These LMs are the worst they'll ever be. They're only going to get stronger and more capable, and you want to be on top of that.
SPEAKER_03What might be a little different with traditional SEO as compared to how I get indexed in the large language levels?
SPEAKER_01If you're already in their training data, they're going to use the sources that they found to be most credible. And that's with the pre-built training data. But when they search the web, they're actually searching established search engines. So sort of whoever's doing regular SEO to get on page one of Google is going to be a candidate for what gets surfaced in your answer in the LLMs.
SPEAKER_03What kind of content should someone be focused on first? And secondly, if it's AI augmented or fully AI generated, does that matter?
SPEAKER_01I think fully AI generated will matter. I think it's the sort of same emotion behind why you can buy a facsimile mona lista, but it doesn't hit the same as seeing the one in real person in real life. Yeah. Google's gonna be striking back quite soon, and I think that's something that a lot of SEOs are quite scared of, is that we may move to a situation where the 10 blue links is the secondary option and the chat is the first option you see when you land on Google.
SPEAKER_02Zach Ollie is the general manager at Finder US, where he's pioneering how AI intersects with SEO and content. From doubling revenue to championing hands-on adoption of AI tools, he's leading the charge into the future of digital discovery and productivity.
SPEAKER_03Welcome to Using AI at Work. I'm your host, Chris Daigle. Each week we'll be learning how today's business owners, entrepreneurs, and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. Right now, every business leader is asking the same question. What are we going to do about AI? If this is you, ChiefAIOfficer.com has the answer. We give you a simple path forward where we provide executive and team training so your people know exactly how to safely use generative AI in their day-to-day. We also manage the deployment and implementation to make sure tools actually get adopted and deliver results. And we'll also guide company-wide transformation so AI becomes part of your operating system, not just another shiny object. The companies that act now will increase productivity, cut costs, and grow faster than their competitors. Those that wait will get left behind. So if you want to make AI work in your business, visit chiefaiofficer.com and see how we're helping companies of all sizes finally get results from AI. Hey everybody, welcome back to another episode of Using AI at Work. I'm Chris Dagel, I'm the host, and today our guest is Zach Ali, the general manager of Finder, which is an award-winning global fintech company helping people around the world make better financial decisions and improve their lives. And Zach and I had the opportunity yesterday to kind of dig in a little bit about what we're going to talk about on this episode. And with what they're doing with AI and just the way that they're looking at things, I think that everybody listening to this is going to walk away with a better understanding of how they can start to use this in their own businesses. So, Zach, before we get started, I'm always curious, like what has the AI journey been like for you leading up to this call today?
SPEAKER_01Yeah. So I've been working in SEO since about 2016. And I remember when ChatGPT launched in late 2022 and just thinking, oh, this is going to change everything. And so immediately I started to adopt it, play with it. I started to see sort of how written content was going to be quickly commoditized because let's be honest, it it's easier to ask a question ChatGPT than to open up 10 blue links. And so immediately was like, I need to understand this. The company needs to understand this. We need to dive in, get our hands dirty. And from there, I've I've tried to stay on top of every update, but the velocity is a bit overwhelming at times. And yeah, so I'm I'm still a you know beginner's mind sort of mindset to all this and and trying to take it on as as it comes. Yeah.
SPEAKER_03Nice. You know, um, whenever we talk about AI and SEO, those episodes seem to get a lot of interest, a lot of feedback, and a lot of listens. So, you know, I've heard it called AEO, GEO. What are you calling it?
SPEAKER_01Um I will call it whatever the client wants us to call it, but at the end of the day, I think it's all SEO. Um the practices are still more or less 99% the same. And I think until there's a significant enough divergence, um, it's easier to just remember the one acronym than the uh than the three. Yeah.
SPEAKER_03So what what would be an approach, just macro level, what would be an approach for a company if they wanted to make sure that as people are starting to search more and more in the uh LLMs for information about a service, a good, a solution to their problem, what might be a little different with traditional SEO as compared to how I get indexed in the large language models?
SPEAKER_01So the way the large language models work is um if if you're already in their training data, they're going to use the sources that they found to be most credible. So they're going to be using signals that they've found across the web. Do they have a strong editorial standards? Have they been cited by really credible outlets like national newspapers? Um and and do they use uh rigorous like external linking to like .govs? And so they're they're really trying to find the sources with the most trust, the highest trip area of trust. Um and and that's with the pre-built training data. But when they search the web, they're actually searching established search engines. So sort of whoever's doing regular SEO to get on page one of Google is is going to be a candidate for what gets surfaced in your answer in the LLMs.
SPEAKER_03So you mentioned Google. Um, however, ChatGPT, it primarily uses Bing, right, as its search engine?
SPEAKER_01So there's been a bit of debate that recently it may have switched to Google. Um they've done some studies that the um search results similarity are starting to line up a bit more with Google than they are Bing. So they may be using a combination of the two.
SPEAKER_03Generally, I guess is more energy going into indexing on Google than on Bing. So would that necessarily mean that it might be easier to get indexed in uh the search results from ChatGPT?
SPEAKER_01So I don't know. I actually have never met a Bing specialist because the pool of traffic is has traditionally been a lot lower. Um but I will say that the conversations just internally at our org when we knew for a fact they were going after Bing results was how do we crack this code? How do we get surface more on Bing so that we can show up on ChatGPT?
SPEAKER_03So that might be uh um an activity that would be worth any of the listeners, if they've exclusiven exclusively been focused on Google, they might want to also explore Bing placement.
SPEAKER_01I think so. I think um when you do a search uh or when you start a chat with ChatGPT and you and it links to external sources, try doing that same search um on Bing and on Google and see where the similarities are. And it can sort of probably give you a way to triangulate um when it's pulling from one source or when it's pulling from the other. Uh there's another thing you can do. This is gonna get real nerdy, but you can you can right-click on a ChatGPT chat, um, do inspect element, go to the network tab, then refresh the page with the network tab still open, and then you can copy one of the JSON files that actually shows you the fan out query thinking process of all of the specific terms that ChatGPT searched. And that way you can say, okay, well, I know they cited me, not when they searched best credit cards, but when they did best low interest credit cards. And so you know, okay, you're already showing up in that citation. Well, one, you want to protect that citation, but maybe you want to go after these other keywords it used to create its answer.
SPEAKER_03Nerdy, but enjoyable. Thank you for that. Obviously, based on that, you know what you're doing. Um what does the internal effort look like at Finder for you guys to make sure that um not only are you showing up in the traditional searches, but you're also uh showing up in LLM results?
SPEAKER_01So one thing that we're doing is um that we found that ChatGPT really likes recent recency. So it's actually probably plays a little bit more of a role in their authority than um maybe the other search engines. So we're trying to make sure that everything we're doing has an updated date of 2025. Uh 2025 is sort of in the title tag when it when it makes sense, um, and just making sure that our information is as accurate as possible because one thing that people are rightfully afraid of is the hallucination that can happen from um yeah, from ChatGPT. And ChatGPT is also not wanting to give people wrong information. So if they see recency uh as a signal on your page, they're more likely to cite it when they show you the results.
SPEAKER_03Aaron Powell It's a great tip. So what kind of content should someone be focused on first? And secondly, if it's AI augmented or fully AI generated, does that matter?
SPEAKER_01Aaron Powell I think fully AI generated will matter. I think people, and if it doesn't matter from a technical search standpoint, it definitely will matter to the humans who have been trained to spot in a digital age, you know, starting every sentence off with in a digital age or in this digital landscape, and the M-dashes, which I'm quite upset about because I used those before. Yeah. They became synonymous with ChatGPT. Um, but there's uh the the tropes immediately folks are gonna be like, ah, well, this was there, no effort went into this. And I I think it's the sort of same emotion behind why you can buy a facsimile Mona Lisa, but it doesn't hit the same as as seeing the one in real person in real life. Yeah.
SPEAKER_03So what does a day in the life of uh somebody on your team who's tasked with making sure that Finder is showing up in uh LLM results?
SPEAKER_01So right now there's a few ways you can do it. So there's one software called Profound, which sort of runs a bunch of queries on your behalf to see where you're cited. But um at Finder, we actually just do the manual searches ourselves. I think there is um a lot of benefit to that actually, to not outsource that because again, chats are deeper than just one query. Sometimes it's gonna take multiple back and forth before you land on the article that the user really wants. So by doing it yourself, you're actually thinking more like a user and more like a human. And that's going to give you the better sort of understanding of what's working and what's not.
SPEAKER_03Yeah. So um obviously you guys were doing this before the release of Chat GPT. I know that um Finder's been in operation for a while. What was that transition like for those experts when this new generative AI boom started happening? What was the the hesitation or what was the excitement? Like, what was the the vibe for your team?
SPEAKER_01Um so our editorial staff, I think there was a immediate fear, right? It was uh this is coming for our jobs. Um there was the anger around the plagiarism, right? Because understandable. I think in even in the Google AIOs, they're they're essentially creating a mini article based on work that you did and not giving you the value exchange isn't there anymore for the creator and the search engine. Uh the search engine is getting much more value out of your content than the creator is these days. Um so I think there was that fear, there was that um a little bit of anger, and there was also the folks who weren't who were inclined to use it, was a little bit of a letdown because to be honest, the writing quality is not that great. Yeah. So and I think once more folks use it, they they sort of breathe a sigh of relief. I'm like, oh, okay, this isn't coming for my job anytime soon. Um what we like to say at Finder is it's pretty good for boilerplate. Um if you have parts of your sort of article that or or content that is really generic stuff sounding stuff, it's totally fine to sort of use ChatGPT to augment that. But when it gets down to the meat and the substance, you do want uh human effort guiding that, I think.
SPEAKER_03So not to get into the secret sauce, but uh ChatGPT seems to be kind of like the you know the brand that most um new users are familiar with. However, um expert level users like the people on your team probably have some preference when it comes to the writing style. Is ChatGPT the tool they use for supporting their writing?
SPEAKER_01So some folks use ChatGPT, other folks have used Claude. I mean, we everyone's playing around with a lot of the different models. We have a paid subscription that the teams use, um, and some of the folks also have paid uh Claude subscriptions, and I think none of them quite nail it quite correctly. You know, you can try to build the skills files or the the stuff that's sort of trained on your writing, but it's still inconsistent. I think that's a huge problem with AI everyone is sort of experienced, is that it it 90% of these sort of fully AI workflows are failing because the can you know it might have one output that looks completely different than the next. And so um, yeah, I wouldn't say anyone is using it to fully replace what they're doing, but um they may use it uh to support them. And and one thing I like to say, one of one of the watershed moments I had for when it comes to writing specifically was um I I liked it because it showed me what I didn't want. And it, you know, it's like so the people who are better at giving out advice than taking their own. It was the sort of the same thing, like you give me 10 headlines and I'll critique those, and now I know exactly what I want. But I had to see what I didn't like first.
SPEAKER_03So for a listener who maybe they don't have access to the level of talent you that you guys do have internally, how much massaging of the output? Let's say they want to create the they won, they want to pursue this, they want to start showing up for their construction company or their CPA service or whatever, um are there particular types of content that seem to get maybe faster or more consistent uh uh results when it comes to showing up in the answers from the models?
SPEAKER_01Yes, I so I think long tail queries get show up much faster because if you're the only one who's written about something that has less pool to pull from. Um and so we've actually seen a lot of success there is is where most folks went really deep on the few head terms that will you know provide 80% of the value for their business. Finder took a bit of a different track where we're like, well, we're gonna cover everything else. We want to be comprehensive, we want to go, uh we want to really understand our users and find answer the question that that, you know, those few people have, which then what we're seeing actually in Chat GPT is it's actually not a few people. Um there's a lot of different nuances you can do in the conversations, but people's needs are ultimately pretty similar. And we went after like the people who are living in New York and make 70,000, you know, having those really varying sort of long tail spaces, and those are getting a lot of traffic from ChatGPT. And so we've seen a lot of success there.
SPEAKER_03That was gonna be my my next question. What has that upward gradient looked like as far as let's say end of 2022, it comes out, how much traffic is coming from on your site, coming from results from LLMs as compared to today? What does that look like?
SPEAKER_01So I actually did that analysis recently, and it was uh quadruple digit increase from year over year. So yeah, the adoption, the traffic is coming, revenue uh I guess I can say doubled this month than it did from the previous month, uh, which was exciting to see. And we're actually seeing a lot of different pages show up that um were previously out of reach for us, um, now appearing in ChatGPT search, which is giving us a higher quality user, um, which we're also seeing across the board, you may have heard that as well, that because folks are walking themselves down the funnel within the chat, by the time they get to your site, they're usually ready to make a decision.
SPEAKER_03Great point. Huh. So how could somebody find out how they're indexing in uh LLM results?
SPEAKER_01To be honest, uh you would have you can use some of the SEO tools, but it's only going to be somewhat directionally correct. And also it's gonna be based on that first chat. So you'll have no visibility on the the follow-up questions. So it's a bit of a black box and to be honest, I've been enjoying that um a lot because it puts the customer back at the forefront of people's minds and and people can't really hack this game because they don't know how it works yet.
SPEAKER_03You know, for for the listeners, I think that this is a good point. I hadn't even considered that, but if you're showing up and you know, somebody's got uh an issue, a challenge, they're looking for a solution, they go to Chat GPT, they go to Claude, they start working with it, they start getting more context on a solution, and then your company pops up, like the pre-selling has been done, it seems like, as far as I mean that that's a that like that's a huge aha for me. So for those of you listening that are kind of like, uh, we don't really do all that kind of stuff, I can tell you, and if you can get quadruple the traffic from uh Chat GPT or Claude or whatever your model is, uh coming in. So now that we're talking about the models a little bit, is Jim and I are are they exclusively pulling those types of links from uh Google's search results?
SPEAKER_01I couldn't say confidently, but I I would be very surprised if they weren't.
SPEAKER_03Sure. That makes a lot of sense. So then that means if if you have been exclusively focused on Google and haven't been necessarily work working with Bing, at least in Gemini you'll be getting uh good placement as long as your SEO results are good in Google searches.
SPEAKER_01Right. And the the even the variance between Bing and and um Google is not so much that so I think like just good SEO works everywhere.
SPEAKER_03Yeah. So I'm starting to see, you know, uh ChatGPT released Atlas, Perplexity has released Comet. So we're now finding this this kind of like LLM meets standard web browsing kind of experience with the sidebar. And for those of you that are listening, if you haven't tried either one of those uh AI-enabled browsers, um do it. I'm I'm getting a lot of Atlas I'm still new to, but Comet I'm using, that's my primary go-to now. How is that going to impact this? You know, that now we've got this regular search, which most people are still doing, they don't even know about these AI-enabled searches. How is that going to impact kind of this paradigm of uh being front and center when a client's looking for a solution or prospects looking for a solution that you can solve?
SPEAKER_01Yeah, so those browsers launching is actually an incredibly big deal. Um, what those browsers do is they give ChatGPT and they give perplexity access to click metrics. So Google, by because most folks are using Chrome, you know, ever they they know who clicks on your link, how uh or how long they're hovering over your link, if they're skipping over it. Once they go on, it follows you onto the page as well. So once they're on the page, they see, oh, are they going to bounce right away? So they're actually getting user data from Chrome. And and OpenAI was like, okay, well, to to know if we're actually giving users what they want, we need we need that same data. And so that's why Atlas and and uh Comet are actually a massive deal. And I think we can expect the search results to um change quite significantly. I don't think they will just be pulling from Google for long. I think they will uh be able to sort of come up with their own ranking methodology.
SPEAKER_03You know, Google's own this space for 20 plus years. How do you see the introduction of these AI-enabled browsers uh eroding market share for Google or shifting the behaviors of how we uh we get information online?
SPEAKER_01Um so I see um it's a it's a credible and serious threat for Google. I think they know um that their business is built on search, that that's still the cash cow that that funds Alphabet. And um any competitor who's launching Me Too products that uh match the same sort of function is a threat to them. And so we know Bing wasn't didn't really pan out right uh as a as a credible threat to them, but ChatGPT is not Bing. Um it's a completely different interface, and they're actually the interface that's synonymous with chat-based search, which is not Google. Google's doing everything in their power now to sort of change the way that they display results. It's starting with the AIOs. Now they have AI mode as well as an option to where you can go into the Gemini search or chat-based search. Um so Google's gonna be striking back quite soon, and I think that's something that a lot of SEOs are quite scared of, is that we may move to a situation where the 10 blue links is the secondary option and the chat is the first option you see when you land on Google.
SPEAKER_03Very cool. I I don't use Google very often. I've been uh really pleased with using Perplexity for the past, I don't know, year and a half or so. But this morning I was in a browser and it defaulted to the Google uh search, and that was the first thing it popped up and said, Do you want to use the AI chat? And I tested it, and uh it's obvious they're feeling the uh the pinch from the others, so they're shifting kind of the direction that they're taking this. Um so uh you know let's kind of shift gears a little bit because I'm I'm particularly interested in in outside of just the uh AEO SEO kind of conversation, how Finder across the company is um encouraging uh AI upskilling of the team and it w what is what have been some of perhaps the challenges, but also the obvious wins that have come from that?
SPEAKER_01Yeah, so AI enablement is um you know one of the guiding sort of um initiatives this year for uh for us is to how can we uh use AI to and automation to enable ourselves to work smarter um and faster and get more done um with their current staff. And so there is that global push, but um how that's translated into an organization. Organizational sort of adoption is still a work in progress. And I think that's probably the same at most companies right now is that usually, you know, it's funny when you read uh the news, you it's like every other day there's a new CEO at a company that's written a memo that's been leaked that's basically saying you have to use AI, like demanding um that it happen. And so that just goes to showing me that people are slow to adopt this technology, and no one has cracked the code to actually get that buy-in um in a way that doesn't require a blunt instrument uh to do it. And so, Finder, we're trying to not use the blunt instrument method. We actually want people to see the value in AI and by getting people to do small tasks that um make their life easier and fun tasks. You know, we had we had an exercise we did last this summer where we had the editorial staff, you know, just say, like, no rules, just create a uh a really cool page, but you have to use AI to do it. And you know, we got calculators out of that and in and graphics, and folks are like, oh, like we didn't put so much pressure on it. And I think that actually creates more buy-in than the you have to use it or you're fired type of situation, or you have to use it or it's gonna show up negatively in your performance review. Um, which again, Finder is not a 5,000-person company. So maybe it's just a speed thing and that's the best way to do it. But uh, we really want to actually see buy-in because I think that leads to that lasting adoption that where we actually create a company of AI natives um instead of folks that just know how to use it or have used it one time.
unknownYeah.
SPEAKER_03So I think that's a great point for anybody listening, is that uh there's certainly two approaches. We've seen what what I guess Duolingo and Shopify and Box have said, and that is before we hire anybody else, you need to prove the case that AI can't handle that role. That's like a mandate. But in your case, you guys are taking a more, I don't know, empathetic approach to uh this this shift in how like the average teammate is thinking about their their career and what the future of this new technology looks like for it. So um you guys have uh you know you've you're not exclusively a US-based uh employment environment, right?
SPEAKER_01Right. Or a global company.
SPEAKER_03What has been the the difference between the cultural acceptance of your uh employees around the globe compared to the US, I guess? And are there any that are kind of like showing real uh enthusiastic adoption compared to the US?
SPEAKER_01Yes, so um our CTO, who's actually in Australia in Australia, he is one of the most enthusiastic folks ever. Um, uh a clot evangelist, uh as uh as I'll refer to him. And you know, he's basically running his entire workday through his terminals at this point. You know, he's instead of opening the JIRA or Confluence or Slack, or you know, he's running it all through his terminals and really showing sort of what this uh the power of these things and and sort of through osmosis of talking to him, and I'm like, oh, well, now I want cloud code and I want to see what I can do. And and that's starting to sort of spread like wildfire throughout the org where we're all like, well, who can cut who can use the coolest MCPs and integrate, you know, have the most integrated workflow and cut the most amount of time down. And so I do think there's some, you know, there are evangelists within the company, but I do think the gap is still too wide between people who use it and people who don't. Um when I what keeps me up at night is thinking, what if there is a US-based competitor who's running training, weekly trainings, right, to get their whole staff on AI. They'll be able to run laps out of us around us by the end of the year, right? Um so we need everyone to sort of embrace this technology the same way we embraced enterprise tools like Google Sheets and Docs and Slack and you know, um it's it's I think these LLMs are gonna be as as commonplace as those.
SPEAKER_03You know, that's interesting. I was having a conversation yesterday, and I've been we we work with a lot of companies, we do the upskilling, we help them with the strategy and stuff like that. And you know, they're not out there talking about it. They're just kind of we we've made a decision, we're gonna get our team to use this, we're gonna start introducing pilots where there's automations or agents introduced. And it's a process, obviously, it's not an event of bringing AI into your business. But what's happening is we ch we we do a weekly kind of AI council that's led by a chief AI officer. So we're seeing that incremental, not even incremental, it's in some cases exponential, impact that it's having on productivity and things like that. And what you just said, this this like my competition could be taking a much more aggressive approach to upskilling their team than we are. And that makes me nervous because I know the impact that will have on their ability to produce more, produce higher quality, produce it faster, less resources required. So for anybody listening, I want you to to, I mean, this is somebody who's uh obviously knows their stuff, our guest today. And uh is your competition doing this? The answer is to some degree, somebody that's a comp competitor is doing this. And if you're not, and you're or maybe you're oh, we did a training in the there, you know, we did it last year or whatever, we we plan on doing something Q1, Q2, and that's the extent of it, your people will get some benefit, but they will not be competitive in 2026, 2026, whatever whatever the timeline is. There will be a point to where those competitors who are doing this behind the scene, that are doing the weekly training, that are listening to the podcasts or you know, reading the stuff that Zach's putting out on his Substack, they're going to be hard for you to like catching up to them once you realize they're they've done it and you're starting flat footed, nearly impossible. I mean, would you concur?
SPEAKER_01I would completely agree. And I think it's best to operate as if they are. Um it's nice, definitely the safer way to operate. Um, and also it's important to start now because the change rate of these tools is exponential. I mean, it's it's slowed down somewhat, right? The jump from ChatGP3 to four was mind-blowing, four to five was less so, but it's still changing. And one thing I like to remind folks is that um these LLMs are the worst they'll ever be. So they're only getting better, they're only going to get stronger and and more capable, and you want to be on top of that. And if you've if you were someone who used it in 2022 or 2023 and wrote it off because it couldn't do what you could do, yeah, jump back in and also try new ones. Um, try Claude, try perplexity, try, you know, try them all. Um there are also ways that have nothing to do with generative AI, you know, have things that have nothing to do with generative AI, at least in the written form, that could be aiding your workflow. I don't know if um if you've used notebook.lm from Google. Yes, but fantastic. But turning your notes into a podcast is a massive unlock for people who learn audibly or are writing the train into work every morning like I am, so that I can get the uh, you know, I can digest some of these concepts in a way that's engaging to me and doesn't require me to be glued to my phone the whole time. Um so there's so much happening here. And I think it would it'll do nothing but benefit you to have an understanding of these tools.
SPEAKER_03Yeah. And you know, you you mentioned getting started now. Even though, let's say, you know, in a month or three months or whatever, GPT six comes out or Claude 4.9 Opus comes out, right? The the important part about starting now is that you're you're having that mind shift of how to think in AI, how to use the tool, how to communicate with, how to iterate and get those results, right? So for those of you listening that, oh, well, we'll just wait till it it all the dust settles and there's a winner. I would say no. Start now, even though the tools are, and you know, Zach, it's kind of tongue in cheek, but it's the truth. They improve. I think in 2024 we used to track about four major updates across LLMs and tools per week. 2025, we're seeing in some cases 11, 12 updates across, you know, Perplexity is releasing the browser and GPT is releasing pulse and bop, bah, bah, all this kind of stuff, right? So please don't wait for the technology to get, you know, as good as it's gonna be. Start now, create the habits. There's gonna be wins. There's been wins measured since, you know, uh 2023 when people were using these tools. So um go ahead.
SPEAKER_01Yeah, well, I was just gonna say um, because you you'd be amazed how quickly you can pick them up as well. Um, if you are a good communicator, it also makes you a better communicator because it teaches you how to specify and get exactly what you want in your outputs. But I have a colleague in our in our New York office who was in the on the product team, and he was someone that I was like trying to convince to download Claude and trying to get access, and then he got access. Um, this was I'll say this was four weeks ago. Um between that when he just got access, and the story I'm about to tell was, again, a month. So didn't use Claude, got Claude, got Claude code, downloaded a bunch of MCPs. And then we had an instance this week where he was looking at um uh he wanted a tool because we always have to view things on mobile, on desktop, on tablets. We need to see what the website looks. He's like, oh man, there's this paid tool that I want to get that allows me to view all of the screen sizes at the same time. And when I scroll down, um they all scroll at the same time. And so this is just so much better than using the inspect element tool um that Chromium browsers have. Um, can we get it? And I was like, well, why don't you try building it yourself? He built it himself in a night. So we just cut costs, he and and it's already being used by our QA team. So how amazing is that to see something go from idea to in production when he didn't even have the tool that built this a month ago. So it's really a transformative sort of technology for us. And the more and he and that's just one person. Imagine if everyone in your org now felt like they were no longer shackled by lack of technical ability or time or whatever. What could you be doing? How much faster could you be testing? How many more ads could you be creating? How many more subject lines could you be A B test? You know, there's so much you can do here to speed up, and that's what gives you your edge. Every founder and startup company will always say the quicker you learn is is what's going to give you the, you know, the quicker you learn and cumulate data is what gives you your edge as a business. This is that on steroids, or it can be for the folks who really lean in.
SPEAKER_03Would you say that he was a technical person?
SPEAKER_01Um no, I wouldn't. I would so we actually had this conversation about where do we fall on the spectrum because we're not we and him are quite similar, like we're not engineers in the fact that we can't just like open up visual code and start, but we're also not complete new novices. Um like we can do CSS and HTML and basic JavaScript and whatnot. And so we understand code um and and sort of that part of the side, but we're not engineers. So technical enough to be dangerous, but not software engineers. So that's what's really blowing my mind is like, what is a really good, what can a really good software engineer do? And some of the conversations we're having um at Finder is with our CTO is like we may be looking at a future where engineers are just managing their AI agents, and those agents are managing their sub-agents. So no one is actually doing the hard coding part, but they're doing more of the reviewing, the augmenting, and the creating the their their these digital employees for themselves.
SPEAKER_03So I'm not a technical uh business person. Um I actually heard that the new term is not non-technical, but post-technical. Like you may not, I mean, I've built apps with AI Studio, which was just natural language stuff, and it's like um that are in production that we're using in our company now. So, but you mentioned this this terminal environment. So I haven't had we've got developers on our team, and and I pick up a little bit as we're talking about some software that we're building and things like that. But for a non-technical user, kind of explain what this clawed code environment with the MCP is like, what is it, what does it allow me to do?
SPEAKER_01So Claude Code is a uh product of anthropic that allows you to operate Claude um in your sort of terminal, which is like you know, your command line, I think that's what Microsoft calls it, but the that sort of interface. And you can do it in your visual code on your screen. You can do it in in any IDE, which is um you know where people write code. Um and so what Cloud Code can do, what makes Cloud Code special is it can connect these things called MCP servers. Um the best way I can explain it is like think of the App Store. And so Shopify has an MCP, so and uh Atlassian products have an MCP, and uh Slack has MCP. Yeah. Yes. And so now imagine in your terminal, you just type in like slash claude um enter, and then it starts the terminal. And you have to, of course, there's a human in the loop process to all of this. You have to approve permissions and access and all that. But once you've done all of that, now you through Claude code, you can say, give me a summary of everyone who slapped me overnight. Or can you go create this doc, create this JIRA ticket for this developer, assign it this many points, and do it. So you're not opening up up, you don't have to have 20 browser tabs open anymore. You're operating everything through conversational interface where you're just telling this agent to do things for you, do things on your behalf. Uh, it can connect to GitHub, it can push code for you, pull code for you, um, it can write documents, respond to emails, reply to Slack, send Slacks. It can do so much within this one interface. And I'm I'm probably explaining it terribly and just scratching the surface. Um, but that's from my point of view as a what post-technical um person. Yeah.
SPEAKER_03So for knowledge workers, when I started getting involved in in generative AI, I saw a stat early on that said that knowledge workers spend, you know, a significant portion of their day looking for the knowledge to work with, the search and retrieval, right? So you're saying that with uh a with the right person or the amount of practice necessary to do it, claude code could allow me to really like I can access my Google Drive or or Teams or SharePoint or whatever, and just through natural language be able to do that search and retrieval to speed up access to the information I need to do my job.
SPEAKER_01Exactly. Um and if you need to do some deep thinking, you have the context of all of those things. And integrate so you can see traffic data at the same time as revenue data, at the same time as search ranking data, and you can paint this picture uh without having to look scan through each three of them. You can just through chat, is there any correlation between traffic and rankings and revenue and season? You know, you can come up with what your question is, but it has all that context. Yeah.
SPEAKER_03So previously, in order to get that same information, that little example that you gave, that's great information, right? Like there's insights that I can get from that that will point the direction of my next step. Previously, without this, or even with, you know, well, I know how to prompt, but you're not doing it at this level, that's multiple people in different departments, you having to make a request to get the information. They got to find the time to pull the information and then provide it to you. And then you still need to assimilate all the inputs from uh the finance team and from the traffic team and from the marketing team and all that kind of stuff. But now in clawed code, if I take the month to learn it, like your uh example earlier, I can now, as an executive at least, anything that comes up in the business, I can go, I wonder what. And now I'm not just getting, you know, Chat GPT saying, oh, here's a suggestion on what you should do. I'm getting real feedback on deep insights into my business without needing to get my own LLM that's trained on my data or anything like that. Like I've got access to that level of insight through a $20 a month subscription to Claude.
SPEAKER_01Yep. Yep, exactly. And um, you know, you'll be limited by token usage, right? So the team plan, the more expensive one, is is really what's gonna allow you to fly. Um which is what, 200 bucks a month? Yeah, something around there, 150 or whatever it is. And um, but the probe account or the the one that's a step below that one is still useful. It's perfect for playing, it's perfect for getting your feet wet um and actually understanding these things. And truly, I can't overstate enough the only way to learn with these is to do. Yeah, it's it's no one is gonna be able to just come in and tell you, and then all of a sudden you're an LLM expert. You have to play with these prompts. You need to go through the pain of it giving you shitty outputs, excuse my language, and then having to correct them and saving prompts that you thought really worked really well, or um and and just sort of go through that whole journey. But on the other end of that, you feel like you could do anything within reason, right? Um and so I mean another thing that uh you mentioned the uh the executive sort of perspective on it. Um one thing I'm worried about is like me and the same colleague who had built that sort of um screen size viewer, um uh were talking about like, oh, like I actually did everything I needed to do this week on Monday. And now here's the dilemma. Do I tell my boss that I actually got everything done that same day, or do I make it feel like it took the entire week? And so of course he would, we were just joking around, and his boss is technical, so we'd be able to know. But imagine you're not technical. Imagine you are someone who thinks you're paying someone for five days of work. They got everything they d did in the needed to do in the first hour, um, but you would be none the wiser because you actually don't know what these things are capable of. And and uh so I'm really scared of the sort of executive class of of leadership, not uh the gulf between what they know and what the rank and file know getting too wide. Um it's gonna really yeah, that I could see that being disastrous uh for productivity. Um good for the person who's getting away with it, but but not for the rest of the company.
SPEAKER_03But then still they're quote unquote getting away with it. And that's not necessarily uh A players don't necessarily want to get away with it. They want like throw more at me, right?
SPEAKER_01Exactly.
SPEAKER_03Exactly. And at least internally and maybe anecdotally from your peers that you talk to or the content that you're consuming, this gap that you mentioned between the, you know, uh management and labor, let's say, right? Is there a gap and who who is holding the advantage in that who who knows more, the leadership or the the teams?
SPEAKER_01I would say um, well, if you're if you're one if you're a leadership that's spending most of their days in meetings, um, it's probably the teams that know more, because unless you're taking up the initiative to after hours or in your the white space in your day to really learn these things. But um they're getting exposure still through like you know AI note takers and on the meetings and seeing how good they are. But really, I think playing with these tools will be important. It's is important for everyone. It really will be like the the sort of like meme of like the older employee not knowing how to create save a PDF, right? It's gonna get to that point soon.
SPEAKER_03Execs, it's time to start playing. Where would you suggest? So obviously, in this conversation, uh you and I, we have a pretty deep knowledge about the landscape of generative AI and how it's being used in businesses. So I was able to follow the clawed code conversation and understand it. I've seen the terminal, we've got people that talk about it regularly internally. But for an executive who is listening to this or or uh you know any level of the business who's listening to this podcast, and they hear that and they're like, oh man, I'm already too far behind. Where what would that what did the upskilling process look like for you guys, taking people from, yeah, I've heard of Chat GPT to now they're using it throughout the day dynamically to address anything that comes up in the business? What what steps did you guys take?
SPEAKER_01So again, it was it we started by giving everyone a taste that would be enjoyable or fun and simple, um, not something where they would associate, you know, LLMs with headaches. And so that that was sort of the jumping off point, at least in that terms for the editorial team, um, to get these things done, uh, you know, to get them to sort of see it in a more positive light and to get them to use it uh more repeatedly. It's also great for research a lot of the times. You can ask it to uh, you know, instead of opening up multiple tabs and scanning them through for information, you can use ChatGPT's uh deep research function and say, you know, give me everything you know about um how industrial revolutions have taken how many industrial revolutions and sources and and whatnot. Um and you can use it in that way. So there's there's many practical ways. And so I'd start by using it in a very simple, practical way for yourself, um, even if that way is saying, how do LLMs work and getting a reply? And and these also are great training tools. A lot of the times, students, and I was one of those students that felt like I didn't really understand things on the first go, and I would have loved to be able to follow up in Chat GPT. You can you can keep following up for as long as you you want until you fully grasp a subject. And so I actually don't think the time to learn uh is going to take too long. Yeah, but the longer you put it off, the the scary of a monster it's gonna be in your head of like, oh, I don't, I just don't understand it. I just don't understand it. Um and you don't want that to happen. Yeah.
SPEAKER_03Well, that's a great tip for anybody that's listening. If you've if you've got a paid account to Chat GPT or or Claude's got their version of it, and again, we're talking about the twenty dollars a month. We're not talking about, you know, big investment here. If you've never run a deep research, I think you'd be blown away. With what's possible with these models, because it's way more than just hit enter and then all of a sudden letters start popping up on the screen. This is like you can go grab a cup of coffee. You can go walk the dog while it's doing its research. So that's a great tip.
SPEAKER_01Oh yeah. And uh you can repeat tasks too. Like if if you've ever set up a Chat GPT task, you can uh have it give you the top five news stories that of that day, right? Or what you know before the market opens. We we we use it a little bit to um research for our market briefing newsletter that we we have where we try and curate the best uh the stories that are gonna be most impactful on the stock market that day. And it's so much easier to have ChatGPT at 5 a.m. using deep research go through and find these stories than to have someone wake up that early and really try and figure it all all out before uh we send it off to the code.
SPEAKER_03They're getting the kids ready and trying to get off to work. Yeah, it's not gonna happen. You know, it's like I'm starting to to see the tools can do whatever, right? But they're sitting idle waiting for us to say, ooh, what if it did this? Oh, and then I could do this and this. So, and the only way you start having those oohs and ahs is from getting started and using it and that that like uh we call it thinking in AI, but you start seeing the matrix. Pretty incredible. You mentioned um that you had been having a conversation with your CTO. Who led this uh push at Finder? Was it the CTO? Was it the ops person? Was it just some enthusiast that said, hey guys, we gotta do this?
SPEAKER_01I I think it's the folks who have like so it was a mixture of folks, but uh so the CTO was one of them, I was one of them, uh the colleague I was talking about was one of them. Um, but really the people who use it and then be overnight become evangelists and are like, I have to tell someone about it. This is just too cool. Like I can just see so many applications. And I think that human word-of-mouth enthusiasm has really started to work its way into people and um they're starting to use it more and see the application. Um, but there hasn't necessarily been a concerted, coordinated sort of effort to get everyone to use them. And everyone's sort of been doing it in their own silos, um, which maybe is not the best way to do it, but that is just how it's gone so far.
SPEAKER_03Yeah, I we've we find that that's already happening and when companies reach out and they say, well, uh we we know people are using it, we don't know what tools they're using, we don't know how they're using it. But the leadership's like, okay, like, yeah, we're we're we're using AI, we just don't know how. Um how would you, for those that are listening that are like, okay, this has inspired me, we we need to get serious about this. They heard the example from Australia. They heard uh the example of, you know, wow, I can wake up and there's tasks that have been run for me on my behalf. Who would you recommend they sit down with? Let's say I'm the owner of the company and like, guys, we got to address this. Who should I bring into that meeting?
SPEAKER_01Um I to be honest, I think you I think it usually like it starts around conversations of automation, right? So I think a common thing that happens at organizations is, and if if you're the one company that uh doesn't have this, please reach out. But there's always this feeling of we have too much work, right? So if you get management to a room uh into a room and you sort of list out the busy work, the stuff where you're like, this is just common data entry entry, it's like you know, uh not something that's really meaningful for me. It's not in it's not high impact work. Uh listing all of those things out and then making an action plan to sort of solve them, or how can AI solve them? And going into one of these uh LLMs and and asking it, hey, how can you solve it? I don't want to do this anymore. Can you solve it for me? And seeing what it throws back. The answer might suck. The answer might be exactly what you needed. Um, but now you've started. And you started on something that's practical and high leverage because it's busy work, it'll free you up to do more things.
SPEAKER_03Yep. So, listeners, great tip. Go into the room and say, like, let's list the stuff. You know that what those pain points are already. You got to eat enough to get started for sure. And then just go into models and say, I don't want to do this anymore. Like, how can AI help me? Like, that's not complicated. Any of you already have the pain point, just ask for help, right? It doesn't have to be an engineered prompt or anything like that. So that's great advice. So, Zach, um, if people like they they've engaged with this conversation, I've gotten a lot of aha's out of this. Um I'm now a little bit more committed to exploring Claude Code now to see how I can start to link all these things together. If they want to plug into what you're like how you're viewing the space and the acceleration of its adoption and all those types of things, where should they go to you know get more of your perspectives?
SPEAKER_01So um they can follow, they can subscribe to my Substack. It's a free Substack where I talk about all things AI and and marketing. It's called thoughtson.substack.com. Um so it's just basically my thoughts. I also post usually an abbreviated version on LinkedIn. So definitely always uh happy to connect or or or have a follow if you um yeah are interested in this stuff and and also just message me on I'm still on like uh X and Blue Sky and all these platforms. If you send me a DM and you want to talk through uh something related to SEO or marketing, I'm always here. I love this stuff. I let the nerd out over it. And yeah, if I can be of help, yeah.
SPEAKER_03Very cool. So, guys, I'm gonna put all these links in the show notes as well. So you should be able to scroll down on your phone and see the links or you know, check it out on our uh YouTube channel as well. Um, Zach, any like closing advice for somebody who's heard this and they're like, okay, I get it. It's time to do something.
SPEAKER_01I guess my advice to you was just start would be to just start. Like it's it's not as scary as you think, and you will you will surprise yourself in a month of what you've managed to figure out that you can do. Um I think uh that it's it's coming no matter what. And so the longer you put it off, it's you're just creating future headaches for yourself. And this is the best way to be proactive.
SPEAKER_03That's like along this um kind of advice that I give people look, the best time to plant a tree was 20 years ago. Next best time is today, because the time is going to pass, and if you keep putting it off, it's gonna be 2026, it's gonna be Q1, Q2, Q3, you're gonna get busy. Oh, we we've got people that know how to do that stuff. I don't need to know how to do that stuff. Wrong, wrong. Just get started. Well, Zach, thank you so much. This was um I enjoyed the conversation we had yesterday as well. Uh and this is like I said, this has inspired me to um say, what can I do with clawed code? What can I do with MCPs? Again, I'm not technical, um, but I'm realizing that you don't have to be a computer science nerd. Uh God bless all the computer scientists, I don't mean that in a derogatory way, um, to to really start to do some sophisticated stuff that you would just be out of your hands 12 months ago.
SPEAKER_01So awesome. Yeah. I think the question for you in a month might be what can't I do with cloud code?
SPEAKER_03I like that a lot. Well, Zach, thanks again for taking the time. I really appreciate it, and I'm looking forward to uh following your Substacks. I've uh I got a lot out of that, again, out of that conversation yesterday and today. So um, and everybody, thank you so much. We're gonna be putting all this again in the show notes. Uh, I would encourage you to follow Zach's advice if you're still sitting, regardless of where you are, even if you're you consider, oh, I'm a power user. Um, there's still so much more to learn, and it changes so frequently. Just plug in and start using the stuff, and I think before you know it, you'll be you'll you'll be beyond where you are today. So all right, everybody, we'll see you on the next episode. Thanks for tuning in to using AI at work. Don't forget to subscribe for more conversations about how to use AI at work. And a special thank you to our sponsor, Chief AI Officer, for empowering businesses with AI education and training. Visit their website for a free AI readiness assessment and AI strategy guide to help you get started using AI at work. That's www.chiefaiofficer.com. Follow us on Twitter at the handle UsingAI at work and visit www.usingai at work.com for free resources to help you harness AI in your role.