Ep 55 - Harnessing AI Tools: Insights, Use Cases, and Future Prospects

Watch the YouTube video version above or listen to the podcast below!

Ep 55 - Harnessing AI Tools: Insights, Use Cases, and Future Prospects Podcast and Video Transcript

[Disclaimer: This transcription was written by AI using a tool called Descript, and has not been edited for content.]

Alex Pokorny: [00:00:00] Welcome to another Enterprising Minds podcast episode today. Dave is out on some travel, so you have Ruthi and myself.

Diving into AI Tools and Use Cases

Alex Pokorny: Alex, topic for today is about using AI tools. What are some current use cases? How are you using them? What do you see as being, you know, most useful about them or missing functionality, something.

You're hopeful for in the future. so we're gonna mix a little bit of everything just to make it kind of fun and interesting. We're gonna mix a little bit of, you know, personal of kinda the fun and interesting things that we're making use outta these tools with, but also some work, you know, applicable stuff.

What you can actually do with these tools today to actually improve your workflow, your career. And your day. So earthly, do you wanna kick off or do you want me to,

Ruthi Corcoran: I'll kick off and I'll just add on as well. One of the reasons we wanted to chat about this is because, first our use cases and sort of what we can do and what we've been trying out are continuously [00:01:00] evolving.

And I think the other thing that's helpful too, of having, People to talk to that have similar skill levels is sometimes I see a bunch of AI use cases and it's like, oh, that, that feels out of reach or it's too much, right? Mm-hmm. Whereas if I know Alex and Dave are doing something, I go, oh. I bet I could do that, or at the very least I could, you know, they could break it down.

And so being able to compare notes with peers seems like a really helpful thing. Um, and as we go, if there's anything that, you know, strike, strike you as something interesting, feel free to reach out. And we can always put together a prompt library or set it out on pathways, um, as, Hey, these are things that we're trying, feel free to latch on to.

Personal Experiences with AI Tools

Ruthi Corcoran: so I'll start, just with some of the personal ones. 'cause it's a little, it's a, it is kind of fun. Um, nothing fancy on my end. and then maybe in a bit we'll talk about some of the various work, work related, uh, uses. Uh, the first and foremost one is, I know deep research isn't. New and I, it's not novel, or at [00:02:00] least it isn't, if you've been working within any of, you know, Gemini, Claude, chat GPT for a while.

I finally got around to actually playing around with it in a real way, and one of the, one of the things I was looking for was sunscreen. So I had purchased sunscreen at the beginning of the year for my kids and it did not work. They came home and it was. They were burnt and application was terrible and there was just a whole bunch of, just, it was pretty rough.

And you go to the, the shopping aisle and you look at a whole bunch of sunscreens and you can look at reviews online. but that only gets you so far. I said, Hey, we've got these fantastic new tools. So I said, let's try out deep research because I really wanna know the nitty gritty of sunscreen. And we think about this, there's so many different elements of it.

There's. Does the thing protect against the various UVA and UVB rays that we care about? Yes or no? Okay. But then we've also got this layer of, alright, there's chemicals that [00:03:00] are scary that may or may not be harmful, and there's a lot of uncertainty and ambiguity. I don't have a lot of time to actually go into the research to know like, is this actually harmful?

Is this just noise? Is this something that isn't harmful today but could have? Like there's a question mark 50 years from now, what's the trade off between those two things? and so what I did is I structured a prompt that essentially said, Hey, you are a product development team creating a new product.

F for sunscreen for kids. Mm-hmm. Talk me through what is the, what is the academic consensus around, best, pro ingredients for UVA and UVB production? also talk me through if there's any promising new research that's upsetting the academic consensus, then talk me through sort of health effects, um, or known long-term health effects.

From, again, the academic consensus, but then talk me through if there's any, any sort of new research coming out. there was [00:04:00] another one on, I think, oh, application and ease of use, right? Because if you're putting it on kids, it's like, is it easy to put on? Is it hard? What do we need to think about from an ingredient perspective in terms of effectiveness?

And then I said, okay, so first do that research, then tell me what your list of ingredients would be for a new product and why. And then tell me what the competitors are, landscape against that, that particular ideal ingredient list you have, and tell me how those competitors, um, different products are better or worse than what you've put forth.

Alex Pokorny: Wow.

Ruthi Corcoran: Super in depth, but what it doubt was, yeah, that's a only love what I wanted to know.

Alex Pokorny: Yeah. Oh, that's interesting. I, you know, I've got a similar, I was going through a little bit of my, I use chat GBT pretty heavily, claw, Gemini, lesser extent, copilot when I need to, there's like one or two prompts honestly I find useful there.

Um, but I was going through [00:05:00] my chat, GBT kind of history, thinking about some of the more personal cases of when I was pulling it up and. It is those, you know, it's the, the two search, three search Google queries that I kind of find from an SEO lens. I'm always thinking about that kind of like, you know, how would I normally, you know, provide this data?

How would I normally search for it? What kind of content would be useful for, I mean, my mind's always a little kind of twisted that way. but it, it was each time I've kind of been looking at these things, it's not the yes or no. I mean, honestly, like the AI overviews thing is great in Google, and if it was those kinds of questions, I consider those questions almost googleable questions versus something that is in depth.

And if it is, I need to know this piece of information and this piece of information. That suddenly negates Google. Like, I'm, now, I'm looking for a new source, or I'm, I'm prepping myself for like, okay, I gotta search this one thing, read three different blogs, figure out what those details [00:06:00] are, you know, can you know things of points of interest that I should be interested in and try to figure out, you know, the real solution.

Okay, now I need to search for that one. And then try to figure out like, kind of you go down this, this like user journey kind of path of query after query after query until you finally get the same information and you're trying to basically like, okay. I, I got that part. Let's, let's pull it together already.

Try to give me something that summarize the package so I don't have to go through that process piece by piece by piece. Yeah.

Challenges with Online Searches

Alex Pokorny: And also Google search, honestly, I gotta say just has gone downhill. Like there's so much crap content, like almost anything. I'm searching now. Oh man. Yesterday I was searching for URL transformations and transliteration, right.

It's where you take like a German word that may have a nom line in it. You put instead of letter U because Latin characters what URLs are made out of. It's the way it's set up. You have to figure out what that transition is. Right. And all I could get was religious articles, transliteration of [00:07:00] URLs is literally my query transliteration of, and I got the transfiguration of Jesus Christ done 10 times and I was like, I literally have in URLs in my search query.

Why am I getting anything religious? Like it was so bad. Or I get so many, you know, I search for something and it's just filled with top 10, best this, top 10, best this. And I'm like, I don't want to,

Ruthi Corcoran: I want don't. And it's the same list that somebody copy pasted or syndicated out to a whole bunch of websites.

So, you know, you had recommended to me, The, co intelligence episode on Our Art of Manliness. yeah. Which is, um, an episode with, um, around ai, of course, with a professor out of Wharton, I believe. Yeah. Ethan

Alex Pokorny: Monik.

Ruthi Corcoran: Oh, say it again.

Alex Pokorny: Ethan Monik.

Ruthi Corcoran: Yes. Who's fantastic. Yeah. and one of the [00:08:00] suggestions he had when, when sort of working, um.

creating prompts and asking questions was sort of, you kind of have to lay out the thinking mm-hmm. To get better results. And so that, this was my first application of that, of here's sort of the background information that should be considered along the way. Hey, go take a look at peer reviewed articles and things like mm-hmm.

And, and also I put an explicit note. Cite your sources along the way so I could go check out. Sure. Is this something that was an often cited thing or was this more of an obscure, PubMed thing?

Alex Pokorny: Or hallucination.

Ruthi Corcoran: Or hallucination,

Alex Pokorny: yeah.

Ruthi Corcoran: so he said, you know, walk through in sort of a logical way I. How you want it to think about the problem and then ask the question, which I think comes back to your point about Google, which is you say like, gimme a product recommendation, and you do get these top 10 lists, but you also don't get the why.

Or if you do get the why, it's on a two sentence superficial level where you go, well, [00:09:00] I don't even know where that information came from.

Alex Pokorny: And the bias is actually a concern there too. So. In all directions actually. Really? Mm-hmm. But I get really concerned every time I do a product related search best, whatever.

Um, I was looking at, uh, smart watches recently, trying to find a really small, lightweight ones versus some of these giant ones that are out there that I can track everything, but I don't care. I just want something familiar with the basic, and. It was so funny because it was the same problem. I basically kept going list after lister list, you know, listicles, basically these, you know, top tens.

And if you read through all of it, you finally got to the disclaimer at the end, this article was sponsored by da, da, da. And I was like, oh, come on. Like, you know, it was always pretty obvious because suddenly the top 10 is only like one brand or two brands, and I'm like, There, there's more than that in this, this market.

Like there's, there's a bunch. [00:10:00] So why, why is this one being pushed? Or you get, like, you look at three of 'em and one of 'em has very weirdly, one of 'em, you know, right near the top where the other two had said, no, this, this one's a dollar. I mean, it's terrible. And the other one said, oh, it's great. And it's like, oh, there's the affiliate link and the other ones don't have affiliate links.

You're just trying to churn affiliate money for these articles. So, and that one was the one that you got a, a tag for. That that's just. It destroys the value of it, and it wastes my time. That's kind of it. It's, it's a sales pitch. When I didn't ask for one, I wanted an unbiased critique. I, I was lucky for a true recommendation of, you know, somebody who's a slight expert or authority of some sort.

I.

Ruthi Corcoran: Yeah. And,

Alex Pokorny: and so I went over to chat GBT basically again asking you for giving a couple recommendations. Show me in that case, show me a chart showing what the different points are, price points, ratings. I want links to their, since these were common products, I want links to their Amazon listings so I can check out the ratings and kind [00:11:00] of double check that those stars, you know, four outta five or whatever actually is four outta five.

there's so many sites out there that now post ratings that are. Brand ratings or offsite ratings or something like that, just to make sure that they can get that, you know. search snippet and yeah, be ranked a little bit better. So,

Ruthi Corcoran: you know, that also reminds me of another use case I've had recently that's failed.

AI in Interior Design

Ruthi Corcoran: And then I, I wanna hear some of the ones you've had as well, which is, I have been playing around with creating an interior designer agent, so. Okay. I am, uh, we're gonna be moving here shortly. and I'm figuring out, we've got a, it's a slightly bigger space. Um, we gotta figure out, okay. Like we got, we need some key furniture.

I wanna think about paint, I wanna think about rugs. Like all the things Yeah. Might go into to making this new, um, house a home. Create an interior designer agent, and the idea was, okay, it'll be based off of some interior designers who aesthetic I already [00:12:00] like, mm-hmm. I'll give it a little bit of prompting or a little bit of information and context around the types of things I like, types of things I don't like.

and let's see. Let's see where this evolves. Right. Let's see how we can sure miss into something that I can have a continuous conversation. so far it's, it's not been a great success. the areas where I found a lot of value, I didn't really need this agent. you know, in terms of, Hey, help me think about how I might ask for this new contractor's price on painting.

Or what considerations do I need to be in the, in the bid that they put back? Things like this, sure, but I didn't need this. And the areas where it's, it keeps asking me if I want product recommendations and I occasionally will say, sure, why not? Flooded with Wayfair, with, with Home Depot, with Walmart, with Amazon just flooded.

And at one point I just said, do not put any of these in. Yeah. And what I found was it found way, well, it didn't find ways that's anthropomorphizing a bit, it was still giving me the same [00:13:00] types of product recommendations, even though I had explicitly told it not to do Wayfair, Amazon, et cetera. like I was looking for more niche or.

Yeah. Is like pulling teeth to get it to recommend even a pottery barn and like, that's not like a, a small niche. No,

Alex Pokorny: no. And so

Ruthi Corcoran: I wonder if back to your listicles, if there's so much of that sort of, dare I say, junk web content. Yeah. It's even been sort of flowing in that, it's almost, it's a, it's a pretty big challenge to get it to, to.

Do what it is I want it to do in this case.

Alex Pokorny: Absolutely true. And. That's been a struggle. I mean, it's kind of US based retailers that we've mentioned, but a couple of 'em, like Home Depot, which is a home improvement store with lumber and nails and screws and all the rest, also has this absolutely gigantic e-commerce side to it, which is a hundred percent shipped in directly from manufacturers.

Has nothing to do with the stores being shipped into. [00:14:00] And it has basically the same list that Wayfair or Amazon would have, you know, millions of products. Which also changes the reputation of a retailer as well, because it's no longer the reputation of the retailer's brand and the choices of products that they've decided to put inside their store.

Instead, now it's, let's just pack it with everything.

That's also Wayfair's model, which, you know, gets into some really junky stuff. This was a funny one. I was on Etsy yesterday and, uh, looking to build a new project. It's a headboard for a bed and it kind of, there's some interesting kind of creative solutions that people have done and some fun woodworking ones.

And I'm kind of looking for a woodworking project about that size and complexity. So I was looking at different ones and I started going through Etsy, because there's a lot of woodworking plans that are posted there. A lot of people try to. You know, make some revenue off of that. And I, I do buy a fair number of plans from people, creators that I support.

I modify [00:15:00] everything heavily, but I still like to support them because, you know, I got inspiration from them and it's like five, 10 bucks. Like, it's just something to be like, you know, appreciate that you put together a video on this. And it was AI generated imagery and AI generated products. These were insane fantasy.

Like wood drift, wood twisted somehow in a bizarre way, and all natural from an ocean somehow. But you can order 10 of 'em and they'll look the same. Okay. And the whole, like the photos of the product in use were completely just a fantasy rum. Like you could notice like little things of like, you know, certain.

Blankets were just merging into the pillows like and stuff like that where it's like bad ai, you know, image outta curiosity.

Ruthi Corcoran: Were they also selling the rainbow colored roses? Oh my gosh. That

Alex Pokorny: hills or the black roses? Yes. Yes. Oh man.

Ruthi Corcoran: Neon blue. Yep, neon

Alex Pokorny: blue, right? Yeah. [00:16:00] That's amazing. You can grow roses that literally every pedal is a different color.

Except, wait, no, that's Photoshop. no, it, it was ridiculous. I mean, I was just starting to go through it, but it started to get really disappointing too because I had a hard time understanding some of these products and product shots of, of the better, more or more basic listings, which had less in product usage kind of ones.

There's just the product itself. Is that actually a real product or not?

Ruthi Corcoran: Yeah,

Alex Pokorny: I, I, I don't know, like some of the stuff is like, uh, ratin or wicker wrapped stuff, and it's interesting to look at and all the rest, but at the same time, it was like, I, I don't have enough knowledge around the area. I don't have enough, you know, photographic or video evidence to prove that this thing is real or not.

And it says it's, you know, $700 or something like that. And then I'm like, there's no way I would have the level of trust. Yeah, to be able to purchase this item.

Trust Issues with Online Products

Alex Pokorny: Now that I've recognized that there is AI generated content being posted [00:17:00] on the site, like the trust factor, sorry, just kind of honing in on that a little bit of,

Ruthi Corcoran: yes,

Alex Pokorny: some of these product recommendations and just the struggle there.

Ruthi Corcoran: I was also thinking about the superstar effect this could create or a variation on it, which is I notice my reaction to what you just. What you just articulated, the experience you just had on Etsy, is to say. Well, now my level of trust for any new emerging small creator is so much lower because I don't know.

Yeah. So what am I gonna do? Yeah. I'm gonna go to the es to the known, established, the, the companies I've purchased from in the past. Mm-hmm. Where is it the best of the best? Is it the, the latest? And maybe, maybe not, but at the very least it's trustworthy.

Alex Pokorny: Sure.

Ruthi Corcoran: And what does that, what does that do for the Etsy of the [00:18:00] world, uh, where

Alex Pokorny: it's harder.

Ruthi Corcoran: Building trust is step one, and it's, it's an uphill battle as it's,

Alex Pokorny: yeah. And yeah, I don't really see that struggle really ending either. Mm-hmm. Like I tested out Timo and a couple other ones, again, around woodworking stuff, but I didn't buy electronics from them. I didn't buy anything complex from them. I bought really, really basic yeah, like metal tools that like our three parts put together kind of a thing.

Again, not trusting that the quality of anything more complex or more precise would actually be accurate or good. Yeah. Um, so I thought, well, you know, if I buy something really basic, it can't be that bad and No. Yeah. It can be. I was so sad up that little junk and I was like, have you

Ruthi Corcoran: ever bought a $5 spade for your garden?

Alex Pokorny: Oh my gosh, no. Yeah. Just bends. Yeah. I could imagine. Yeah, there's a price floor to some of this stuff, but the difficulty, again, from the the market standpoint, I can understand too of being, [00:19:00] there's thousands of manufacturers out there, how are you really vetting each and every one? Unless you're doing some sort of like private label thinking of like Costco has Kirkland and a few other, you know?

Mm-hmm. A lot of companies, retailers have their own house brands that they've developed, but it also means that they've worked with those manufacturers and seen those products and decided to put their name on it, you know, creating their own little. Universe of products on top of their market.

Ruthi Corcoran: This is such a cool tangent, but I also wanna hear what, yeah, I've been trying out and testing, so we'll move on, but we might come back to this in a later episode.

Exploring IKEA's Interior Design App

Alex Pokorny: Yeah, a random interior design one. I struggled through a whole project for that. Ikea. IKEA has a super cool. interior design app because you take a photo of like your basement or some living room and then you can delete everything out of the room and then place in their products and size appropriately.

And all the rest stuff snaps to like wall in corners and stuff, and you've gotta re read. Look at areas. I found it [00:20:00] useful because I, you know, a house that we live in has, you know, paintings on the walls or pictures on the walls. Kids aren't everywhere. Kids toys, I mean, you name it. But I was able to take a photo and to wipe it all out and it gave me at least a blank slate.

That is my room, uh, is that room. And then I was able to kind of like, okay, kinda reimagine the space. Doing that you have an empty house. I mean, different situation. But that was kind of cool.

Ruthi Corcoran: That's, that is really cool. or even just being able to, to place the furniture in the room. I, I have yet to actually use one of those apps, but I will be in the next couple of weeks.

So appreciate

Alex Pokorny: I, I've struggled to find a good one. Whenever you're actually trying to place, get visuals of a room and actually layouts and all the rest,

Ruthi Corcoran: which is so important 'cause you've got the scale question of, you know, making.

Alex Pokorny: Yeah. And I think that's the piece that hasn't been quite developed quite yet, is not just creating the image in the first place, the diffusion aspect.

You know, what pixel should be next to what pixel, but instead of what pixel next to what pixel. Okay, now we have [00:21:00] complete images. Okay. Now we have to analyze those images and properly resize to fit. I've seen some that just try to paste it on top, that kind of thing, and it's just,

Ruthi Corcoran: yeah.

Alex Pokorny: Doesn't really work.

Ruthi Corcoran: This is

Alex Pokorny: dubious. Yeah. All right. Uh, other ones you wanna switch over to work stuff or, yeah.

Ruthi Corcoran: Yeah. Let, let's do it. Okay. You kick off the work stuff and then I'll take on a couple.

Alex Pokorny: Okay.

Challenges in Modern Marketing

Alex Pokorny: I've got a, a fairly small team at a fairly small company, but we're very nimble. what that does mean though, I have a lot of freedom in the role of changing the way we're gonna do our work.

literally from, we could change it from week to week if we really wanted to. We could just notify our stakeholders and now we're doing something different. Like we can move very, very fast that way. It's kind of fun. Awesome. the difficulty of course is we're trying to do everything, kind of boil the ocean approach instead of boil a teacup.

Right. And we're struggling with the ocean. Turns out it's kind of [00:22:00] big, I would've thought. I feel like there's some people that might've noticed that a few thousand years ago. but we're figuring it out ourselves and yeah. So I've been trying to create a number of kind of scaled systems to handle. The situation that we're in, which is modern day marketing has a giant MarTech stack.

And trying to keep a handle on all of these different systems and all of these different programs, all these different tactics and all the rest is very difficult. Uh, we have agency partners, but even with that, I want to have oversight over those agencies and seeing, you know, actually, are they running these paid search accounts well or are they just let and run and then every month is kind of questionable data.

You know, it's like you gotta have. Some insight all over the place, and if issues arise, you have to be able to move on it fairly quickly. so part of this is one I've been trying to create standardized automated reports, [00:23:00] been fairly successful with those.

Automating Google Ads Reporting

Alex Pokorny: So Google Ads has Google Scripts and there's a bunch of 'em out there, but I had.

Judge GBT write one that was custom to what I wanted. And it took a variation of some months that they had in the past, which was checking for four fours on landing pages. every 10 minutes it checks it and they gives me a daily summarized report of how often they occurred, missing tracking codes and final URLs, which we had a problem with that for a couple of 'em.

So that was awesome that it flagged it. Um. Not enough spend to cover like the budget was off so it wasn't enough spend and it wasn't spending thoroughly enough, uh, high number clicks with low conversions, like a sudden change there, you know, kind of things like that. Or sudden spikes, you know, there's some thresholds and some stuff like that.

It checks a few other things too, but mostly focused on the four fours, the missing tracking codes, kinda the bare basics kind of stuff, because I like my reports or any report, honestly, that gets [00:24:00] shared out to be one short sentence summarized. So the sentence that I was going after was, is my. Fill in the blank of tactic Google ads screwed up.

That's what I want the report to say. That's it. It is a confidence boost that basically says I don't need to log into that system and run through a bunch of tests myself, and that's all I wanted. Was a confidence boost that that was being handled enough, that it wasn't screwed up? Not not recommendations, not enhancements, not expansion.

Nothing like that, just. Is it screwed up or not? Like, do I need to deal with this? Is this a fire? That's it. That's it. And I like reports to be stupidly simple, down to that summarized, uh, moment. Because a chart that is that simple can be reused and shared with a whole bunch of different audiences. If I'm doing my own investigation work, I'm gonna be pulling crazy amounts of data and very, very little ad hoc things and I never really, probably won't ever see it again.

It was [00:25:00] just a question I needed to answer. I wrote, go down that rabbit hole. I figure it out, I move on. But for routine reporting or presentation reporting, every chart should be simple enough that I can basically display this to somebody who doesn't really know and they get it and they get the point of it and they move on.

And if it has to be three different charts, it should be three different charts. Don't make one super complex 'cause that's crazy to do to people.

Ruthi Corcoran: Couple of questions. How, like, talk me through some of what you did and how you did it. For example. Sure. An agent, is this a prompt? Do, are you, attaching a spreadsheet with information?

Do you have in integrations? What tools are you using?

Alex Pokorny: Absolutely. I can work for this. Um, so first off, what's basically is I worked with chat GPT trying to make sure that I had my couple of ideas. And really what I wanted was like, I just need to know if it's screwed up or not. What does that mean? I. What are some things that I should be checking on my Google Ads accounts to make sure that they're running correctly?

I do not want recommendations for [00:26:00] improvements. I don't want recommendations for enhancements. I just need to know kinda a status of them. so I had listed out like three different things. It added like a maybe one or two more, so it became a better list. Okay, so now I can use that in Google Ads. Google Ads has a section called Scripts.

They're under the bulk edits section. You click that and it literally just has an empty window. You paste in your script, you try to run it. There's problems. You go back and forth, back and forth, back and forth, which at GPT saying, Nope, this didn't work. This didn't work. This needs to be authorized back and forth.

about an hour later, I got it to finally work. And what it does is it runs that information and it pulls it into a, uh, Google sheet, and it basically just appends the bottom of that same Google sheet and keeps basically adding it with a date stamp on it. So that's what it does in, uh, Google Sheets.

There's also, if you click towards the toolbar, there is Google apps and Google apps scripts basically. And that is a [00:27:00] separate thing in kind of the Google Sheets area. So that's the interface you're using then? same thing. Basically it's a blank window you can paste in a script and I'll run that automated script.

Um, so what that does is it takes this, list of data that's basically being appended inside this Google sheet done on a daily basis from Google Ads. So it's feeding that sheet now, the script formats that data and emails it to a list of people that I can easily add more people to It, took a bit to get it to email out.

But it does. So every single day there's a group of people in my company who receive an email that says Google add status, a little hyphen, and it says the number of issues found. Um, so if it's zero, you just delete the email. If it's 10, you got a prom. Do false flags happen Occasionally? Yeah. But.

Honestly, the script has been super solid. it's not wrong. It has always been a question [00:28:00] of, no, we're doing this in a new way or a different way, or something like that. That's a new modification. It's not that the data that it actually pulled was inaccurate. It was that we've changed our methods, so it had to be updated.

Been super useful. It, it found things on the first run of stuff that's actually been hanging out there for months and we didn't catch it. So super useful agency. Didn't know, nobody knew some stuff that should have been fixed. Yeah. Super handy. Um, so I'm creating another one. I. kind of keep going down the line.

Integrating Multiple Data Sources

Alex Pokorny: So Bing ads as an API, same thing, trying to pull that data into that same Google sheet, and it's being appended to that same email. So now the email basically has Google ads and Bing. And then the next one is gonna be an SEO one, which is partially based off of ga, which I've got the 4 0 4 report done from ga.

It's in just an automated report that it's doing.

Ruthi Corcoran: Yeah,

Alex Pokorny: and it's just an exploration inside GA four. I [00:29:00] just had to find. Our four fours page titles get translated. So we have like 18 versions of four, four page not found, you know, in every ation, you know,

Ruthi Corcoran: always good.

Alex Pokorny: Yeah, right. You know, kind of hard to make a rule.

but luckily they all have four, four and then a space and we don't have a product ID that pops up like that. So it is all our four fours happen to have that same piece in them. So I could say in the page, you know, name. Using GA four language, four, four with a space, find me everything that, and then I made an exploration report from that that again, kicks over to basically a Google Sheet, Google App script grabs that puts in the same Google sheet as the email one.

So now they're both there together. And then again, the Google app script gets updated again to basically email that out to me. So it says, if I have four oh fours across the entire website, at least if GA four picked up. There's other [00:30:00] reasons why GA four doesn't always pick up stuff, but, so that's that one.

Uh, GSC is one of the, the later editions. And then I've got Screaming Frog, trying to get that one, um, running. So that is, uh, screaming Frog has a scheduler running bats against all the URLs inside my X ml. Uh, site maps. So it just starts at all. My under Screaming frogs includes you can list out all the different site maps, which we don't have a site map index, so we have to do 'em all individually.

so it hits all of those, it crawls every single URL that are in my site maps, and then it basically creates a, a report and Screaming Frog has a scheduler, so it runs automatically. It runs headless, which basically means it does not open on your computer, it just runs in the background and it's super fast compared to the actual user interface that you normally would work with.

It runs like 15 year Ls a second versus like. Three super fast to run it. So anytime actually you're doing [00:31:00] bulk crawls really nice, if you can run the thing headless because it just exports the report. You, you literally have a folder, you have like your, uh, window explorer on your, uh, desktop open and you've got the folder and then something like a new file just pops into it and you open the file and there's all your data, like just crazy.

You can also get it, which is part of this project, to automatically upload to, To Google Drive, and that's where my sheets start to appear in Google Drive. And then same thing, can hook it into the sheet, can hook it in the email.

Ruthi Corcoran: This reminds me very much of, um, a conversation I was having over the weekend.

I, so I remember a few years back. You had shared some of the, the cool connections that you could do when you, you, where you could share various data from Google products directly into Google Sheets. Yep. And it would all, you see the data, manipulate the data and, yeah. That a lot of that automation was already present, but what what seems to make this [00:32:00] moment unique is that it's sort of a convergence of multiple technologies.

It's bringing that automation in with the, the natural language processing. Yes. And perhaps some of the other marketing automation activities that may have just in incrementally improved and now it's allowing you to do a step further that feels like a much larger step change.

Alex Pokorny: Absolutely. And yeah, the old version was basically it's Google Ads has these different scripts, but they were pre-formatted, they're written by Google themselves or various, you know, people online through LinkedIn or whatever.

And it was just whatever it was, was what it was. Right. I wasn't, you know, smart enough with, or knowledgeable enough with, in terms of the code to be able to manipulate it and change it. But yeah, now it's completely custom. It does exactly what I want it to and does nothing else that I don't want it to, you know, that kind of thing.

Yeah. it is connecting all these different products together.

Leveraging AI for Troubleshooting

Alex Pokorny: There's probably other ways that would be easier to do it too, but it's also troubleshooting for me. Like I ran into so many issues with Screaming Frog even, and I [00:33:00] would ask Chad, you know, I got this problem, what's going on? Or Google App Scripts has a weird authorization.

I need, I don't know what this means. It's like, oh, it's because of this and it gives me the reason behind it. Like, there was so much information that would've been, you know, rabbit holes on forums online. Yeah. To try to maybe understand why this random flag is happening and even doing, and also I should say this isn't seamless.

It is way better than it used to be. Chachi VT has improved a lot. I've done similar projects even like three, six months ago and is way better than it was. but even just getting the ads one out that just said the four oh fours and problems like that, yeah, that took me like six or seven hours to do straight and I'm familiar with this stuff.

Yes. Not easy. And getting the, even the 4 0 4 1 from GA integrated in this and sent out, that was probably another five hours on top of that.

Ruthi Corcoran: I mean, that's a good call out of some of these projects. They're not [00:34:00] easy.

Alex Pokorny: I. No,

Ruthi Corcoran: they're not like a, oh, quick, just do the thing. But, or an

Alex Pokorny: intern like, Hey, intern, go do this.

Ruthi Corcoran: Right. No, that's not fair. You have so much background knowledge.

Alex Pokorny: Yeah, exactly. Right. But

Ruthi Corcoran: it's making things possible that didn't exist before.

Alex Pokorny: Mm-hmm.

Ruthi Corcoran: And there are some things, like your comment about the troubleshooting, that's, that's half of the use cases I have at work is like, yeah. I, I, to your point, they're in this forum, or the company has documentation, it's public facing.

You could go find the documentation. Mm-hmm. But. Saying, Hey, copilot. Even, I know, I know we talked about copilot last time, but even copilot, I can, I can toss in documentation or I could say, go find the documentation and it can come back to me with something that's relevant to the question I'm asking.

Alex Pokorny: That's extremely useful. I mean, especially in the world of remote work, I, I feel, I use. LLMs more frequently for that than probably anything else, which is sounding bored kind of things where I have [00:35:00] an idea or I gotta, you know, I got a need of some random kind of sort gimme something to react to basically.

And it would totally, if it was in person, be the times that I would look around the office and say, that person looks super bored. Hey, do you mind grabbing coffee and we can talk through this. Or can we go on a walk quick? I need, I need to like work out this idea and I can't quite. I, I need to kind of like, you know, parse it out somehow.

Yes. And in remote work I always feel like I'm bugging somebody 'cause I don't know what they're doing.

Ruthi Corcoran: Right.

Alex Pokorny: They could be really, really busy. And I really don't want to just frustrate them by constantly paying somebody, being like, Hey, do you have five minutes to talk about a random topic? That really just affects me,

Ruthi Corcoran: which are still good to do because it builds connection and they might have been thumbs, but it's a, it's a really good point of you have no idea.

You have no idea of knowing. Yeah. the, A related one, That I wanted to chat a little [00:36:00] bit about is, um, some of the troubleshooting thing. 'cause also some of the information finding things you and I both work in a, in a world with. Um, I. to your point, you there's stuff in forums or it's through documentation that perhaps wasn't written for either just the average person or potentially, um, it, it was written for a specific purpose and now you have to use it for a different purpose since you've gotta do that contextual piece.

that's been another huge use case that I've had, um, even just in the last couple weeks. Hey, you know, there's, um. June 28th, there's some EU accessibility requirements coming into place. We gotta make sure we're compliant. We've done audits. That's fantastic. Boy, the WIC ag documentation is, yeah, something special to read through.

It's nuanced. It's, it's, sometimes it's clear and sometimes it's not. And being able to say, Hey, copilot, even, here's, here's, um, the [00:37:00] section of the, the documentation, here's the bit of code that we think is, um, not compliant. How do we need to tweak this in order to fit that compliance? Spam right there. It's, it's there and we send it into Jira.

And what I found really helpful is it, gives you options. They can say, oh, here's three different ways you might be able to solve that problem. Nice. And here's the pros and cons of those. This is stuff that would've, to your point, an intern couldn't have done it or it would've taken a long time for somebody to sit down and do it.

Um, and instead I was able to do it, you know, 10 minutes, 20 minutes if I was really sort of digging in and curious. And, and we can move on. It's an important thing that we want.