AI DTC WTF

Florian Litterst, German paid social agency owner since 2012, joins the show to break down exactly how his team has wired AI into every corner of their operation. The conversation gets specific fast, from a Claude-powered agent that scrapes competitor ads from Meta's EU ad library and surfaces untested angles every 30 days, to a another that mines Trustpilot, TrustShops, and Amazon competitor reviews for creative ammunition. Florian walks through his image generation workflow using krea.ai, why he now runs AI-ready photo shoots so every product can be "prompted into any situation," and how AI-generated B-roll is solving the impossible shot problem (think: socks on a runner's feet filmed from behind). The episode closes with a forward-looking debate on where AI is taking agencies from execution focused shops toward data-connected consultancies and a unanimous take that experienced human taste is the one thing AI still can't replace. Key Takeaways

  • What a real competitive intelligence AI agent looks like and how you can build one using Meta's ad library, Airtable, APFI, and Claude.
  • Why pulling competitor negative reviews from these places are one of the most underrated sources for ad angle discovery.
  • How the EU's ad transparency rule actually gives European agencies a data advantage over US-based media buyers.
  • Why investing in "AI-ready photo shoot" is cheaper than repeatedly shooting new product content everytime
  • When AI-generated video becomes a liability instead of an asset
  • How this one site is being used to create scalable image and video workflows.
  • Why creative iterations are losing their value in Meta's algorithm, and what agencies should be doing with their production time instead.
Products & Software Mentioned

To learn more about Florian and his team at Ads Venture go here: adsventure.de

To connect with Andrew Foxwell reach him here Andrew@FoxwellDigital.com To connect with Will Sartorious DM him here https://x.com/will_sartorius To Connect mith Thomas Moen DM him here https://x.com/thomasmoen To learn more about Foxwell Founders and conversations like this one, go here: www.foxwellfounders.com

What is AI DTC WTF?

Your Home For Tactical Tips, Strategies and Ideas for Your DTC - AI Workflows

If you work in D2C and you use AI and

you're wondering what the

F is going on every week,

this is your podcast, the AI D2C WTF

podcast, your home for

tactical tips, strategies, and

ideas that you can implement right now in

your AI workflows to make

your brand or agency more money.

All right, welcome to another episode of

AI D2C WTF. Glad to have

Florian Litter's German

legend on the show today to talk about

all the things that he is

doing with AI and in his business

and agency. Florian, how's it going, by

the way? It's good to have

you. This must be the most

complicated last name you've ever said,

right? The German, my last name. No,

sounds very complicated.

Absolutely not. Yesterday, I had Marin,

who is from Croatia, and he

is a very well-known Croatian

marketer, and his last name is Isch Dwan

Tick. I had obviously been

mispronouncing this for years,

and I probably just did again. So it was

tough, you guys.

Thanks for having me, guys.

Glad being here. Yeah, I'm Florian. I'll

just do this very quickly. I run an

agency here in Germany

doing paid social since 2012-ish. So

quite a long time. We know each other

since, I don't know,

2000, what? I don't know, 15, maybe? So

long. Long, long, long time ago. And

yeah, playing around

with lots of AI stuff in the agency to

find new ways to navigate

through this AI shift, basically,

for our clients and for ourselves. Yeah.

So Florian, like, what are

you, how are you using AI now?

What are the things that you're doing

that you've seen the most impact on? For

us, a big breakthrough

actually started last year. We did a team

workshop, and we discussed

about how can we make use of AI?

What's the best way? What's the best

strategic way of using AI? And

we stepped back a bit, and we

thought about how can we think of AI, of

using AI in different

roles, in different departments,

and different use cases. And that brought

us to, okay, we have creative. In our

agency, when you run

paid social as a brand or as an agency,

it doesn't matter. You do need creative

strategy. You do need

surely some kind of copywriting. You do

need design and video editing. That's the

biggest part you do.

And analytics and reports. And in all of

those roles, basically, we developed

workflows, tools. We're

using multiple tools for that, that help

us with strategy for

strategy. And quite lately, two weeks

ago, maybe, we implemented some AI agents

that helped us to find new ideas,

basically, for creative

strategy done in cloud. We have workflows

for image generation, and

we have workflows for video

generation. And when it comes to

creatives, I do believe the best case

right now is like a hybrid

case where you use real content and AI

content and mix this

together. That's basically a quick

rundown of what we do in each department.

Cool. So there's, I think

there's a lot of interesting

stuff there. But what I also heard a lot

of agency owners say is

that it's when they start

to see kind of real value out of it. And

it's, it's easier to grasp. When you guys

had this team meeting,

and you're kind of decided how to go into

this, what was the first kind of things

you started doing? And

what kind of tools did you

start playing around with?

I mean, that was back in June last year.

So in AI world, that's quite

a long time ago, I would say.

So we honestly, we started mostly using

ChatGPT, using deep research for creative

strategy, finding new

ideas, new angles, new hooks. But we knew

where in which direction we need to go,

because we know which

type of content performs on, on ads. So

that's, that's where I

started being like an architect for,

for prompt. So I was more like a prompt

engineer and engineered

different prompts to help us,

yeah, make use of ChatGPT to find new

angles on the strategy side.

And then we played around with

lots of image generation tools. I mean,

back then the first big breakthrough, I

think, was the image

generation of OpenAI, which was not so

good from today's perspective. Then

Google introduced Nano

Banana and from there took off basically

an image generation and what

we can do in image generation.

And with video, it's still kind of

okay-ish, I would say. I

honestly don't think there is

already a big breakthrough in video

generation. There are new models like Sea

Dance, which came out two

weeks ago, I think, which is quite good

and quite strong. But video is

like, it's hybrid. So we have

real content with, made with content

creators, and then we mix

and match this with AI content,

which helps us basically to produce stuff

that we couldn't produce with content

creators or that they

basically forgot to produce. And this

solves a big problem. Yeah.

So let's, let's get into specifics on

this. So of the things that you've built

thus far right now with

your agency, what are the things that are

the most high leverage

and, and specifically like

exactly how, how did you set those up?

You talked about agents. Is that maybe

where you want to start?

I would say, yeah, I would say. So how

is, how are you building them with Cloud?

Like how specifically

are you setting them up? Like with, you

know, what do they look like? What kind

of tools are you using?

I would say it's, it's the agents, um,

which are quite new because they are more

on autopilot right now.

Up until then, it was like, we need to

prompt something, we

need to wait, and we need to

guide it a bit more. And now it's way

more on autopilot. What we

did is we created a database

in the backend using Airtable. And in

Airtable, we have like different

informations about our

clients, about the brand, about the

products. And based on this...

And what kinds of stuff are you giving

them in Airtable? You're

giving it like the, the name,

the vertical, the website, you're giving

it like the persona. What

else are you putting in there?

Yeah. Persona. And it depends on the

brand, but sometimes you

need to go down on product

levels. Sometimes you can just say on the

brand level, depends on

the brand and each case.

And based on this, we've created, um,

yeah, two different

agents. One agent is, that's,

I would say that's unique to the EU

because in the EU, in Meta's ads library,

you have a transparency

guideline, which means in the EU, you can

see how much reach an

advertiser had on one ad or the

other ad. That's unique in the EU

compared to the US. And

based on this, my idea was, um,

why can't we create an agent that always,

um, analyzes the ad

library and checks for

ads with the highest reach from up to

five competitors,

analyzes those, and then pushes

back angle ideas that we might be

missing. So we always like monitoring

competitor brands from our

clients. Their ads library, we always see

which ads, uh, or we see

based on the transparency, uh,

rules and ads library, which of those ads

get the highest reach. And

I would say in most cases,

not in all cases, but in most cases, the

highest reach means the best

performance, not in all cases,

but in most, uh, and based on this, we,

this, this information, so

the ads, uh, scraped using

APFI, which is like a tool for APIs,

basically. Um, we're scraping

data out of the ads library.

We are also, um, uploading, um, the

images into air table, and we

are also pulling the videos.

The videos is then, the videos then

pushed to whisper from open

AI, which is transcribing the

video. And then we have everything that's

set basically in the

video in air table. And then

Claude is going for everything and is

ranking. Okay. This is a good

angle. This is a good angle.

That's my idea, uh, for you guys that you

create this or that angle

that you may be missing this or

that, or this is good by done by whatever

type of brand that's quite helpful.

So you're having, so the, the agent or

the agent goes through and you're

continuing to feed it all

this information. And other than the

tools that you mentioned,

you're using Claude as the

one that's plugging in air table, APFI

using whisper. Are there

other things, other tools in

there that are helping like pull it

together? And if so, what are those?

It's, it's pretty much that it's a

Claude, which is like the architect and

which is building my ideas.

Um, air table, um, APFI and whisper.

Yeah, that's it for that case. We also

have another case where we,

so this is like the, um, the ad library

agent and we, we have kind of the same

idea. We had kind of

the same idea with customer reviews and

we are not pulling customer

reviews from our clients. We are

pulling the reviews from their

competitors, which means we always see

the top three positive and top three

negative comments from the competitors,

from our clients, which is

kind of the same idea. We are,

we always get ideas for like, okay, these

are angles you're missing.

These are angles that are

maybe like a anti pitch angle, whatever,

because they are lacking

in quality here or there.

You could go into that direction as an

angle with, with, uh, for

your client and, uh, basically,

yeah, uh, get this information out of

Trustpilot, TrustShops and

Amazon. Also done with APFI

and Claude and air table in the

background. Just a follow up. Um, um, a

lot of people I talked to

today use Superbase versus, uh, air

table. Um, I know they both do the same

things. Uh, but what I'm

interested in trying to understand is,

um, um, what are you doing then with all

the learnings and the data

clock gives you based on what it's

analyzing? How, how do

you, how do you use that?

So right now it's right now it's

basically pushing, uh, the idea. So it's

pulling this data every 30

days for each of our clients, not daily

every 30 days, because you

need new data that needs to be

analyzed. And that's why we decided to go

with every 30 days. And then

it's pushing like a message

into the Slack channel. Basically we have

a Slack channel for each client, uh, in

our agency and there,

our agent is pushing a message with, I

analyzed whatever 50 ads

here, the top ads. I think this

is good. This is good. So I asked Claude

to score those ads or to score the

reviews. Uh, and based

on the scoring, it's basically sending us

the ideas. And then we have

a set of ideas and a set of

ads that we might be, uh, producing. Are

you then kind of looping back what is

created in terms of how

it delivers? I mean, is it like a

feedback loop of, was it actually a good

idea? This did this ad

perform or is that maybe kind of maybe

the next level of what

you guys are thinking on?

This is not done by AI yet. Um, this is

done by our creative

strategist. Um, and I think a creative

strategist, a good creative strategist is

always looking what worked

in the past and is always

looking into like what might work in the

future. And this is

basically just at the end, it's just

in another angle that we're testing, but

the idea is coming from AI

and we save lots of time, uh,

in research here. Yeah. So you have, so

you have the, this agent set up. So the

other stuff you have

in terms of, uh, image generation and

video generation, talk

through those and how those are

set up and, um, yeah. And what kinds of

things they're, they're

doing and what they're built on.

Well, we played so much with different

models and different tools

out there. And we came to the

conclusion that, okay, the most important

thing is having good raw

material, a good kind of

imagery from each product to basically

train a model or to basically use an

image as, as a reference

image, for example. That's why we

started, um, a month ago, uh,

doing photo shootings, AI ready

photo shootings in a studio, which

basically means we get the product and

we, um, create images of a

product in each angle with perfect light.

Um, so we can use this image, this

product image to train

models or to use it as a reference image.

That's where we are right

now. Um, and then the tool we're

using or the platform we're using is

called, uh, kreia.ai. That's what we use,

uh, kreia.ai with K,

not C, not C, kreia.ai. AI. And in, in

that, uh, that's like Higgs field, for

example, it's pretty

similar to Higgs field. Basically it's a,

it's a wrapper. It's a kind

of a wrapper, which, uh, um,

is plugged into, uh, every model through

the API. And in CREA, we can

basically create images with

nano banana, nano banana 2, uh, whatever,

every, every version of nano banana,

every version of, uh,

the latest video models like C dance, for

example. And then in CREA, we built,

or we're building workflows in a way that

we say, okay, we do know

which type of imagery works as a

performance ad. So for example, product

enhance is classic one. Uh,

each product shot in a hand from

like a eco perspective is usually a top

performing image ad. Um, and

we created a workflow just to

create multiple images of a product in

different hands, basically

different settings, different

hands. So that's a workflow for us where

we basically upload the

imagery from different

products. It's always the same workflow.

And then it creates

different images that are from our,

that we know perform as an ad. So it's

like the workflow is built on,

built, uh, built in a way that we created

each, uh, different

workflows for each use case where we

know, okay, this type of imagery works.

Um, and, uh, yeah, that's

for image and for video. It's

basically my experience with video is

like so-and-so sometimes it's

quite good. Sometimes it's, um,

slop, AI slop. And, uh, for our

customers, um, I'm trying to not create

AI slop. Um, that's why we started,

uh, in a way that we try to, um, try to

widen in a way, our B-roll library for

video content in a way

that we create content that we haven't

created before with videos.

And this is done kind of like

an art director, I would, uh, art

director, I would say. So we create

different, um, different

images based on the images, uh, the

photos we've created, and then those

images, we animate those.

And then based on this, it's like a

sketchbook, I don't know. Right. So we

have different images,

different images getting animated, and

then we have five different B-rolls and

these five different

B-rolls are then fed into the video team

and mix and matched

with our existing content.

That's, uh, what we do there. So, and so

basically a client signs

with you and they come and say,

these are our products and you're, you

look at the top products and then you

make sure to load those

into the like static workflows that you

have, which is a product holding in hand.

And they're the things

that you have seen that have worked

before. And then with the video,

basically taking like

animated gifs essentially. And then

you're building those into the

preexisting video assets that you

have is like the hook. Is that

essentially kind of what it is? Exactly.

Exactly. Yeah. Exactly. Yeah.

And, and you're going through, so when

you're building these out and

you're, you know, starting to

build them new creatives and new creative

direction and stuff like is, is there,

there's the research

phase that's taking place that we talked

about before with the

agent and that's then spitting

out as well, some ideas for just the

creative strategist or does

the client also see those or

like what's, you know, cause the

innovation on not finding new personas

and stuff, I assume maybe you're

telling the client as well, but so maybe

some of that's

internal. I'm just curious.

So what I'm trying to build is, uh, I

would say for most processes,

it's still human in the loop,

which means we are still screening the

ideas and we still say, okay, that's a

good idea. That's a bad

idea. Even with designs or with images

that are created based out of those

workflows, we always

need to make sure that there's a good,

like a mood board, um, at

the beginning, which is like

style trading for the AI, I would say

done by human. And we always

make sure that the output is

basically, um, uh, controlled by human,

um, to really make sure

that this is on brand and this

is not sloppy in a way. Be really clear.

Sorry. One more question.

So you're, you're taking the,

let's say that you did that with video.

So you take some of this

stuff that the agent has learned,

the mood board, you then take it to this

software that you mentioned, or the one,

I forgot what the name of

that one is. That is kreia.ai.

KREA. And then you say to, to kreia, Hey,

this is all of the stuff that we know.

Here's a, here's a preexisting video

asset that we want to be added

or want to be like included in

this. Create the, a 30 second version of

this for me. Or like, what are you

telling kreia to do?

You, you're speaking about videos, right?

Yeah. Yeah. Yeah. Okay.

It's like kreia is basically,

it's no difference in using Google V or

free or for whatever, or

kreia. It's, it's basically the

same model. Um, what we do is we, we, we

create for videos, we create like a

sketch board. So it's like

having someone who's thinking of the

video, we, which scenes do we need in a

video? And we create

images for each scene. This is based on

the idea. And then

based on this, we say, okay,

we think C-Dance 2 is the best model to

animate this scene. And then

we have 10 different scenes.

And those 10 different scenes is

additional B-roll footage for our video

team. That's how we,

how we think about this. And the, the,

the 10 different scenes are being

sketched out by what

you're learning from the agent. In some

ways, yes. In some way it's

still made by, by us, by our

creative strategies. It's by you're just

saying, okay, actually this is what we

know. This is what the

client says. And then like, here's the

scenes and there, and

these little scenes.

I do have one good example, which is

quite hard to, how to shoot as a video

actually. So usually we,

we, we still do lots of videos with

content creators out

there. We have roughly 300 that

are working with us. Um, and we have a

client that's selling socks, sports

socks. The issue with

sports socks is it's really hard to, as a

content creator, it's

really hard to shoot a video,

wearing your socks, running and shooting

a video from, from the back, uh,

basically recording yourself

running. That's very, very hard. So you

definitely need a video team

or someone else, uh, recording

that person, which is expensive, which,

um, makes costs going up. And what you

can do is you can create

this scenes or multiple scenes where a

person is wearing socks and

it's running basically. And the,

video is, uh, recording this person from

the back, from the side, whatever. This

can be perfectly done

by AI. And then you can basically mix and

match the scene of whatever the content

creator who's talking

about the product with this B-roll

footage. So that's quite a,

quite a good example, I think.

But, but, but, but then again, which I

think is really interesting is that

you're using that so your

experienced team can, can make the

decision if it's good enough or good

enough and give feedback.

So it's, it's, it's just like you're

working with a AI photographer, uh, and

the rest of the workflow is

pretty much the same that it's always

been. Is that kind of correct or?

Kind of. I do think that human taste is

still super important. Um, but everything

else should be enabled and

enhanced by, by AI. And, uh, that that's

how we think about this.

Yeah. I also built another quite

fun workflow, something else I did. I did

build quite a fun

workflow, which is pulling the best

performing image ad out of the ad account

and creating iterations

based of this best performing

image ad, which is quite easy for AI. So

it's basically downloading the best

performing image ad,

uploading the best performing image ad

into Google drive done in cloud. Uh, then

it's sending the image

into cloud cloud is a reading basically

what's on the image. It's

creating new hooks basically. So

text overlays for the image. This is then

again, double checked by

another, uh, um, prompt basically,

because sometimes the grammar was wrong

or the spelling was wrong. And then

basically it's creating

three different versions of the same

image with different hooks

and overlays. This is end to end

automation of, of creatives. But from all

what we guys probably know is that

iterations are not like

the thing where you should spend too much

time on because of meta

andromeda and that are there

creative diversification. So I would not

spend too much time on this.

That's why I thought, okay,

it might be a good idea to automate this

end to end. That's, that's what you do

with, with image ads

and iterations. Yeah. It's pretty cool.

And, and, um, are, are the, are these

versions, um, performing

well when you're testing them? Like the,

the flow is working, but it is giving

good quality as well.

It's, it's working. We still double check

every output because I

mean, I would not upload random

stuff into ads manager, at least not when

I'm the one who's responsible, uh, for

our clients. So I always

want to have someone double checking

this. Um, and yeah, their

output is good, but, um, from

everything we see before this cloud

transition and AI shift,

iterations are not the thing that

are driving performance in ads manager

that much, at least sometimes yes,

sometimes no. That's why we,

before we had this workflow, we decided,

okay, we don't put, put much

effort into iterations just

because of that shift on meta side. And

we spent more time finding

new concepts. And that's why I

thought it might be a good idea to have

to still like autopilot in

the background who's creating

animations, but we don't spend human time

on this. Yeah. Yeah. And how are the

clients reacting to,

to this kind of shift? Do they, do they

know, do they care about it

being more AI in the picture?

How, how are, how are they feeling? One

year ago, I would have been afraid,

honestly. Um, and I thought,

um, that, um, that maybe clients don't,

um, at least for, I'm

speaking now for the German speaking

market, which might be different to the

international market, just to be, just to

have this disclaimer

here. And looking at the German market, I

thought that, uh, most people

think that we basically spend

less time managing their things, which is

not true at the end. And, uh,

then our, the perceived value

of what we do as an agency is going down.

That was my, uh, very, to

be honest, but turns out it's

completely different. People really love

when we do this. People

are really open. Uh, people,

all clients always, uh, yeah, they, they,

they're super happy when we

create new stuff. For example,

another thing we do, and all of our

clients are, are really loving, loving

this is we, we create CGI

type of content, which is basically you

take the product and you

put it somewhere in front of,

I don't know, the Empire State Building,

for example, that's

trademarked, would not do this,

but it's possible technically with AI. So

you take your product and you put it

somewhere on a monument,

someplace that most people know. And this

is the CGI type of stuff,

which is quite cool. It's still on

brand and people love this when we do

this. That's one example

where we understood that our

clients really like AI stuff. Before I

was hesitant and I was skeptical, but

right now I'm all in. I'm

going all in into this and I'm pushing

this really hard, um, to the team and the

team is really open.

Um, our clients are super open. We do

lots of content around this and we are

very, very open to our

clients that we use AI in different

scenarios. They trust us that we are not

creating AI slop. Uh, and

I take care that we are not doing this.

Um, and that's why I

think it works quite well.

So if it's done right, AI basically makes

the work that you're doing

more effective because you're

able to give the client more creative

options and therefore more scale, right?

Because scaling is so

much dependent upon persona development

and the expansion of

matching personas into that. Um,

what do you think in terms of places

you're spending your time

and what you're seeing and

hearing from other leaders? I think

there's, there's a lot of people throwing

shit against the wall right

now, but what is your opinion about the

future of this moving forward and where

you are spending time and

where you'd recommend others spending

time? Because I think a lot of us

initially with this have gone

into better agency owners have gone into

spending time on reporting

and automating the reporting and

spending time on creative iterations and

obviously building up more iterations and

workflow and processes

around that. The next phase of this to my

opinion, I will, I won't

poison the well. So I have an idea

of like what I think is going to happen

with this, but I'm curious of your

opinion of where it's,

where it's headed and where people should

be spending their time. Um,

you, uh, you mean spending

time as an agency or in general in, in

this digital market? As a

marketer, let's just say,

as an agency, you could, you could

certainly answer it any way

you'd like. Um, I do think,

I mean, there's a, there's a saying, and

this is quite common that the person

using AI is definitely

outperforming the person not using AI. So

there's no way not using AI

definitely. And I would be

curious. I would be open. Um, um, and I

would, I mean, at the

end I, I would spend time

thinking about what's actually a good

system and how can I

connect together multiple,

multiple things, databases, tools,

whatever, to find a good and effective

workflow. So being more like

an architect being more like, uh, I mean,

pretty clear, you need to

move from, uh, the day to day

work more to a strategic work. I mean,

that's nothing new that

happened already with media buying

five, six, seven years ago with iOS and

every, every change there.

Um, this is, this is, um,

accelerated by AI and you need to, you

need to be the one who's really

connecting multiple dots and

maybe multiple channels insights from

multiple channels. For us,

this means we, we are, we evolve

more into like a consultancy, which is

also doing agency stuff

basically for our clients. So we help

with multi-touch attribution,

cross-channel, branding, upper funnel, I

don't know what else,

which is usually that's not what we do,

but we know this stuff and we can help

with this stuff and we

really, I think with AI as an agency, you

have the option to really cut

down costs for your clients.

Um, and that's what we try to do with the

photo shoots we do. This is

a photo shoot you do once,

and then you have perfect imagery for

endless AI creations of your, of your

product. Um, and that's

how we try to help being more agile in

terms of content creation. Um, which

basically means you can

prompt your, your product into every,

every situation. And before you are, you

would need to create a

photo shooting. Um, that, that, that's

how I think about this in general, I

would be interested in your

take, uh, because I mean, the end goal

of, I would say every ad platform and

also meta clearly is to automate

ad creatives and to basically give us

your money. We do everything for you.

What do you guys think? Do you

think that's, that's happening and when

would it happen? What do you think?

What's your take on this?

You started the loop. So you just, just,

uh, what's, what's the next phase and

also specifically around

these platforms and automating

everything? Yeah. So I think there's a

couple of different pieces of

this. One is as it relates to the way

you'll use it as a marketer right now,

it's been used in silos.

So we've been utilizing it in creative

because that's where so

much of us spend our time in

terms of scaling an account. I think that

it'll continue to expand

into utilizing it for full

funnel. So one of the most powerful

things that we can do is build a better

offer as build better

landing pages that match personas. And

then the ads match those

personas too. So that it's a full

thing where if you have a particularly,

and then you, and then you cross

reference this with financials.

So you have a product that if you're

selling water bottles that

has a better margin than a

product that if you're selling

microphones and looking at things on my

desk, um, then you're

going to push a, the water bottle more

and AI will say, Hey, look,

your margins are better on this

product and you have more of this in

stock and your new customer

acquisition on this particular

product or this particular funnel is X

over the last 50, 60 days. I recommend

that you start doing

more creative for this and it will then

spool up persona development that's

around the water bottle

angles that you have in the markets. And

then it'll have corresponding landing

pages or corresponding

advertorials that are built and you'll be

able to say, yes, try this

idea. Um, and then it basically

builds from there looking at, you know,

competitive analysis and continues to

like go on the things that

you've been talking about. So I think

it'll become a more full funnel play. And

I think that that also

extends to agencies. I think that where

it comes with agencies is

really going to be, how are you

making the case for AI with your clients

in agencies to say, look, this

isn't just me messing around,

right? And Florian, you talked about AI

slop. This isn't just us

trying things. And I think to some

degree in agencies, this is actually

people messing around. But I think right

now, what you want to try

to go forth with is look, we're going to,

this is allowing us to go into deep

research with more ideas,

and it's allowing us to build our process

better so that we get to

spend more time effectively

analyzing data for you and doing and

being a better advertiser. It's not just

advertising all products

and saying you have a four row as it's

like looking at them by category, looking

at them by contribution

margin, looking, you know, looking at

sort of the results in relation to the

financial metrics and

calculations. And if you're able to do

that, utilizing AI as an agency moving

forward, you become a true

marketing partner. You're not just one

person that's building the

meta ads, right? You're doing

a lot of other things. This could even go

into things like building

incremental channels, right?

Where, look, we have this idea, we're

going to launch connected

TV now, and this is how we're

going to be able to do that. So, you

know, I think this is, it's

headed in the direction of

making it more effective with more people

and, or with less people

involved, I think. But with the

people that you have, it's going to be

more data that's going to

really actually be able to really

help folks. So that's kind of what I

think. I don't know.

Thomas, what do you think?

I agree. And I think that, which I also

feel like Florian's point

in this whole conversation is

around that you're still going to need

real people with real experience,

understanding the data,

understanding the workflows and

communicating it clearly to the clients.

So I think that people

will be more and more valuable, and at

least people who have experience in this

industry from before.

And I also think that it's going to be

harder and harder for younger

people to kind of get up to a

certain level and get that gut feeling

experience because, you know, younger

people are just agreeing

to whatever AI is telling them more than

us grumpy old farts who are

just like challenging and no,

not that way. And you think like this

and, you know, all that stuff. So I

definitely think it's going to be

more of the whole funnel. And I think

that Meta, Google, everyone

is going to try to build that

platform. But I think that the more the

agencies can take ownership

to creating their own software

and their own platform and plug into the

different APIs, I think

that's the one who can do the best

job for their clients going forward. It's

quite funny because I was

playing around with Cloud,

which is basically, it's pretty easy to

duplicate product detail pages using

Cloud and then change

something and messaging on a product

detail page or any landing page. So it's

pretty easy to spit that out

because I had the same idea in like,

okay, we have the messaging on the

creative, but it's not

really, really matching the landing page.

And the idea was to

basically create a system that's

that's using the existing page and

changing the headline,

changing the image to,

to have a better match with, with ad

creatives, which could be

done quite easily using Cloud,

honestly, and Netlify. But

yeah, that's a different story.

Yeah, I mean, so much of it is just like

trying and testing and

I'm thinking about it,

but I think you have to make sure that

you're trying and testing

things in a strategic framework

that add into the greater whole. So it's

not just like, oh, it's like,

you know, so much of this is

like, you see this on X, oh shit, I got

to try that. You see this on X, I got to

try this, I got to try

this. And it's like, you can try it if

it's, but you also have to have a

destination because otherwise,

you're just taking all these side

journeys that aren't necessarily

improving the overall things.

It's like, where are the hard costs of

our agency? Where are

the hard times that were,

or where are the hard costs of working

with a particular client? You know,

is it client communications? What are the

things that we can do to

make that more efficient and

effective and saving money there? So I

think it's like, if that

contributes to that goal,

then absolutely. Right. But yeah,

Florian, I appreciate you

joining us. Florian will

be able to answer any questions. If you

have them, you can email me

andrew at foxwelldigital.com.

And all of his details and notes, if you

want to follow Florian,

are going to be in the show notes. But

thanks. Thank you for being here.

Thanks for having me guys.

Bye.

Healthy options.

ある程度 最初 使用