Speaker:

Well, hello and welcome to the eCommerce Podcast.

Speaker:

My name is Matt Edmundson and it is great to be with you today.

Speaker:

Of course, this is your first time with us, a very, very warm welcome to you.

Speaker:

Make sure you like and subscribe and do all of that good stuff if you think

Speaker:

there's going to be some value to staying connected with us, which of course

Speaker:

I truly believe there will be.

Speaker:

And of course, if you're a regular, welcome back because

Speaker:

today I'm going to be talking about

Speaker:

how I use AI in my e-commerce businesses right now.

Speaker:

This is a big question I get asked a lot.

Speaker:

It's a big topic and we're going to get into it.

Speaker:

A stat I came across recently, which really surprised me is

Speaker:

56% of CEOs report a zero ROI from their AI investments.

Speaker:

In other words, there is zero ROI on their ai.

Speaker:

Oh, I love it.

Speaker:

Anyway, just little jokes that keep me amused really.

Speaker:

But I don't think we're, they're experiencing that zero ROI on their ai.

Speaker:

Not because AI isn't working, but I would venture to say it's probably

Speaker:

because we're not using it in a way that creates that ROI on our ai.

Speaker:

Anyway, enough for that.

Speaker:

So I, my experience here is most of us, like the idea of ai, some

Speaker:

of us are using it, the majority of us that are using it are dabbling with

Speaker:

it, sort of playing around the edges, not quite sure how to dig into

Speaker:

it and how to get the most out of it.

Speaker:

And because it's moving so fast, how do you keep up with the whole thing, right?

Speaker:

I thought we'd talk about it in this episode and hopefully move you a little

Speaker:

bit closer to getting an ROI from your ai.

Speaker:

Okay, that's honestly the last time.

Speaker:

So let me give you a quick overview of the tools that I'm currently using,

Speaker:

and then I'll do a little bit of a deep dive into each tool and why I

Speaker:

think each tool deserves its place in my tech stack in my e-com business.

Speaker:

Okay.

Speaker:

Now, across my whole company, organization, we currently use amongst

Speaker:

the whole team, including me, four tools.

Speaker:

Okay.

Speaker:

We've tried more.

Speaker:

And there are definitely lots and lots out there at the moment, but they

Speaker:

tend to be the four that we stick with.

Speaker:

Now, the main one that we use is Claude.

Speaker:

We probably pay more to Claude than any other,

Speaker:

software platform, with the exception of Klaviyo,

Speaker:

but with Claude, I personally have the Max plan subscription.

Speaker:

I don't know its specific name.

Speaker:

It's the Max Max plan, as high as I can go, because I do hammer it.

Speaker:

And it costs about 150 pounds a month plus VAT £180.

Speaker:

It's about $250, I guess.

Speaker:

$220, $250. You definitely don't need to spend that much money, by the way.

Speaker:

It's just, I hammer it and for the longest time I was on their

Speaker:

£90 a month plan, which was great.

Speaker:

But I just needed a little bit more with some of the stuff that we're doing.

Speaker:

The second one.

Speaker:

So we use Claude.

Speaker:

The second one is called Perplexity.

Speaker:

And again, I'm going to go into each one of these in detail.

Speaker:

But just to give you some idea of cost.

Speaker:

Currently, Perplexity costs about $20 a month.

Speaker:

For the sort of the platform that we're on.

Speaker:

I also

Speaker:

love Google NotebookLM, which is my third system, it's part of

Speaker:

their Google Gemini AI system.

Speaker:

I think I pay 20 bucks a month for.

Speaker:

For that.

Speaker:

I think I pay 20 bucks a month for, I really should check.

Speaker:

I think we just pay 20 bucks a month for the whole Google suite of AI

Speaker:

tools, of which NotebookLM is one.

Speaker:

And the fourth tool we use is also part of Google Suite.

Speaker:

You may have heard the name Nano Banana.

Speaker:

Or I suppose to my cousins across the pond, my nano banana.

Speaker:

Which actually stands better when you pronounce it that way.

Speaker:

But anyway, it's not actually a separate tool.

Speaker:

It is part of Google Gemini's sort of image generation model that went viral.

Speaker:

And I guess Nano Banana is a better name than image generation tool,

Speaker:

so, but that's part of Google Gemini as well.

Speaker:

Now, if I was on Shopify, which currently.

Speaker:

I am not specifically on Shopify with any of our e-com businesses.

Speaker:

I've talked about that before on the show.

Speaker:

There is a company, there is a possibility of one of our websites,

Speaker:

or a website that we're getting involved with potentially being on Shopify.

Speaker:

The reason I'm saying that and being slightly cagey, is if you are on

Speaker:

Shopify, if I was on Shopify there, there might be a fifth platform

Speaker:

that I would add to this mix.

Speaker:

Because ai is now available on Shopify.

Speaker:

It's called Shopify Magic, and Shopify Sidekick, and the

Speaker:

guys in the e-commerce Cohort groups have used it.

Speaker:

I personally haven't, like I say, our businesses aren't on Shopify,

Speaker:

but they have, and the, they.

Speaker:

Actually really, really like it.

Speaker:

They're loving it right now.

Speaker:

So if you are on Shopify, go and try Sidekick.

Speaker:

Maybe that would be the fifth one that I would add.

Speaker:

I think it's available on all plans, but it's definitely worth exploring and

Speaker:

like I said, it sort of comes,

Speaker:

recommended to me by the chaps in Cohort.

Speaker:

If you don't know what Cohort is, if you're new to the show, Cohort is a monthly,

Speaker:

group that you can connect with.

Speaker:

And every month we jump on Zoom, in the various groups.

Speaker:

And we just talk about eCommerce.

Speaker:

They're free to join.

Speaker:

You get to meet some sort of fellow eCommerce traveling,

Speaker:

the world of journey of eCommerce.

Speaker:

Some of them have big businesses, some of them are just starting

Speaker:

out, but it's really great.

Speaker:

We get that mix.

Speaker:

And the calls last for about I guess an hour or two.

Speaker:

I love them.

Speaker:

It's one of my favorite things we do every month actually, is the different

Speaker:

Cohort groups and just chatting with people about e-commerce, learning

Speaker:

from them and throwing ideas in.

Speaker:

And we're all.

Speaker:

Throwing ideas in and learning together and it's really, really good.

Speaker:

If you want to know more about that, go check out eCommerce

Speaker:

Podcast dot net slash Cohort.

Speaker:

I actually think you can go to, I'm just going to check it now.

Speaker:

This is really unprofessional.

Speaker:

I should've.

Speaker:

I should have done this beforehand, but I think if you go to eCommerce,

Speaker:

Cohort dot com takes you to that.

Speaker:

Yes, we have the domain and it's working.

Speaker:

So just go to eCommerce Cohort dot com.

Speaker:

If you can't remember any of the domain, it'll take you to the

Speaker:

exact page you need to go to anyway.

Speaker:

All of that said, the Shopify AI might be the fifth

Speaker:

one you want to add to your list.

Speaker:

So I think across the whole company, we probably spend about 3, 3 50, something

Speaker:

like that on ai, bearing in mind, I'm taking up the vast majority of that with

Speaker:

my Claude subscription and our dev team have some pretty.

Speaker:

They have the big subscription as well.

Speaker:

So I think that's about what, 4 50, 500 bucks a month?

Speaker:

US dollars, something.

Speaker:

Along those lines.

Speaker:

And what we did was we were actually spending quite a bit more.

Speaker:

And so what we have done is culled,

Speaker:

if that makes sense.

Speaker:

We sort of stripped back the.

Speaker:

AI tools that didn't make sense for us.

Speaker:

And so we definitely have stopped chasing the shiny objects, you know,

Speaker:

which I'm really bad for actually is just that.

Speaker:

I just come across 'em and go, oh, that sounds a good idea.

Speaker:

We could do that.

Speaker:

And we would try it and then we'd forget about it.

Speaker:

You know, some of the image generation stuff and things like that.

Speaker:

We've tried.

Speaker:

So we finally cut everything down.

Speaker:

We've simplified it and this tech stack of four AI tools work like

Speaker:

insanely well for us and our company.

Speaker:

Okay, now.

Speaker:

Full disclosure, this may well change in the future, starting like maybe tomorrow.

Speaker:

I don't know of anything that will be different tomorrow, to what we're

Speaker:

doing now, but it's ai, it's moving quick.

Speaker:

You don't know.

Speaker:

For example, one of the things that we did in this culling was

Speaker:

we dropped ChatGPT, the whole open AI thing.

Speaker:

We don't subscribe to that anymore other than the fact that we do have a few.

Speaker:

Again, just being totally honest, we have a few apps that use open

Speaker:

AI's Whisper, which transcribes things, and it does it super, super well.

Speaker:

But we dropped chat.

Speaker:

GPT is still a great tool, genuinely a great piece of ai,

Speaker:

but I just prefer Claude.

Speaker:

It's what I do.

Speaker:

There's a bunch of reasons, why I do.

Speaker:

So let's dig in.

Speaker:

Let's dig into each one of these tools.

Speaker:

And specifically how I am using them.

Speaker:

And I'm going to spend a bit of time explaining to you how I'm using Claude.

Speaker:

You might be like, Matt, surely it's easy.

Speaker:

You open up the app type stuff in and away you go, haha.

Speaker:

No, it's not.

Speaker:

Like, like that.

Speaker:

For those of you don't know, Claude is an AI assistant made

Speaker:

by a company called Anthropic.

Speaker:

Like ChatGPT, you can have conversations with it, you can ask it questions,

Speaker:

you can get it to write things, you can analyze documents, you can, it

Speaker:

can help you with all kinds of tasks.

Speaker:

And it, and like sort of basic ai, a lot of it is down to prompting.

Speaker:

And prompt engineering.

Speaker:

It does come in different versions, and you can use it through the,

Speaker:

a web interface, like in a browser.

Speaker:

You can use it on an app, like a desktop app or on your phone.

Speaker:

And it's also got this built in coding tool, which is actually what I use.

Speaker:

It's how I use it.

Speaker:

And that's called Claude Code.

Speaker:

Okay.

Speaker:

Now, Claude Code is basically, it is Claude.

Speaker:

It's the same system.

Speaker:

But it's, it's having Claude live inside your computer's terminal.

Speaker:

Which without getting too technical, if you're not sure, is the black

Speaker:

screen of texts that developers use.

Speaker:

You'll see that quite often.

Speaker:

So instead of chatting into a browser, what this means is Claude

Speaker:

can run directly on your computer and it can see your files on your computer,

Speaker:

it can run commands on your computer.

Speaker:

It can make changes to computer code.

Speaker:

And originally this was designed for coders and to do coding.

Speaker:

Right?

Speaker:

And it was like having a developer sit next to you who can actually

Speaker:

touch the keyboard in, who can code.

Speaker:

So why does this matter?

Speaker:

If you are not a coder and you're just, you know, something like you're using

Speaker:

Shopify, you're not writing code, you know where you've got, you've got a dev,

Speaker:

a dev team, you know, whatever, right?

Speaker:

You're not, you're not doing the code.

Speaker:

Why would you want to use Claude Code?

Speaker:

Why would I want to use Claude Code?

Speaker:

So imagine asking, say, ChatGPT to fix something in a

Speaker:

document or basic clause, right?

Speaker:

And it's going to reply with the comment, okay, well here's what you need to change.

Speaker:

And then you would copy it and you would paste it into your document,

Speaker:

or you would download the document from ChatGPT and you would save

Speaker:

that somewhere on your computer and you do whatever you need to do.

Speaker:

Right?

Speaker:

That is the normal sort of Claude / ChatGPT behavior.

Speaker:

Imagine if you could just have AI make that change for you.

Speaker:

So it changes the file, right, right there.

Speaker:

Right.

Speaker:

And it tests it, it tests it once it's changed it to make sure it works and

Speaker:

then it can move on to the next task.

Speaker:

Right.

Speaker:

So that's, that's in essence Claude Code.

Speaker:

It doesn't just advise, it actually does.

Speaker:

In relation to the files on your computer, and the key

Speaker:

thing this then means is it has.

Speaker:

Context.

Speaker:

Okay.

Speaker:

So it, Claude Code then can see your whole project.

Speaker:

So it could see your whole coding project, it could see,

Speaker:

what's going on in there.

Speaker:

It could see all the files, it could read them, and it could understand

Speaker:

how all the different files and pieces connect and it can make

Speaker:

changes that fit with everything else.

Speaker:

So rather than giving it isolated snippets of information,

Speaker:

you know that sort of give it part context or you have to figure

Speaker:

out how to plug that all in.

Speaker:

It's way more than that.

Speaker:

So think of it like the difference between texting a plumber for

Speaker:

advice versus having the plumber in your house with their tools.

Speaker:

Okay.

Speaker:

That's a really good sort of analogy of Claude Code's.

Speaker:

Okay.

Speaker:

It's the plumber in your house with your tools now.

Speaker:

Stay with me here because we are going to supercharge this a little bit.

Speaker:

Okay?

Speaker:

Now what I do then, personally with Claude Code, bearing in mind it can

Speaker:

interact directly on my computer with files and make changes to those files.

Speaker:

Okay?

Speaker:

I combine Claude Code with a tool called Obsidian.

Speaker:

Now if you haven't heard of obsidian, it's basically a note

Speaker:

taking app, which I appreciate.

Speaker:

Does not sound exciting in any way.

Speaker:

Why am I telling you about a note taking app?

Speaker:

Get ready.

Speaker:

Okay.

Speaker:

Most of us use note apps like, I've used Apple Note, I've

Speaker:

used Evernote, I've used Craft.

Speaker:

Never really played around with Notion.

Speaker:

I know that's a big deal for a lot of people.

Speaker:

Google Docs, you know, I've played around with all of them.

Speaker:

So why obsidian?

Speaker:

Why is this different?

Speaker:

Why, why do I use it?

Speaker:

And more importantly, what's this got to do with Claude Code?

Speaker:

Well, remember,

Speaker:

Claude Code interacts with files directly on your computer, and obsidian

Speaker:

might be another note taking app.

Speaker:

Okay?

Speaker:

But it's got one crucial difference.

Speaker:

To the other note taking apps, your and what it does is when it creates

Speaker:

the notes, okay, they in effect become very simple text files

Speaker:

sitting in a folder on your computer.

Speaker:

And if you're used to obsidian, it's called a vault.

Speaker:

And so you have a vault, which is a folder full of text files.

Speaker:

They're not locked away in someone else's cloud like my Apple notes are in iCloud.

Speaker:

Okay.

Speaker:

When I was using craft, uh.

Speaker:

Thus the notes were stored in their system somehow and they

Speaker:

synchronized down to my computer.

Speaker:

But

Speaker:

not only does Obsidian store the files on my computer, they're in

Speaker:

this plain text formats, right?

Speaker:

They're in a format that other apps can use and access.

Speaker:

Okay?

Speaker:

So think of, think of them as like a. I don't know.

Speaker:

A a, a Google Doc or a Microsoft Word document, but much simpler now.

Speaker:

Stay with me.

Speaker:

We're going to go a little bit deeper.

Speaker:

The files that obsidian creates on your computer are something

Speaker:

called markdown files, which if you've never heard of markdown,

Speaker:

it's just a way of formatting text using very simple characters.

Speaker:

So you have these sort of text files, which carries some kind of

Speaker:

formatting, for example, if you wanted.

Speaker:

A line of text to be the equivalent of a heading or your fur, your

Speaker:

main heading, you would use an asterisk, or a, sorry, a hashtag.

Speaker:

You'd put a hashtag in front of the text and that signifies

Speaker:

that this is heading one.

Speaker:

If I put two hashtags, it's a heading two.

Speaker:

So you can see as you're typing documents, it's really easy

Speaker:

to just go one, two hashtags.

Speaker:

This is heading two.

Speaker:

Type the heading and go to the next line.

Speaker:

You can use asterisk to help identify bold text, for example.

Speaker:

Um.

Speaker:

It sounds maybe a little bit more complicated than it actually is.

Speaker:

It takes literally like five or 10 minutes to pick the, pick up the

Speaker:

idea of what markdown files are, and they're just really quick, easy

Speaker:

ways to create text documents now.

Speaker:

This is how Obsidian works.

Speaker:

It makes these markdown files for all of your notes.

Speaker:

And the beauty of obsidian is why do I need obsidian?

Speaker:

I could just put them in any text document you could.

Speaker:

But the beauty of obsidian is, is it, is it formats that so, you're

Speaker:

not staring at symbols all day.

Speaker:

Actually looks quite, quite lovely on the screen.

Speaker:

You know, it's got some nice formatting.

Speaker:

The other cool thing about obsidian is it has this internal linking

Speaker:

system so you can link notes together.

Speaker:

And so every day I write a daily note and I'll put in that daily note

Speaker:

the different things that I've been working on, and I can link through.

Speaker:

To those da, to those items, to those projects, to whatever it is,

Speaker:

using a very simple internal linking system that obsidian has.

Speaker:

So what this means is over time you build this sort of web of connected thinking,

Speaker:

your ideas, your projects, your meeting notes, your daily notes, all the research,

Speaker:

everything links up, and you're in effect building your own personal Wikipedia.

Speaker:

And so people, including me, will often call this building your second brain.

Speaker:

So that's what Obsidian does.

Speaker:

It's brilliant for note taking, for interconnecting files and

Speaker:

building this second brain.

Speaker:

So let's plug that into Claude Code and see where the magic happens.

Speaker:

Okay.

Speaker:

Because your notes are just simple text files in a folder.

Speaker:

They're not trapped inside an app's database on the cloud somewhere.

Speaker:

Claude Code can work with those files directly.

Speaker:

It can read through them hundreds.

Speaker:

I mean, I've got thousands of notes on my system.

Speaker:

It can understand.

Speaker:

How those notes connect.

Speaker:

It can create new notes and it can slot them right into my system.

Speaker:

Okay, so a simple workflow here might be, like I say, I capture my thinking

Speaker:

in obsidian throughout the day.

Speaker:

I have a daily note.

Speaker:

I write into the different project files.

Speaker:

Meeting notes, transcripts, blog posts, even, you know, EP episode transcripts.

Speaker:

I can put them in there.

Speaker:

They're all in my vault.

Speaker:

And I write this sort of note that connects 'em all together.

Speaker:

I know where they are.

Speaker:

It's a simple folder system, simple file system in effect.

Speaker:

So then Claude Code, the second thing comes across all of that,

Speaker:

and at the end of the day, it sort of distills my daily note.

Speaker:

And it makes sure, make sure, make sure that everything's organized,

Speaker:

everything's in the right place.

Speaker:

It pulls out tasks for me so I know what I'm doing.

Speaker:

It updates project files and it connects dots that I've missed.

Speaker:

Okay.

Speaker:

Which is beautiful.

Speaker:

So the third thing then, so I create the notes first thing, second thing,

Speaker:

Claude gets involved, and so then.

Speaker:

The third thing to bear in mind with this flow is everything stays on your

Speaker:

computer in your files, which you own and only you can open, right?

Speaker:

So there's a security thing now, which is quite nice.

Speaker:

It's a bit like having a brilliant assistant who can read your entire

Speaker:

notebook collection in seconds and actually file things properly without

Speaker:

you ever handing your data to a third party, which is, is a beautiful thing.

Speaker:

So I hope you're with me still on this, right?

Speaker:

This combination of obsidian, which is all my thinking, my second brain and

Speaker:

Claude Code, which is my AI assistant who can see and act on the thinking on

Speaker:

that thinking, well, this is genuine, genuinely one of the most practical uses

Speaker:

of AI that doesn't really require massive technical ability beyond the ability.

Speaker:

To type.

Speaker:

Okay, so.

Speaker:

I get excited about Claude Code and Obsidian, and it's one of my big things at the moment.

Speaker:

All our company information, for example, our branding documents, our

Speaker:

playbooks, like on email marketing, the playbook that I gave you a couple

Speaker:

months ago on the stuff we did around narrative binding that's in there.

Speaker:

Our content plans, all past scripts from the eCommerce Podcast

Speaker:

episode, you know, the blog posts, the transcripts are on there.

Speaker:

I can search them, I can have Claude interact with them.

Speaker:

Everything is in its own separate markdown file stored in my obsidian vault, or

Speaker:

just a simple folder on my computer.

Speaker:

And again, it's secure because it's not storing stuff in the cloud, it's just

Speaker:

working on local files on my computer.

Speaker:

So Claude Code then knows our brand voice.

Speaker:

It knows our product range.

Speaker:

It knows what we've tried before.

Speaker:

It knows what's worked.

Speaker:

So I'm hoping you can see why this matters because AI needs

Speaker:

context and it has all the context it could possibly need.

Speaker:

So the default way I think most of us use ai, certainly I did for a

Speaker:

long time, was to open up a chat.

Speaker:

On a browser window or, you know, using the app, type in a question

Speaker:

and I would get my generic answer.

Speaker:

And then we realized, well, we could actually give it some kind of

Speaker:

context maybe with project files.

Speaker:

That was, that was helpful.

Speaker:

But it did feel a little bit like every single time.

Speaker:

You used it, it was like, you know, you're hiring a new employee, right?

Speaker:

They've got a little bit of information, but then they're,

Speaker:

they're still quite dangerous.

Speaker:

And so, felt like every morning they didn't really

Speaker:

need know a whole great deal.

Speaker:

And you'd have to sort of explain your business again and again and again.

Speaker:

Right.

Speaker:

But what I've built with Claude Code and Obsidian

Speaker:

is more like having a team member who has spent six months reading

Speaker:

every document you have ever written.

Speaker:

And the difference then is night and day.

Speaker:

So take writing my ideas out for this podcast episode.

Speaker:

I can use Claude Code, I can brainstorm that.

Speaker:

Right?

Speaker:

And because I'm using, because I'm doing that brainstorm in obsidian.

Speaker:

Then that's stored with all my notes as local text files.

Speaker:

It's got access.

Speaker:

Then Claude's got access to all my previous notes, all the

Speaker:

scripts from previous podcasts.

Speaker:

We've got all the transcripts, the blog posts.

Speaker:

It's got my voice guide, my brand documents.

Speaker:

It's got access to my complete slingshot framework.

Speaker:

How I think and understand about e-commerce, it's

Speaker:

got access to all of that.

Speaker:

It knows how I write what I've covered before and what our business context is.

Speaker:

And so you can see when you put all this together, the beauty of this is

Speaker:

Claude learns as well and it, because it can update its own files when it

Speaker:

learns, when I give it more info, it can update its system files.

Speaker:

Because I'm giving it more context every day.

Speaker:

That system is constantly staying up to date.

Speaker:

It's a remarkable thing.

Speaker:

Now, I don't think it's a common setup.

Speaker:

I've not many met many people that use it, but I do think it is properly powerful.

Speaker:

I've explained the system to quite a few people and it's, I

Speaker:

think it's quite transformative.

Speaker:

There are some videos on YouTube if you want to find out more.

Speaker:

If you're using any kind of note system.

Speaker:

That stores plain text files on your computer, whether it's

Speaker:

obsidian or something else that you've maybe found, you could.

Speaker:

Use this system.

Speaker:

And for me, when I stumbled across it, I thought, actually, I'm going to do that.

Speaker:

That makes all the sense in the world.

Speaker:

To me, that's going to solve so much headache I have with ai.

Speaker:

It's going to boost my productivity and it's going to make everything run a lot quicker.

Speaker:

And so I, the start of that, which actually was only, three months

Speaker:

ago, at the time of recording, three months ago, three and a half months ago.

Speaker:

Anyway, I spent, two solid days.

Speaker:

Exporting my notes from my note taking app at the time.

Speaker:

Which was craft.

Speaker:

I exported all of those as markdown files and I imported them into my obsidian

Speaker:

vault, but I cleaned them up as well.

Speaker:

That's why it took two days.

Speaker:

The whole operation would've taken an hour, but I wanted to go

Speaker:

through them, and do a bit of housekeeping because it was really

Speaker:

needed, and absolutely worthwhile.

Speaker:

So I'm glad I spent the time doing it.

Speaker:

And since then, the whole thing has just gone from strength to strength.

Speaker:

Okay, so Claude Code is my thinking partner, my co-pilot, if you like,

Speaker:

for building, for creating, for doing a whole bunch of stuff.

Speaker:

So let's talk about perplexity.

Speaker:

What, what is it about perplexity that I like, given that, you know,

Speaker:

I've just wiped lyrical about Claude Code and obsidian.

Speaker:

There's a number of things.

Speaker:

Every Wednesday morning I walk from my house to the office.

Speaker:

It's one of my favorite times of the week.

Speaker:

It takes about an hour and a half, and for me it's Prayer time, it's just

Speaker:

thinking stuff through, processing, but it's also learning.

Speaker:

And this is where perplexity comes in.

Speaker:

I love in perplexity this chat feature, which it has, which is where you

Speaker:

can have a voice conversation with it and it will talk back to you.

Speaker:

So, the other day I was in the van, I had like a 30 minute drive.

Speaker:

I just put this voice feature on.

Speaker:

I just, nothing to do with e-commerce.

Speaker:

I was just like, why did the Roman Empire fall?

Speaker:

Just this question in the back of my head.

Speaker:

I had some thoughts around it and I just wanted to clarify.

Speaker:

I must have spent 45 minutes just going back and forth in conversation,

Speaker:

with their chat feature.

Speaker:

I absolutely love that feature, but that's not the main reason I use perplexity.

Speaker:

I use Perplexity when I need to research something specific.

Speaker:

I will open up perplexity on my computer or my phone and I will go for it.

Speaker:

So for example, I mentioned the episode we did a couple months ago

Speaker:

about narrative binding, which if you don't know what that is, it's a really

Speaker:

powerful idea for writing product copy.

Speaker:

For the products you sell on your e-commerce site.

Speaker:

Now, I'm a big fan of this idea and I've spent a lot

Speaker:

of time thinking about it.

Speaker:

And that whole episode came about because I spent the day locked away

Speaker:

doing a lot of research on the topic.

Speaker:

And I did that research with perplexity.

Speaker:

And the great thing about perplexity, the reason why I like it for research

Speaker:

is because you can sense check.

Speaker:

It gives you all of the sources, so it will give you some information,

Speaker:

and then there'll be a link to the source where it got that information from.

Speaker:

Which is awesome because let's be real, AI has the ability, an extraordinary

Speaker:

talent to do this whole hallucination thing quite well, doesn't it?

Speaker:

It really, really does.

Speaker:

And so it's good to verify.

Speaker:

And you can follow the links through the sources and see what makes

Speaker:

sense or what doesn't make sense.

Speaker:

And I'll tell you what, I came away from that research session.

Speaker:

It was, I spent literally like a whole day doing it.

Speaker:

I got so engrossed in it.

Speaker:

I came away from that knowing that strategic product copy optimization

Speaker:

delivers, measurable substantial conversion improvements, which

Speaker:

is all very AI speak, isn't it?

Speaker:

Measurable, conversion improvements, et cetera, et cetera.

Speaker:

What it actually did, it, to put it in real language, is it

Speaker:

gave me the confidence to invest a little bit more time and energy.

Speaker:

Into that area by looking at the research, by looking at case studies

Speaker:

and seeing how other people were using that idea, and then understanding how to

Speaker:

apply that to my own business as well.

Speaker:

The result of that day was a 30 page document on the topic.

Speaker:

There's a lot of stuff in there, let me tell you.

Speaker:

And all that learning then became the backbone of the episode, which is episode

Speaker:

2 74, all about narrative binding.

Speaker:

If you've not listened to it, check it out.

Speaker:

It's really good episode.

Speaker:

So perplexity is a great piece of AI when it comes to doing really deep research.

Speaker:

I know now it's got new features like computer and it's trying to compete

Speaker:

with Claude Code or Claude Cowork, which is sort of another version.

Speaker:

It's more sort of user friendly version of my Claude Code obsidian, setup.

Speaker:

It started to do things like that, but I use it specifically for deep research.

Speaker:

That's where I sort of got used to it.

Speaker:

So I've got Claude Code with obsidian.

Speaker:

I have perplexity.

Speaker:

Like I said to you, the next tool on my list is Google's NotebookLM, and

Speaker:

this is a different beast entirely, and for me has a very different use case.

Speaker:

So I mentioned that perplexity will go to the web and it'll do

Speaker:

a whole chunk of research around a topic like narrative binding.

Speaker:

It will come back with its finding and give you the links to those.

Speaker:

Sources, Google NotebookLM is a little bit like that, but I give

Speaker:

it the sources to look at and it restricts itself to those sources.

Speaker:

Why would I want to do that?

Speaker:

So let's say I want to learn about a topic or I'm trying to synthesize

Speaker:

large volumes of information.

Speaker:

This is where NotebookLM sort of becomes your co-pilot for

Speaker:

getting smarter and faster?

Speaker:

I think.

Speaker:

Lemme give you an example.

Speaker:

Yesterday, I wanted to spend a little bit of time learning about negotiation.

Speaker:

The topic intrigues me, and the reason this came about was I had breakfast

Speaker:

with a friend, and he recommended to me the book, Getting to Yes.

Speaker:

I'd not read it.

Speaker:

I had read the book, Never Split the Difference with Chris Vo

Speaker:

and thought it was a great book.

Speaker:

But I'd not read.

Speaker:

Getting to Yes, which was Harvard's Strategy for Negotiation.

Speaker:

And so I got that book and realized actually these, the, they

Speaker:

approached negotiation in a, in, in, there's some similarity between the

Speaker:

two books, but there's some things where they, they're quite different.

Speaker:

So.

Speaker:

What I did, because reading a book is great and you get some ideas,

Speaker:

but I actually want to download the information into my brain and learn it.

Speaker:

Which for me means doing a whole bunch more stuff than just reading.

Speaker:

So I uploaded my, notes, my research and some other

Speaker:

stuff I'd found on the web.

Speaker:

I uploaded, I uploaded those to a notebook in NotebookLM.

Speaker:

Now I think you can add up to 300 sources of information.

Speaker:

And that can be blog posts, it could be YouTube videos, it can be

Speaker:

audio files, it can be a PDF, it can be text files like from your

Speaker:

obsidian vault, whatever it is.

Speaker:

You can upload that to your notebook and Google NotebookLM will

Speaker:

restrict itself to that information.

Speaker:

So when it comes to negotiation, I can ask NotebookLM questions

Speaker:

and it will give me answers from the research and the notes that I uploaded.

Speaker:

It doesn't give me answers from the web, okay?

Speaker:

It keeps its answers strictly to my documents, which is a great way for me to

Speaker:

ask questions and also learn from large volumes of information that I may have.

Speaker:

So I specifically want to learn.

Speaker:

The Getting to Yes, the Never Split the Difference, the and sort of what

Speaker:

that means amongst the two systems and what that might mean for my businesses.

Speaker:

So when I'm talking to NotebookLM, I want it to restrict itself to

Speaker:

that information, which is great.

Speaker:

And technically I can now do that with Claude Code and the

Speaker:

Obsidian Vault, but I think Google NotebookLM is a little bit.

Speaker:

Stronger in that area, but for me,

Speaker:

the most amazing feature about Google NotebookLM, is that you can have

Speaker:

it generate 20 minute audio files.

Speaker:

Now, these audio files.

Speaker:

It generates based on the what you want it to do.

Speaker:

So with the negotiation, I wanted an overview of the two systems.

Speaker:

It generated a 20 minute audio file.

Speaker:

In essence, it gave me a podcast.

Speaker:

Okay, and this podcast is two voices having a conversation

Speaker:

about the topic of negotiation.

Speaker:

I love podcasts.

Speaker:

I'm a big fan of listening to podcasts.

Speaker:

So what this gives me is a podcast around a very specific topic that I'm

Speaker:

trying to learn and understand about.

Speaker:

It's like podcasts on demand.

Speaker:

It's like you calling me and going, Matt, can you do a 20 minute

Speaker:

podcast on, email marketing?

Speaker:

In France selling flowers to women in age 50 to 60.

Speaker:

I mean, you can get really super specific, you know, and I go away, do that research,

Speaker:

come back and do a 20 minute podcast on it to help you and your business.

Speaker:

You've got that with Google NotebookLM. And it's the most extraordinary thing.

Speaker:

So if I take this case, like I said, negotiation, I

Speaker:

listened to that podcast.

Speaker:

I call it a podcast.

Speaker:

Is it a podcast?

Speaker:

It's an audio file, isn't it?

Speaker:

It's an overview.

Speaker:

I listened to that thing this morning and it was a remarkable

Speaker:

overview of the key information in a, sort of this audio conversation

Speaker:

base, which, like I say, works super well for me and it's so well done.

Speaker:

That I think it's pretty much impossible to tell them apart from actual humans.

Speaker:

Now, that's both scary and fascinating at the same time.

Speaker:

But if I'm trying to learn about a topic, I can upload my notes, my

Speaker:

information into NotebookLM and I can have it generate a bunch of different

Speaker:

podcast episodes for me about the different aspects of that, and I can

Speaker:

listen to those whilst I'm out walking.

Speaker:

To the office every Wednesday morning.

Speaker:

And so, you know,

Speaker:

if you fancy creating your own podcast around a topic that's

Speaker:

going to help you learn, go for it.

Speaker:

I mean, I love it.

Speaker:

I love, love, love that feature.

Speaker:

Okay, so let's move on to the last tool.

Speaker:

Nano banana.

Speaker:

Nano banana, Google Gemini's image generation system.

Speaker:

Yes, nano banana.

Speaker:

It's so much better, isn't it?

Speaker:

Than, than the other name.

Speaker:

So what I use this for, I use this product for lifestyle images,

Speaker:

product lifestyle shots.

Speaker:

So what happens here is you can take a, a product photograph,

Speaker:

upload it to Gemini, and then have it generate, products, shots,

Speaker:

lifestyle shots in different contexts.

Speaker:

Okay?

Speaker:

So the stupidest example that I could think of was the

Speaker:

very first time I tried it.

Speaker:

I uploaded a picture of our Omega-3, which if you're

Speaker:

watching on YouTube, is this here?

Speaker:

Not that you can see.

Speaker:

I need to cover my eyes.

Speaker:

There you go.

Speaker:

It take,

Speaker:

So, yeah, we've got this Omega-3 from our supplement site.

Speaker:

And the thing about our Omega-3, you'll have heard me talk about

Speaker:

this before, if you regular on the show, is it is fish friendly, okay?

Speaker:

Because the Omega-3 in those supplements comes from

Speaker:

algae and not from fish.

Speaker:

Fish don't make Omega-3.

Speaker:

Fish ingest Omega-3 from the algae.

Speaker:

So we just take it from the algae and as we like to say, cut out the middle fish.

Speaker:

And it's one of its unique selling points.

Speaker:

Every jar of Omega-3 you get from us saves the lives of 50 fish, which

Speaker:

is, you know, just fascinates me still.

Speaker:

And it's the purest, strongest form you can get.

Speaker:

So we've got a really great product.

Speaker:

It's a really good hero product for us.

Speaker:

And it worked super well.

Speaker:

So I took a shot of Omega-3, which was on a, I think it was on a transparent

Speaker:

background, or it may have been on a white background, I can't remember.

Speaker:

But I uploaded that to Google.

Speaker:

Gemini and I said to it, right, I want you to take this bottle of

Speaker:

Omega-3 and put it so it's like being held up slightly by the ocean.

Speaker:

I want the sun setting behind it and I want dolphins jumping over it, sort of

Speaker:

making that heart shape as dolphins do.

Speaker:

And I want it to be as realistic as possible.

Speaker:

I mean, it is the cheesiest image.

Speaker:

Idea that I've ever had my whole life.

Speaker:

But I just wanted to see, it was so well done.

Speaker:

I was like, we will never use that image, obviously.

Speaker:

But the fact that we could do that was amazing and just playing

Speaker:

around with it was great fun.

Speaker:

So, like I say, you can take product photos and generate them

Speaker:

in different contexts, different settings, different moods.

Speaker:

So you can take a product photo on a plain background.

Speaker:

You can put it on a kitchen counter, you can put it in the hands of kids.

Speaker:

You can do, I mean, there's all kinds of things that you could do.

Speaker:

And there's no need for the expensive studio setup, right.

Speaker:

I mean, there's no way I'm getting dolphins to simultaneously jump over

Speaker:

a bottle of Omega-3 while the ocean.

Speaker:

Lifts it up, it's just not going to happen.

Speaker:

But for AI it was a pretty straightforward thing to do.

Speaker:

Now, full disclosure, because I know I'm going to get asked.

Speaker:

Yes.

Speaker:

We still use, our photographer Lindy, who is just awesome.

Speaker:

She's brilliant.

Speaker:

And AI supplements what Lindy does.

Speaker:

AI is not replacing Lindy and.

Speaker:

This, I think is true of all AI tools, right?

Speaker:

They're supplements.

Speaker:

They are co-pilots.

Speaker:

They're not always replacements.

Speaker:

So I think learn to use AI as your co-pilots is the way forward

Speaker:

for me, for your website, social media, quick lifestyle

Speaker:

images, all of that sort of stuff.

Speaker:

Use nano banana.

Speaker:

It's brilliant, but use it as a co-pilot.

Speaker:

I will say this, video generation tools like vo and similar ones

Speaker:

are starting to make some really interesting waves, aren't they?

Speaker:

We haven't actually played around with them that much yet.

Speaker:

Played around a little bit with them, but I, not enough for me to be able to talk

Speaker:

with confidence about it on the show.

Speaker:

They are definitely on our horizon.

Speaker:

But for me it was all about getting the foundations in place where we really

Speaker:

start to try and get complicated things.

Speaker:

So that for me was all image stuff.

Speaker:

So, you know, try nano banana.

Speaker:

It's not perfect, but it is pretty good for your website, social media stuff.

Speaker:

A fraction of the cost.

Speaker:

The only thing you've really got to do well with that is write

Speaker:

the prompt and give it context.

Speaker:

But guess what?

Speaker:

If you use Claude Code and obsidian, then Claude Code's got all of that context.

Speaker:

You can use Claude Code and say, write me a prompt to use on Google's nano

Speaker:

banana that gives it enough context.

Speaker:

This is the kind of thing that I want.

Speaker:

So please write a, a really great prompt and it will write

Speaker:

the prompt for you, which you can then paste into nano banana.

Speaker:

Your whole life becomes a lot easier.

Speaker:

And actually what you could also do if you really wanted to supercharge it, is

Speaker:

you would go to perplexity and go, right, I want to do some research on how to write

Speaker:

the best prompts to generate the best images from nano Banana, specifically

Speaker:

these types of images that I wanted.

Speaker:

Do I want it to take product shots and I want it to create

Speaker:

these lifestyles type shots.

Speaker:

I want it to be on the website.

Speaker:

So I want both a, a sort of landscape version and a square

Speaker:

version and a portrait version.

Speaker:

I want this to be high res. I want 'em to be, playful in nature.

Speaker:

I want this sort of color tone to the image.

Speaker:

Whatever it is, right?

Speaker:

You'll go away to perplexity, have it go away, do some research, do some,

Speaker:

and use the deep research feature.

Speaker:

It'll go away and it'll come back.

Speaker:

You'll go back and forth, get a whole chunk of research.

Speaker:

You then put that research into Claude Code and Obsidian,

Speaker:

or put it into obsidian.

Speaker:

You tell Claude Code.

Speaker:

Here's a whole bunch of research we've done around how to write prompts.

Speaker:

I want you to write a really detailed playbook, using that.

Speaker:

As the skeleton, as the foundation, as the principles, but figure out

Speaker:

how that all applies to our business and write me a detailed playbook.

Speaker:

Then it writes you a playbook which you store in your vault.

Speaker:

So when you ever need to create an image, you go right Claude, using our

Speaker:

Google image, nano banana playbook, I want you to write for me a prompt

Speaker:

that will take a product image and create this kind of output and it will

Speaker:

create a, I mean, you're talking about.

Speaker:

Creating a prompt that is so specific and so context rich, you can put that

Speaker:

into nano banana and voila, everything suddenly becomes a lot easier.

Speaker:

And because you've got the playbook written you, you then go back to

Speaker:

next time you want to do another image.

Speaker:

It's already got all that context there, right?

Speaker:

It's already got that context there.

Speaker:

And in six months time you'll go back, you'll take that.

Speaker:

Notebook, you'll put it into perplexity and say, this is our

Speaker:

current notebook for image generation.

Speaker:

It was done six months ago.

Speaker:

What has changed in the last six months that needs to now update in here?

Speaker:

It'll go away, do some more research.

Speaker:

You'll give that back to Claude and the whole process starts again.

Speaker:

So you're always up to date.

Speaker:

That'd be one way to supercharge it.

Speaker:

Okay.

Speaker:

That is my current AI stack.

Speaker:

So hopefully you found this helpful.

Speaker:

Now I am rolling the AI out in our company and everyone in our

Speaker:

team uses AI with one exception.

Speaker:

I'm not going to name names.

Speaker:

They know who they are,

Speaker:

but here's the thing, right?

Speaker:

I found even in our own office adoption.

Speaker:

Was slow for ai and we're a tech company, right?

Speaker:

Even the dev team, were slow to use Claude Code and I wonder if this

Speaker:

is because we don't understand it.

Speaker:

And let's be frank, I don't think AI has always lived up to the hype.

Speaker:

And we may, you may have been like me, you may have spent

Speaker:

more time trying to sort out ai.

Speaker:

Sort of output than doing the actual job that AI was supposed

Speaker:

to help within the first place.

Speaker:

And I, I get that frustration having been there on many occasions.

Speaker:

But I also get that AI is moving at such a rapid pace.

Speaker:

Even this episode may well date quickly.

Speaker:

I don't know.

Speaker:

And I might have to do another one, an update probably in six or 12 months time.

Speaker:

I'm definitely going to have to do an update, aren't I,

Speaker:

to talk about what's happened.

Speaker:

But neither of those things really should be a reason that you don't

Speaker:

do anything where AI is concerned.

Speaker:

You know, I think there's a real fear factor here.

Speaker:

It does feel alien.

Speaker:

I get it.

Speaker:

We've got different competence levels across our team,

Speaker:

different technical levels.

Speaker:

We've got the dev team that's going to use it more for coding the marketing team

Speaker:

that's going to use it more for content.

Speaker:

But it's my responsibility in my company to make sure everyone's

Speaker:

getting better and better, that we have an AI policy and that we are

Speaker:

growing in our learning, because I don't think it's going to go anywhere.

Speaker:

I could be wrong, I just don't see it right now.

Speaker:

And even for some of the, say our admin team, we use in, you know, my system,

Speaker:

Claude Code and Obsidian, it's easy for me to create specific prompts that they use.

Speaker:

So they might not have the Claude Code Obsidian set up.

Speaker:

They use Claude and Project Files, so I can give it the context

Speaker:

and the prompts for those project files and they can use those.

Speaker:

It's just, it's really straightforward, right?

Speaker:

Like, the prompts we used in the narrative binding episode,

Speaker:

an SEO product description is the one that I've mentioned.

Speaker:

Well, guess what?

Speaker:

Our team who, who look after the website are using those exact prompts.

Speaker:

They're all inlaws and they're just using them.

Speaker:

They're putting the product information in and getting the

Speaker:

output out, which is great.

Speaker:

So that's sort of what we found out with the transition.

Speaker:

And I think the real lesson for me with AI is it's not, I

Speaker:

don't think it's going anywhere.

Speaker:

There is a lot of hype around it.

Speaker:

So understanding how to use it specifically for your business

Speaker:

I think is where it's at.

Speaker:

And the good news is you do not need to be a power user.

Speaker:

You really don't, you just need to find the use cases that, that sort of

Speaker:

help you do what you need to do better.

Speaker:

Right?

Speaker:

I think that's the key.

Speaker:

Like, like for me,

Speaker:

the, one of the beautiful things about AI is if I get stuck and I don't

Speaker:

know what I'm supposed to do next.

Speaker:

I just ask the ai, I go, I'm stuck here.

Speaker:

I don't understand this, or explain this, or what's the next step?

Speaker:

And you know what?

Speaker:

It's starting to really get smart because it's been really, really helpful anyway.

Speaker:

I hope this has sparked some ideas for you, and I hope you have the courage

Speaker:

to go and try some of these things.

Speaker:

And like, you know, like I always do on these solo episodes, I have put together

Speaker:

a freebie for you this month, which is my guide to using AI in e-commerce.

Speaker:

It covers the tools that I mentioned, simple getting, starting guides, and

Speaker:

of course I've added some more of the prompts in there that's going to help you

Speaker:

get started, with some of them.

Speaker:

And I've also added a priority matrix.

Speaker:

Which is going to help you know what to try and where to start because

Speaker:

I think a lot of this stuff feels overwhelming and because of the

Speaker:

overwhelm we don't start anywhere.

Speaker:

So just use that to kind of, sort of answer a few questions.

Speaker:

It'll help you figure out where you are right now and then make

Speaker:

some decisions based on that.

Speaker:

You will of course find the link to said freebie in the show notes or in

Speaker:

the description if you're on YouTube.

Speaker:

But of course you can head over to the website eCommerce Podcast dot net.

Speaker:

Just click on the resources link and you'll find it.

Speaker:

There.

Speaker:

No problem at all.

Speaker:

So go check that out.

Speaker:

It's a free download and it will be super, super helpful for you.

Speaker:

Let me, let me close this episode with a challenge.

Speaker:

Start with one thing, just one thing.

Speaker:

Pick the use case that resonates most with where you are at right now and just give

Speaker:

it a right old go for the next two weeks.

Speaker:

Not a quick play, but a proper go.

Speaker:

We're going to learn what this does and means for us.

Speaker:

You're going to get your head around it.

Speaker:

I'm going to watch YouTube videos that talk about your specific use case.

Speaker:

I'm going to do this, I'm going to do that, but I'm going to give it a right go.

Speaker:

I do not think.

Speaker:

I know there's for, again, this is just my brain and where it's at.

Speaker:

I could be wrong.

Speaker:

I do not think AI is replacing us.

Speaker:

I think it's definitely going to replace some jobs, don't get me wrong.

Speaker:

But I think.

Speaker:

Some of the roles it will replace.

Speaker:

Yes.

Speaker:

But for most of us, I think AI is a co-pilot, and I think that's

Speaker:

a really good way to look at it.

Speaker:

I think the roles that'll be replaced will be the people that

Speaker:

don't actually learn how to use AI to help them with their role.

Speaker:

The people that learn how to use AI as a co-pilot, I think will be some of

Speaker:

the most valuable people on the planet.

Speaker:

That's just my own personal opinion.

Speaker:

Time will tell whether I was right or wrong.

Speaker:

And I think the people that will figure out or figure that out are

Speaker:

going to be the ones that can pivot quickest, if that makes sense.

Speaker:

And it's going to give you, certainly in your econ business, a bit of an

Speaker:

edge, because I would say 98% of people aren't doing the stuff that we

Speaker:

just, the stuff that we've talked about.

Speaker:

In this episode, it just puts you above everything.

Speaker:

So yeah, just my thoughts.

Speaker:

I'd love to know what you think.

Speaker:

Of course.

Speaker:

So reach out to me on social media, on LinkedIn especially.

Speaker:

That's Find me at Matt Edmundson.

Speaker:

I would love to know how you're using ai, what your questions about AI

Speaker:

are, where you think it's going to go.

Speaker:

Because this whole topic I find absolutely fascinating.

Speaker:

And of course, if no one has told you yet today, let me be the first.

Speaker:

You are awesome.

Speaker:

Yes, you are awesome.

Speaker:

It's just a burden you have to bear whether you use AI or whether you don't.

Speaker:

Now.

Speaker:

Like I said, the freebies available on the website eCommerce Podcast dot net.

Speaker:

Check out Cohort as well.

Speaker:

If you'd like to come join us in the Cohort groups, it'll

Speaker:

be great to see you in there.

Speaker:

It is a free community to join.

Speaker:

There's definitely no strings attached and we just all help

Speaker:

each other out, grow in eCommerce.

Speaker:

So hopefully I'll see you in there.

Speaker:

But that's it from me.

Speaker:

Have a wonderful week wherever you are in the world.

Speaker:

I will see you next time.

Speaker:

Bye for now.