# SEO Rockstars 2026: Day 1 - Dan Kurtz

**Video URL:** https://www.youtube.com/watch?v=y1felTXj_OU

---

## Transcript

**[00:04]** Right on. Hey guys. So, as Mike said, I

**[00:08]** do a lot of automation stuff. Um,

**[00:10]** constantly dabbling in things, figuring

**[00:12]** out new tech, new stacks, new whatever

**[00:14]** in order to get the results that I'm

**[00:16]** looking for because I hate doing stuff

**[00:18]** and I want computers to do it for me.

**[00:20]** Anybody who's been vibe coding probably

**[00:22]** feels the same. So, we're going to kind

**[00:24]** of blitz this. Um, I'm going to show you

**[00:26]** guys some examples and stuff. I'll give

**[00:28]** you guys all the prompts, the systems,

**[00:29]** the GPTs and everything else at the end.

**[00:31]** So you too can go home and go do it

**[00:32]** yourself.

**[00:34]** So

**[00:38]** obviously core problem with AI

**[00:40]** automations, most implementations wind

**[00:43]** up failing or your NAN crashes or your

**[00:45]** make automation doesn't work right or

**[00:48]** clog code does something with one of the

**[00:50]** files and you can't [ __ ] open it

**[00:52]** again. But

**[00:54]** >> damn, what' you say? the cloud code.

**[00:56]** It'll go edit a line of code and then

**[00:58]** the whole app crashes.

**[01:01]** >> Everybody's favorite.

**[01:02]** >> So that's why I said you

**[01:06]** [ __ ]

**[01:06]** >> Yeah.

**[01:07]** >> So

**[01:09]** obviously if you can figure out the

**[01:10]** problem, it's a lot easier to build the

**[01:12]** tools around the problem and then

**[01:14]** eliminate the problem. So we're going to

**[01:15]** kind of go through some of those today.

**[01:18]** Vibe coding you guys are all familiar

**[01:19]** with probably. Anybody in here not vibe

**[01:21]** coding on some platform? All right,

**[01:24]** cool. So basically

**[01:25]** >> the only one

**[01:26]** >> no Jose

**[01:30]** sitting in front.

**[01:31]** >> Yeah. So you can do a ton of stuff with

**[01:34]** it. Um a lot of people have been

**[01:35]** building apps and applets and deploying

**[01:37]** stuff on PCEL and cloud waves and all

**[01:39]** different sorts of sites and building

**[01:41]** out these massive projects. Um

**[01:45]** far than that. So I decided to go ahead

**[01:48]** and figure out how to do arbitrage using

**[01:49]** AI.

**[01:52]** So basically go describe the problem in

**[01:54]** chat GPT or cloud AI generates the code

**[01:56]** run sandbox do the feedback iterate till

**[01:59]** it works deploy and monitor. For this I

**[02:02]** just use cloud code. Um and I built two

**[02:05]** different things. I built a mini app and

**[02:07]** then a companion a uh companion chrome

**[02:10]** bookmark lit to do all the work. So or

**[02:14]** extension I think they're called. So

**[02:17]** went ahead built an extension and this

**[02:20]** thing goes out and basically

**[02:23]** does everything it says here. So it does

**[02:24]** competitor research. Basically I wanted

**[02:26]** to figure out how to build a second

**[02:27]** income stream using AI and didn't want

**[02:30]** to spend more than 30 days uh 30 minutes

**[02:32]** a day doing it. Spent two weeks coding

**[02:34]** it but saved 30 minutes a day. So

**[02:38]** decided I wanted to resell stuff on

**[02:40]** eBay. So build out my app. This lives

**[02:43]** inside of Chrome as an extension. does

**[02:45]** competitor research, figures out who's

**[02:47]** selling stuff, what products. Basically,

**[02:49]** it's a it's an ecom sniffer. So, it goes

**[02:52]** out, you plug in your competitors,

**[02:54]** whatever sellers you're going after,

**[02:56]** shows you all their items, their

**[02:57]** purchase history, what they're currently

**[02:59]** doing, scrapes all their products,

**[03:01]** figures out the best ones, and then goes

**[03:03]** and finds the pricing for them and

**[03:05]** alternatives on Amazon. that goes in,

**[03:07]** does browser automation, runs through

**[03:09]** the whole process, figures out the

**[03:11]** markup, the markdown, the margins,

**[03:13]** everything on the fly, and then I just

**[03:16]** paste it into the web app, which is just

**[03:19]** this screen. Um, and then I just let it

**[03:22]** rip. The numbers up in the top right

**[03:24]** hand corner is the limit of how many

**[03:26]** products per month I can push into eBay.

**[03:31]** So, example, found a random product.

**[03:34]** just picked something off the bestseller

**[03:35]** list this morning, add it to the

**[03:37]** presentation, and then went ahead to

**[03:39]** search that same thing on eBay.

**[03:42]** Found out that yes, people are reselling

**[03:44]** it on eBay. And that little blue line

**[03:46]** down there that you can't really read

**[03:48]** from back there because picture's too

**[03:50]** [ __ ] big. Tells me my break even

**[03:52]** point on that product. So, it's doing

**[03:54]** all the calculations on the fly. And

**[03:56]** then I'm importing and the AI is doing

**[03:59]** all the product matching for me.

**[04:04]** So, what that looks like,

**[04:07]** because everybody here likes rankings,

**[04:10]** here are my current sales ranks in all

**[04:12]** of these categories. And I've been doing

**[04:14]** this for two weeks.

**[04:18]** I am not doing anything. I'm not

**[04:20]** pointing back links. I'm not writing

**[04:22]** content. I'm not doing anything. I have

**[04:23]** the entire system on a learning loop.

**[04:26]** So, the AI is continuously figuring out

**[04:28]** what's working and what's not. importing

**[04:30]** products, deleting products, selling

**[04:32]** products, doing the arbitrage, going

**[04:34]** back to Amazon and buying them and

**[04:36]** shipping them to the customer for me.

**[04:37]** >> Can you tell us what that means?

**[04:39]** >> Sure. Yeah. So, that's my seller rank in

**[04:43]** each of those categories. So, so out of

**[04:46]** 21, 291 sellers for rugs and carpets,

**[04:50]** I'm number 600.

**[04:54]** >> Okay. So it's out of however many people

**[04:56]** are actually selling that that's and you

**[04:58]** can figure out the percentile on that

**[05:00]** but it's not the same as SEO but it

**[05:03]** sells products

**[05:06]** and you can see obviously the machine is

**[05:08]** learning very well. Um I don't do

**[05:12]** anything with it at all. It figures

**[05:14]** itself out. It does the posting and I

**[05:17]** took the screenshot at like five or six

**[05:20]** this morning. So today's impressions are

**[05:23]** a little bit low, but they'll catch up.

**[05:25]** So in the past week, week and a half,

**[05:29]** 350,000 impressions, 20 products sold,

**[05:31]** and I've done nothing.

**[05:36]** >> Well, because the system is going

**[05:37]** finding the products.

**[05:38]** >> No, I wait until somebody goes and buys

**[05:40]** them, and then I go to Amazon, order it,

**[05:42]** put their shipping address in, and send

**[05:43]** it to their house in an Amazon box.

**[05:47]** >> I missed that part.

**[05:48]** >> Oh, yeah. I'm not I I have zero

**[05:50]** inventory until somebody buys something.

**[05:52]** I my cost is zero.

**[05:55]** >> Yeah.

**[05:55]** >> So, are you still having to physically

**[05:57]** buy it from Amazon or do you have that

**[05:59]** automated?

**[05:59]** >> Oh. Oh, no. It takes everything that

**[06:01]** they put in the address, shipping

**[06:02]** information.

**[06:03]** >> Okay.

**[06:04]** >> Yeah. So, it's completely automated.

**[06:06]** >> It goes and buys it

**[06:07]** >> and then it just ships it.

**[06:09]** >> Yep.

**[06:11]** >> How does it pay for it?

**[06:12]** >> My card. I have the card on file. It's

**[06:15]** right there on Amazon. It does the

**[06:16]** browser automation. picks that card.

**[06:18]** >> Wow.

**[06:19]** >> Dumps it.

**[06:21]** >> And again, it's kind of early. I've only

**[06:22]** been doing it for like two weeks, but

**[06:24]** I'm sure once I get up the speed, it'll

**[06:26]** be a little bit better.

**[06:28]** But, uh,

**[06:30]** >> so, and again,

**[06:30]** >> that's what you made this week.

**[06:33]** >> Yeah. So far.

**[06:34]** >> Yeah. So, 813 after taxes and fees from

**[06:37]** eBay because they take out like 14% plus

**[06:39]** tax.

**[06:41]** >> So, my markup is 60%. After they take

**[06:44]** out their 20, I'm left with a margin of

**[06:46]** 40% on a product I don't own. I don't

**[06:48]** have to warehouse and I don't have to

**[06:49]** worry about

**[06:52]** a return.

**[06:54]** >> Nope. No, there's because there's no

**[06:57]** nowhere to return it to. It's like,

**[06:58]** okay, cool. Ship it to Amazon.

**[07:02]** So, it that's handling all of it

**[07:04]** basically. Um, but again, just an

**[07:07]** example of what you can kind of do. Um,

**[07:10]** just playing with cloud code, playing

**[07:12]** with you know, different ideas and stuff

**[07:14]** because a lot of people are like, I need

**[07:15]** to build a software. I need to build

**[07:16]** this massive app that does all these

**[07:18]** things. I built half an app and maybe a

**[07:22]** Chrome extension and it's currently

**[07:24]** doing its own thing.

**[07:26]** >> Wow.

**[07:26]** >> But yeah, I'm not I'm not doing

**[07:30]** anything. Like I check it once a day to

**[07:31]** make sure it hasn't crashed. It just

**[07:33]** runs on a $2 BPS.

**[07:35]** >> How uh much time do you spend building

**[07:38]** it?

**[07:38]** >> About two weeks. Total actual hours is

**[07:41]** maybe 20.

**[07:43]** So,

**[07:45]** but for 20 hours worth of work over two

**[07:48]** weeks, I'll take it.

**[07:49]** >> Yeah. Just let it run.

**[07:51]** >> Yeah.

**[07:52]** >> So, but like I said, it's constantly

**[07:54]** learning off of its own mistakes, which

**[07:57]** is why the impression graph is the way

**[07:58]** it is. It's just continuously learning

**[08:01]** and testing and figuring out what works,

**[08:03]** what doesn't, what gets banned, what

**[08:05]** doesn't, and just reports all that, logs

**[08:07]** it to the central AI, and then uses

**[08:09]** those rules the next day going forward.

**[08:15]** So,

**[08:18]** just I want to put this in here because

**[08:20]** a lot of people talk about, you know,

**[08:21]** building stuff in in Vibe coding and

**[08:24]** doing whatever, but they never actually

**[08:25]** launch anything. We'll show what happens

**[08:27]** if you like actually do something with

**[08:29]** it.

**[08:31]** So,

**[08:33]** one of the other things I wanted to talk

**[08:34]** about was AI video generation. In

**[08:37]** specific, we're going to talk about

**[08:38]** building AI avatars. Kind of what um was

**[08:42]** talked about previously, but slightly

**[08:43]** different. Um, so I'll kind of show you

**[08:46]** guys what we're doing. So, this is the

**[08:49]** stack I normally use for all this. It's

**[08:50]** chat GBT, Google Whisk, Flow,

**[08:53]** occasionally uh VO free pick to upsample

**[08:57]** the images because the quality of the

**[08:59]** images coming out of AI are usually

**[09:01]** pretty bad. Um and then hey genen to

**[09:04]** turn that person into a permanent

**[09:05]** talking avatar.

**[09:08]** So it cost about a buck per video if you

**[09:10]** do it right doing this method because

**[09:13]** most of the stuff we're using with the

**[09:15]** exception of hen and free pick are free.

**[09:18]** So,

**[09:20]** examples of avatars you can make and

**[09:22]** then these get fully animated by the

**[09:24]** system. So, this guy's a influencer,

**[09:27]** digital marketing guy,

**[09:30]** regular like business or finance person

**[09:33]** and then Gen Z, Gen Alpha,

**[09:36]** inspirational, whatever you want to call

**[09:38]** it. Uh, but these were built off of

**[09:40]** literally five questions I have baked

**[09:42]** into a custom GPT and then as soon as

**[09:44]** you answer those, it will generate that

**[09:46]** person for you and then you can tweak

**[09:47]** it.

**[09:49]** So,

**[09:51]** going through the personas and all that

**[09:53]** sort of stuff. Um, like I said, I'll

**[09:56]** give you guys all the GPTs. There's only

**[09:57]** three and you just use them in the order

**[09:59]** I give them to you. Um, first one will

**[10:01]** generate the actual character sheet of

**[10:03]** who that person is for your business.

**[10:05]** Second one will generate that the uh

**[10:07]** head shots and a few others. And then

**[10:09]** you take that import that into Flux. Um

**[10:13]** and that will give you uh the ability to

**[10:15]** put them in different locations,

**[10:17]** different scenes. It'll lock the model.

**[10:19]** Um so it'll keep that person exactly as

**[10:21]** they are. And then you can push them

**[10:23]** into any scenario you want to make

**[10:25]** reals, videos,

**[10:27]** whatever you want. Kind of like what um

**[10:29]** was just being shown with uh Chad.

**[10:33]** But um we're doing this right now for a

**[10:35]** couple clients.

**[10:36]** We're doing about I want to say

**[10:40]** three videos a day across nine social

**[10:42]** media platforms. We're just using go

**[10:43]** high level to push them out. Um so we're

**[10:46]** getting about 27 reels out per day right

**[10:50]** now. Um and some of those accounts are

**[10:52]** in the 50 to 150,000 followers. Again,

**[10:55]** this is just in the past two weeks

**[10:58]** building these out,

**[11:01]** but it's and you guys will get the

**[11:03]** scripts obviously baked into the GBT,

**[11:05]** but also uh there's a whole framework

**[11:07]** that we use in order to get the reels to

**[11:09]** actually land with somebody. So, like

**[11:11]** grab their attention, pull them in and

**[11:12]** actually have them like go to do a CTA.

**[11:15]** Um you guys will get all that and you

**[11:17]** can just dump it in. It'll write your

**[11:18]** scripts for you.

**[11:21]** Um,

**[11:23]** and then what I wanted to show um was

**[11:26]** some video stuff that I've been playing

**[11:28]** with. So, taking ideas like these guys

**[11:32]** or different concepts and actually

**[11:34]** animating them or creating different

**[11:36]** types of like film stuff with it cuz I

**[11:39]** like messing around with video a whole

**[11:40]** bunch. So, I'll show you guys some of

**[11:43]** that

**[11:47]** if it closes. There we go.

**[11:56]** So, like as an example, this was done

**[12:00]** I want to say three, four months ago.

**[12:04]** Um, and this was a follow-up to the last

**[12:05]** video I had shown at Rockstar. So, if

**[12:07]** you were here for that one, this is a

**[12:09]** different version of kind of doing the

**[12:10]** same thing.

**[12:24]** But using the same GBTs I'm about to

**[12:27]** give you, if you guys wanted to make a

**[12:29]** video, you could.

**[13:14]** Yeah, that's what this is based off of.

**[13:19]** So,

**[13:50]** say again.

**[13:53]** >> Oh, no. So each one of these is an

**[13:55]** individual image that was then later

**[13:57]** animated. But all of the video you're

**[14:00]** seeing, the rendering, the compositing,

**[14:02]** the upscaling, and the background music

**[14:05]** was entirely done by AI.

**[14:11]** So that was a couple months ago before I

**[14:13]** started playing with making hyperreal

**[14:15]** people like this. So that's an example.

**[14:19]** Um, this one's a little bit more recent.

**[14:22]** Um, this one was done like right before

**[14:24]** the holiday.

**[14:35]** >> Whenever you're alone, I'll just hold

**[14:38]** you.

**[14:40]** I'll do anything you want. I'm not home.

**[14:46]** You're a little taste of heaven on my

**[14:49]** tongue.

**[14:51]** You

**[15:01]** can't get

**[15:34]** But again, completely done with AI. Um,

**[15:37]** everything from front to back on that

**[15:39]** animation, all of it. Um, using the same

**[15:42]** prompts and templates that I'm going to

**[15:43]** give you guys.

**[15:45]** >> Question for you.

**[15:45]** >> Yeah. on those. How would you modify

**[15:48]** them to I mean script not withstanding

**[15:50]** let's say that that first one with all

**[15:52]** that destruction.

**[15:53]** >> Yeah.

**[15:53]** >> Could you theoretically modify it for a

**[15:55]** local business like water damage

**[15:56]** restoration more destroyed than

**[15:59]** >> Oh yeah of course. Yeah.

**[16:00]** >> But

**[16:02]** because that would blow away everything

**[16:03]** else at the local level.

**[16:06]** >> Yeah. So being able to do something

**[16:08]** cinematic like that at the local level.

**[16:09]** Absolutely. Um you can do that with some

**[16:12]** of the um stuff that I have inside the

**[16:14]** GPTs. Um, so for example, if you plug in

**[16:17]** a script, it'll go ahead and write the

**[16:19]** entire thing. You would need to render

**[16:20]** the image that you're starting from

**[16:22]** first, you just need that one composite

**[16:24]** image and then follow the script that

**[16:26]** the system outputs and then just dump it

**[16:28]** in your favorite video maker of choice

**[16:30]** and you'll get probably 85 90% of the

**[16:33]** way there. Go back and maybe edit once

**[16:35]** or twice and you're good to go. Um,

**[16:37]** yeah, absolutely. Um, so

**[16:42]** >> um, so I'm taking those and running them

**[16:44]** through. Um, right now we've got a

**[16:46]** couple that we're using. I'm using V VIE

**[16:49]** or a key, sorry, KIE.AI.

**[16:52]** Um, so we're just pulling based on the

**[16:54]** type of shot, the uh, type of framing

**[16:56]** composite. So, it's multiple depending

**[16:58]** on what the shot style is. Um, like

**[17:00]** certain ones work better up close, other

**[17:02]** ones work better with human faces. So,

**[17:04]** depending if it's a landscape shot or

**[17:05]** close up, different renderings. Um,

**[17:08]** we'll pick different models for those

**[17:10]** different scenarios. I have the AI

**[17:12]** pretty much trained to be able to figure

**[17:13]** it out on the fly. So, it picks its own

**[17:15]** model and then does it.

**[17:18]** But yeah, so

**[17:21]** >> yeah,

**[17:21]** >> you mentioned last time.

**[17:25]** >> Yeah.

**[17:26]** >> Is that the same key?

**[17:27]** >> Yeah, key. Yep. Key.AI. Um, it's

**[17:31]** basically it's open router like you said

**[17:33]** for video and and images. So, if you

**[17:34]** don't want to have 50 different

**[17:36]** subscriptions, just use that. Um, and

**[17:39]** you can just plug everything in. Um, the

**[17:41]** system that I'm using obviously to make

**[17:43]** the videos is vibe coded. Um, a lot of

**[17:45]** it lives in Air Table, so it'll build

**[17:46]** itself through Air Table, save all the

**[17:48]** scenes, and then I can go through and

**[17:50]** like manually tweak if I need, and then

**[17:52]** have it bulk publish, generate the

**[17:54]** audio, compress, and then spit it out.

**[18:00]** So, that's that. And then if you guys

**[18:03]** want to go ahead and play with any of

**[18:04]** that stuff in any format, go for it. Um

**[18:08]** there's three GPTs at the bottom of it.

**[18:10]** And then there's a sevenstep framework

**[18:12]** that we use for all short form videos.

**[18:14]** So if you're creating any kind of short

**[18:15]** form video, um whether it's with one of

**[18:18]** the people that you generate like the

**[18:19]** examples I've given or you have a custom

**[18:23]** avatar that you've already done or

**[18:24]** you're figuring stuff out in Hen, this

**[18:26]** will actually help you to get it right

**[18:27]** from a marketing and sales perspective.

**[18:29]** So, aside from the shiny stuff, the

**[18:31]** stuff I'll give you, the stuff you're

**[18:32]** getting on that will actually work.

**[18:37]** >> Yo, Dan, you know you owe me money for

**[18:38]** letting all these people scan that QR

**[18:41]** code.

**[18:42]** Got all your data now.

**[18:45]** >> Everything.

**[18:46]** >> Yeah.

**[18:47]** Does anybody have any questions,

**[18:49]** comments, accusations of heresy?

**[18:53]** >> Heresy.

**[18:54]** >> I've got a comment. I just want to

**[18:56]** commend you on that first thing because

**[18:58]** I used to sell on Amazon

**[19:03]** um as a real form of income and I've

**[19:06]** been thinking about getting back to that

**[19:08]** and uh yes, I just wanted to commend you

**[19:10]** for doing it because

**[19:11]** >> Thank you.

**[19:12]** >> You did a great job.

**[19:13]** >> Thanks. I appreciate it. It was just

**[19:15]** like I said, I'm only using one API

**[19:17]** that's ebase. Um the rest is through

**[19:19]** that little Chrome extension. It goes

**[19:21]** opens a new tab, looks up the product,

**[19:23]** scrapes the data, figures it all out,

**[19:25]** and then it does computations in the

**[19:27]** back, figures out if it's good or not,

**[19:29]** if there's enough margin, if there's

**[19:30]** enough quantity left in the store and

**[19:32]** Amazon, all that sort of stuff.

**[19:34]** >> And it it just does all the viability

**[19:36]** shopping for me essentially.

**[19:38]** So

**[19:39]** >> yeah, I think I was halfway not hearing

**[19:41]** you when you first started, right? But

**[19:43]** did you show them the Amazon auto ship

**[19:45]** thing you got?

**[19:46]** >> Yeah. So guys, right, like please

**[19:48]** understand, even if you don't do it, put

**[19:51]** someone in your family to do it.

**[19:53]** Especially if they're not in this

**[19:55]** [ __ ] realm. We talk another language,

**[19:58]** bro. What he's saying is just go babysit

**[20:01]** this [ __ ] and make some income and that

**[20:03]** lead them into your world and and more

**[20:05]** importantly, right, an income. He

**[20:07]** started the session off showing you a

**[20:09]** semi-automated way to make side money.

**[20:13]** How amazing is that? Thank you, Daniel.

**[20:15]** I didn't know you was going to add that

**[20:16]** in there. I know you told me you were

**[20:18]** doing it. Yeah,

**[20:18]** >> cuz you just started it. Yeah.

**[20:20]** >> And you're already getting sales. That's

**[20:22]** crazy.

**[20:22]** >> Yeah. And again, most of it is like I'm

**[20:25]** like I said, I'm not babysitting it. I

**[20:27]** wake up and see if there's any sales I

**[20:28]** need to push out in the morning, let it

**[20:30]** run for the day, check at the end of the

**[20:32]** night.

**[20:32]** >> That's it. Like the amount of stuff. And

**[20:35]** then it's just like answering customer

**[20:37]** questions, messages, questions about

**[20:40]** products, spam, any of that stuff. So,

**[20:43]** did you just pixel all the stuff there?

**[20:45]** >> I did not. That is just a straight cut

**[20:47]** through to Google Docs. Nobody's been

**[20:49]** pixelled.

**[20:50]** >> I'm playing nice this time.

**[20:56]** >> Cool. Any questions? Anybody?

**[21:00]** >> Guys, give it up for Dan, please.