_03 57:11

Yeah, it was a fun, no, fun conversation. I think we yes, we drifted, but it's funny how these conversations with me tend to drift. I think this is uh it's my my problem, my fault. But you can find me at bidviewmarketing.com. Uh, if you email us, that it goes to me and our account manager. I do like to post on LinkedIn, although I am not somebody who writes with AI, so I don't post as much as I probably should, but I am actually do writing my posts when I do post them. Uh, and then a couple weeks ago, I did start an AI business uh podcast. We call it the AI Business Lab. You can find it on YouTube right now. Uh, and it is not about the the it's not about AI, it's it's it's about how a business owner should think. A lot of kind of what we talked about here is like how do you operationally introduce AI to a business? We do talk a little bit about the nitty gritty details, but I don't think that's where business owners care to know. Like they're they want if if AI solves revenue problems, they'll do it. If they they're not gonna do it because it sounds cool, right? So um we try to stick to those topics. Sometimes we drift, but uh he's uh he comes from the aeronautical uh defense uh arena, my partner in that. So he has some pretty cool um he's he's the one who got access to Mytho. So he has some pretty cool um he has a completely different viewpoint on a lot of things than I do coming from the business side and the marketing side. So um yeah, those are the few places you can find me. Um and I I'm I'm available. So thank you, Matt, for having me on.

SPEAKER_01 58:44

Yeah, awesome. Uh yeah, there's a lot more like def like as I've moved more into AI, a lot more defense contractors and people reaching out. I mean, this is a powerful, powerful tool, everyone that's listening. Um, you know, if you want to grow your planet with the most powerful, uh strongest, like awesome tool uh in in well, guys, I missed it. I messed up. I messed up. Like you know what I'm trying to say, but there's like I spill we have at the end. But uh if if you're looking for that, like you know, you you should really keep listening to this podcast, go check out what Cameron's doing. Like, I think the world is changing. Um, I just recently um again uh took over EWR digital, so I I have full control of that. We're actually uh spinning out some of the companies that that we have inside of that. We're moving uh some things around. Um we're starting some stuff. So if you have any interest in what we're doing or getting involved, um we are also raising some money. Uh I mean, AI is absolutely powerful and it's going to be doing a lot of really cool things. And uh, we have a lot of long-term clients that want to get involved. And so, yeah, if you have any interest, uh reach out. Um, you can find me on LinkedIn. Uh, and uh until the next time, my name is Matt Bertram. Bye bye for now.