July 11, 2025 • 36 mins
In this episode of The Digital Marketing Podcast, Daniel Rowles sits down with copywriting and AI expert Kerry Harrison to explore one of the most exciting—and often misunderstood—areas of artificial intelligence: AI-powered copywriting. Together, they dive into how Anthropic’s Claude AI compares to tools like ChatGPT and Gemini, and why Claude is quickly becoming a favourite for writers who care about tone, style, and ethical AI. Kerry, a seasoned copywriter and trainer, shares her hands-on experiences using Claude to create high-quality marketing content. She explains how Claude's intuitive interface, stylistic flexibility, and alignment with ethical AI principles make it a uniquely powerful tool for communicators, marketers, and brands. But this isn’t just a tool review. It’s a masterclass in how to approach AI copywriting strategically. Daniel and Kerry explore a human-first methodology called the AI Sandwich, a practical framework for combining human creativity with AI efficiency. They tackle big questions about authenticity, originality, and prompt engineering, and show how to avoid the cookie-cutter feel that plagues so much AI-generated content. Highlights include: Why Claude’s tone of voice and writing capabilities often outperform other models for consumer copy How to train Claude on your own style and get responses that reflect your brand’s voice How to use Claude’s Projects feature to build custom tools like tone checkers, newsletter editors, and interactive customer personas A breakdown of the AI Sandwich Methodology (Human → AI → Human) for crafting content that feels human, not robotic Techniques for fact-checking, editing, and sense-checking AI outputs to avoid hallucinations and blandness How Kerry uses Claude for marketing strategy, persona creation, positioning, and ideation Tips for integrating Claude with Gmail and calendar tools to improve time management and workflow efficiency This episode is packed with actionable ideas, whether you're a content creator, strategist, marketer, or business owner. You’ll learn how to write better prompts, extract more value from AI tools, and remain fiercely authentic and creative in an age of digital automation. Key Takeaways: Train AI on your tone of voice to stand out from generic outputs Use Claude’s extended thinking and reasoning models for strategic ideation Human guidance (before and after AI) is critical to compelling content Projects in Claude can act like custom GPTs tailored to your exact needs Staying human, thoughtful, and ethical is still your biggest advantage 💡 Whether you're new to Claude or already exploring AI content workflows, this episode will give you a fresh, grounded, and strategic perspective on what’s possible, without losing your creative soul in the process. 📥 Access the show notes, tools, and links at: 📩 Join the free newsletter for access to Daniel’s monthly live sessions:
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Welcome back to the DigitalMarketing Podcast, brought to
you by target internet.com.
My name's Daniel Roll, and in this episodewe have an expert guide to copywriting
using AI and a hands-on guide to Claude.
(00:20):
So if you're anything like me, you've beenusing chatt pt, maybe Google Gemini for a
while, you've been doing some copywriting.
You think you're pretty good at writinggood prompts for copywriting and getting
good quality copy out of these toolsyou may not have played with Claude.
You may think there's a littlebit more you can explore, and this
(00:41):
is the interview that is gonnareally help you with this as well.
So I have got an interviewwith Kerry Harrison.
Kerry is an expert in using Claude.
She's worked out the real differencesbetween Claude and the other AI and
how you get the most out of them.
And she's also.
Originally by trade A copywriter.
So writing is their craft.
(01:02):
And I thought it was really interestingto see someone as a copywriter not
feeling threatened by ais, but actuallyreally embracing 'em and working out
how you get the most out of them.
And what Kerry's gonna take us throughis a step-by-step guide to approaching
copywriting for any given scenario,but also while we're doing that,
we look at how you use Claude thedifference between it and the other
ai and what you can get out from it.
(01:24):
I really learned some greatpractical tips in this one.
Uh, so over to the interview.
Daniel Rowles (01:30):
Okay, so I am here with Kerry.
Just kind of explained to peoplethat I'm not familiar with Claude.
It's been off my radarfor a while and chat.
GBT is the default for a lot of people.
So why do you recommend Claude andwhat kind of separates it, do you think
from chat GBT and Gemini and like.
Kerry Harrison (01:45):
Yeah, it's a, it is a good question.
It is so interesting 'cause Chat GTis the default for so many people
in most of my training that I run.
Literally every, if you had anyonetried any tool at Chat, gt and I
totally get it, it was the firstthat came out November, 2022.
So chat PT was the first there, andso people have just sort of grown with
it, which, and it's a brilliant tool.
It's not that I don't use it, I'm justa really big fan of Claude as well.
(02:06):
So I think it's a really great, youknow, not that dissimilar to chat
pt. It's not quite as innovative.
I feel like Cha t is always likethe first to to be, to be ahead.
They're a really innovative company, but.
I'm a copywriter by trade, soI write a lot and Claude is
really good for copywriting.
I just think it's got areally nice default home, so
(02:27):
I tend to use it for that.
In terms of how it separates, and this issomething I talk about in my training, is
that I. It's one of the things I reallylike is they're very, so it's, it's, it's
run by a company called Philanthropic.
Philanthropic is the, is the bigcompany and includes their product and
they have trained their model to alignwith a constitutional AI document.
So this document talks about thingslike opposition to inhumane treatment.
(02:50):
It talks about freedom, ittalks about, um, safety.
And so essentially the model'sbeing trained to kind of align
with those values, which Ithink is really interesting.
If you go onto the philanthropic website,the first thing you'll see is this.
Their interest in safety they're sortof slower to move, so you haven't
got, you know, all the fancy bells andwhistles that you get with our GPT, but.
(03:14):
They take their time to releasethings, they really wanna make sure
that they're doing things properly.
They talk a lot about thelong-term implications.
They talk a lot about doingthe right thing for humanity.
So I really like that kind of ethicalside and that approach that they take.
So that's one reason to use it.
In terms of writing, there's a littletool in it that I, that I use for
writing, which is helpful for anyonewho's doing any kind of writing,
(03:36):
whether it's for social media orblog writing, whatever it might be.
There's something called Usestyle, which is just accessible
from the main prompt window.
And if you press that, you, firstof all, you can decide whether how
you want Claude to respond to you.
So it might be that you want,so you can do a normal one,
which is just the default.
Or you can choose something like precise,where it'll give you something that's
(03:59):
very, um, a much more concise answer.
Or you can go explanatory.
So if you're learning something,then you click on the explanatory.
Response and it will give you somethingthat's a bit more like a teacher.
So will reply in that kind of way.
But what you can also do with that isyou can train it on your own style.
So you can upload a document, itwill scan it and have a look at the
way that you write, and then it willname that style, it'll absorb that
(04:21):
style, and then that's also thenbecomes one of the dropdowns as well.
So if you wanna train it on somethingto sound a bit like you, you can just
then go to that drop down and it reply.
It's not perfect.
It's not ever gonnasound exactly like you.
But it just gets out of the default tone.
And I re I just think that's reallyimportant with writing, considering
more and more people are now usingthese tools for their writing.
It's just like, how do we notsound like chat GPT basically.
(04:45):
So I like the fact that I cantrain it on different styles.
So I've, I've trained one for my sortof newsletter, so where I'm very much
me and I just sound exactly like I am.
And then I've got one whereI'm a bit more professional.
So I've trained on different ones,so I can just literally click and go.
I want you to reply like this.
So, yeah, things like that Ijust think are really helpful.
Daniel Rowles (05:03):
And, and that default writing style you've said is probably
better than the kind of standard chatGBT for writing as well, do you think?
Kerry Harrison (05:08):
Yeah, I, it's interesting because I've gone through
this real journey with Claude.
So there's been two model releases.
Literally in all my courses, I waslike, if you're a copywriter and you're
writing B2C copy, always go to Claude.
It's just really, really good.
And pretty much everyone I taught,PR agencies, copywriters, normal
agencies, they would just be like,oh yeah, now I've seen Claude.
This is amazing.
(05:30):
And then they did a an upgrade to3.7, so they've just upgraded to four.
And when they went into 3.7.
I started playing with it and Iwas really disappointed in it.
I think it was the first time whereI'd explored a new model and it had not
been better than it was before, justfor copywriting for everything else.
I know coders had said like, oh mygosh, the New York grade is amazing.
(05:51):
And then for me as a copywriter,I was like, oh no, it's just, no,
it's just not as good as it was.
So I was actually dropping downto the former model, and then I
reverted back to chat GPT, and Iwas talking to fellow copywriters.
Who were doing the same 'causethey were like, oh, we are really
disappointed with the Claude one as well.
Have you found the same?
I'm like, yeah.
So I felt a bit sad 'cause Ijust really like Claude for all
the reasons we talked about.
(06:11):
And then I was like, oh no.
But then they've just upgraded to four, sothey've upgraded the models again and then
it's gone back in and I've kind of got,yeah, I've like, I'm back into CLA again.
I really like Claude again.
So I was really glad that.
Daniel Rowles (06:22):
I remember we had this scheduled in and you kind of messed me.
We went, do you mind if wepush this back a little bit?
Because I'm a little bitconcerned about the new model.
And it's really interestingbecause the different versions
of chatt T can be the same and
Kerry Harrison (06:31):
Yeah.
Daniel Rowles (06:32):
and it will give you a different, slightly
different tone of voice becausethe way the model's been trained.
So it's a
One to pick out that if you'rewriting, you need to know the
model and the response you get.
And don't expect it tobe the same every time.
I
Kerry Harrison (06:42):
yeah.
Yeah.
I think that's true and that was thefirst time where I'd actually was
disappointed because I think with chatGPTI often found chat GPT was a little
bit stiff and actually I think theirnew upgrades as they've gone along, it's
actually got better for B2C copywriting.
So, you know, there's probably, it's.
There was a point where I thought,actually now Chapter GP is actually
better than Claude, and now we'rekind of somewhere in between.
(07:04):
I just, I just quite like Claude's kindof approach and this kind of slower
approach to releasing their tools.
They, they just seem to be taking thetime and thinking things through and I
really like, like, I think another thingwith Claude that I quite like, where is, I
feel like chat GPT and I know there's beenquite a lot of talk about this lately.
That it's quite syco antic, youknow, it'll tell you that even your
like, really quite mediocre ideasare genius and things like this,
(07:26):
it will, it loves to flatter you.
I feel like Claude's not quite like that.
It's a bit, it doesn'tsort of flatter you.
'cause sometimes it's like, it's lookslike occasionally you'll be like,
yeah, I know my ideas are pretty good.
Right?
Um, and then occasionally just like,do you know what JGBT just tell me
straight 'cause I don't want you tojust keep me nice to me all the time.
Um, I feel like Claude's a bit morerealistic and I quite like that.
Daniel Rowles (07:47):
A critical friend, which is good.
Kerry Harrison (07:48):
Yeah.
Daniel Rowles (07:49):
So they've upgraded the model recently,
Kerry Harrison (07:50):
Yeah.
Daniel Rowles (07:51):
you think about it?
And I think there's like a defaultand a reasoning model, isn't it?
Kerry Harrison (07:54):
Yeah.
Yeah, that's right.
So the new model you have now, ifyou're in the, if you pay, if you're
free, if you're in the fee version,and, and really do use the free version
if you've never tried call before.
'cause it, it, you know, it'sa great place to start if
you are in the paid model.
You also get the reasoning option as well.
So you've got four Sonet, which you getfor free, but you just don't get it.
(08:18):
You just have a limited use of it.
And then when you're on a page, you canobviously use it for much longer, but
you also get four Opus, which is likethe most advanced model that they've
got, which is quite high powered.
It's really good forthinking things through.
Critical thinking and, and you tendto think with chapter pt, with our
reasoning models, they tend to bemore suited to things like stem and
(08:39):
maths and coding and things like that.
I feel like with the Claude one,it's also pretty good for content
creation, for things like strategy.
So I do use it for, if I'mthinking more strategically.
And more recently I've been lookingat my own business and my positioning
and how I might move forward.
And actually, Claude's been a really good.
Thought partner for workingthrough those kind of things
(09:01):
as well with this larger model.
Just it's, it, it's just much more,yeah, it's kind of critical thinking.
It's reasoning, it's, it movesthings through step by step.
So I found that really power,a really powerful model.
Daniel Rowles (09:14):
And there, there's a lot of debate about AI content and
chat creating stuff that's quite,you know, generic sometimes as well.
How can you use Claw to makesure that that copywriting is
kind of authentic, compelling?
It's gonna be original as well.
I.
Kerry Harrison (09:26):
Yeah, it's, it's a good question.
I think there is a danger with chat GPTbecause it's the default model for a lot
of people, and I, I think we're, we'refinding now that because so many people
are using it for copywriting or content.
Actually people are getting pretty savvyand they know that it's generated with
our GPTI, I think you don't have to useit for that long to, to recognize it.
(09:47):
The certain phrases that it picks up,like every time I set create any kind
of, uh, blog or if I ask it to thinkabout anything that's related to AI
and and writing or AI content or AImarketing, it always starts the line.
Something like, in this fast baseddigital world or something like
that, like every single time.
Daniel Rowles (10:05):
is the one I've seen a lot.
Yeah.
Kerry Harrison (10:06):
And you see it a lot.
And then I see it in blogs and emailsand stuff like this that it's like
you've, you've generated that with GPT.
But I was talking to and someone theother day, she's a recruitment consultant
and she was saying literally nine overlike 90% of the covering letters she's
getting at is are made with G pt.
And she can just tell straightaway now, and she's just kind
of working out, you know.
(10:27):
Maybe I need to look at ones thataren't generated by charity, GBT.
'cause I've actually made the effort todo it, which is, which is interesting.
And so I, I think with chatGBT, and I think it's the same
with all the models, really.
We have to be coming out of thedefault tone as a de a default.
You know, we have to make sure that we'renot just sticking with what's there.
I suppose with Claude, just becauseit's a bit more of an underdog, I
feel like less people are using it.
(10:48):
So you might be able to get awaywith just going with the default
tone for a little bit longer.
People pull before people go like, oh,it's, it's, she's doing it with Claude.
Which I don't think youwould necessarily pick out.
So there's that.
But I think also just the fact that it'sgot that little abuse style thing where
you can train it on different styles.
But I think a lot of authentic, creatingauthentic content with AI comes down
(11:10):
to also bringing in the human element.
And that's regardless of whether you'reusing Claude or Chacha, bt, or Gemini, it,
the same rules apply really, you have to.
Put in the work as in you have to use,you know, we've got amazing brains.
We should keep using them.
One of the key things that I worry aboutwith AI is this idea that of de-skilling,
you know, we'll just kind of go, oh, I'llget to do it, or I'll get cla to do it.
(11:32):
And, and we just, I, I reallyfear that we'll just stop.
Doing our thinking, our critical thinking.
So, I have this methodologythat I talk about in all of my
presentations at conferences and stuff.
Uh, it is called the AI sandwich, whichI know is not the most sophisticated
name for a methodology, but it, it worksreally well and it people seem to remember
it and go, oh, I love your sandwich.
(11:53):
So for me to get the best writingout of AI models, and I guess it,
it expands wider than writing is.
It's a hu a human AI human process.
It's a three part process.
I. And it started with this idea,'cause I wrote this creative
methodology in, in January, 2020.
Where are we now?
2024. So I wrote it.
(12:14):
So back end of 2023.
I was thinking actually, I think thebest way to get great writing out of
something like charge BT or code is touse AI to generate a first draft and
then for me as a human, to go in andedit it and craft it with my own skills.
And then what I realized towards theend of 2023 was actually the prompting.
(12:34):
Prob is equally, if not moreimportant than what we do afterwards.
So that's where things reallychanged me back at, back at 2023.
So almost like a year into chapter GBT.
It's like, hang on a minute.
The prompting is so blooming important.
So I thought actually we need to dowork before we even come to the tool.
We need to have thought about,want to come to tool with.
At the very least an objective.
(12:56):
'cause a lot of people now justgo, oh, I'll just go to chat GPT.
But they haven't really thought throughwhat they wanna create or why, or how
that links to their wider marketingstrategy or their content strategy.
You know, just kind of go,oh, I'll just do it with that.
So for me, I think we need tocome with an objective, but also
to think about other things.
So how does it link to our strategy?
What kind of what?
Of my own ideas might Iwanna incorporate into this.
(13:18):
You know, if you're writinga blog, what's the angle?
Tell the model what your angleis so it moves in that direction.
So doing all of that thinking,taking time on your prompts to
get as much detail as you can.
I tend to say in my, in my courses, thatif you think about it as an intern, so
if you are asking an intern to write copyfor you, if you're asking 'em to write
(13:39):
a blog or a social post, you are notjust gonna say to them, go write me a.
Blog post on, I dunno,workplace wellbeing.
You're gonna say, I need to writea blog on workplace wellbeing.
This is who it's going to, this is whereit's, it's gonna go in this publication,
or it's gonna go on our website.
It needs to be, you know,500 to 800 words long.
You probably would alsosay, I've got these ideas.
(14:00):
I'm thinking it probably needsto include this, this, and this.
So maybe start there.
Uh, have you got any questions?
And basically that's what you woulddo in your prompt, you know, that,
think about it as an inter, whatinformation would you give it?
Think about context, bit of background.
How long do you want it to be,where it's going, who's your target
audience or all of that stuff.
And so for me, putting, being reallydetailed in the prompts then guides
(14:23):
the model in the right direction.
So.
And I think what also really kind ofstill slightly intrigues me is the fact
that it takes me a lot longer to writea prompt that does for AI to generate.
So it will take me awhile to get my prompt.
Perfect.
And then what It will generate theresult in like a matter seconds.
It's like that's wild.
But I, yeah.
(14:43):
But I do think that itmakes a, a huge difference.
And also to bear in mind that a lotof people are coming to the models
and just giving them a one sentence.
Very unthought through prompt.
And so what you're gonna get is prettyvanilla, pretty generic, pretty,
oh, it's definitely been writtenby SHA G pt. Um, so I think doing
that e early thinking, so that's thesort of human bit of the sandwich.
(15:05):
And then in the middle, whichis the filling, is the ai.
So then you get the AI to do onceyou've done your thinking, get
the AI to go off and do its thing.
And it's brilliant and it's fastand it's quick and it will give you
loads of it, give you 20 ideas in afew seconds, give you overviews or
insights or data analysis, and thenyou come back to the human again.
And when it comes out and you'vegot your res, your response from.
(15:26):
Whatever model you've used, the firstthing we need to do is, is fact check it.
So as you know and anyone knows who'susing these tools, hallucinations
are a thing and we, it's not,it doesn't hallucinate all the
time, but it may have done.
And so we have to check it.
So fact checking everything, anykind of stats or facts that come
back, just make sure they're truebecause they sound so viable, but
you know, it might not even exist.
(15:48):
And then we also sense checking itis another thing that I do because
I just think that it's really easy.
It's so easy to get wowed by the,the AI thing, the fact that you
generated a blog in what, 20 seconds?
You kind of go, oh great, I've got a blog.
Brilliant.
And I also think on first read,sometimes you can go, oh, that's
(16:08):
great, but then on second readyou're like, oh, do you know what, it
hasn't quite got the depth I wanted.
Or maybe it's a littlebit repetitive here.
So.
I think sense check and just pushing pastthe kind of, wow, and actually looking
at what you've got, and again, thisis where we need to use our analytical
brains like am I happy with that?
Another question I ask myselfis, if I wrote this from scratch
(16:29):
or created this from scratch,would it look anything like this?
And if the answer's no.
Probably need to think, well, youknow, maybe need to go back in and
think about the problem because it,you know, it, the fact is we're,
we're quite lazy naturally, aren't we?
And if, if we've got a blogdone in 10 seconds, it's like,
brilliant, I'll go and walk the dog.
Um, actually you justneed to like analyze it.
Uh, and then the last thing is craft.
So I still feel like in this.
(16:51):
Point in time that we still need peoplethat know their stuff, people who have
taste, people who have experience.
And so for me, adding the human craft ofcopywriting, which I've done for 23 years,
you know, I've been writing for 23 years,I still think it makes a difference.
You know, I know what goodlooks like, so it it, it helps
me to analyze the copy and go.
(17:11):
I think that's good.
Or this, it needs more ofthis, or it needs less of that.
So yeah, I still say to copywriters,there's a lot of copywriters are like
really worried about their work, but I,I still think there's a real need for
people who have deep expertise becauseI also think, you know, there's a lot of
talk around AI at the moment with the,the deep research tools where it says,
oh, you can have a, a PhD research 24 7.
(17:32):
I always think it's a bit weird 'causeif it generates something at PhD level,
but I haven't got a PhD in that thing.
It's like how do I actuallyknow if there's any good or not?
So I still think we need to beapplying or critical thinking.
So that's the kind of sandwich.
So prompting thinking.
Briefing it like an intern, you know, ifyou are really putting in your thought
and actually thinking of your own ideas.
So rather than just relying on the modelto come up with ideas for you, what
(17:55):
would you like to see in your blog thatyou're just about to write or your social
posts that you're just about to create?
Then the AI does its thing and then you goin sense check, fact check ad your craft.
So that's how you get more authentic.
Write writing.
I mean, there's a load more tothe prompting and stuff, but
I basically, if you wanna makesomething sound really beautiful,
you need to add a little bit of you.
Daniel Rowles (18:13):
Brilliant.
I love that.
I mean, I keep coming back to thisphrase of like, lean into your humanity.
When everything's AI andeverything's deep fake, that
becomes really, really important.
What about the pro, theproject feature in Claude?
How can you use that inyour kinda working process?
Kerry Harrison (18:24):
Yeah, so project, the projects is really good.
The projects is a pretty much anequivalent to the custom GP and gpt.
Daniel Rowles (18:31):
Right.
Kerry Harrison (18:32):
For me, that is a reason for me to pay for Claude.
And now also because they've got the, youknow, this more advanced reasoning model.
So projects allows you to create alittle area, or what they call a project.
You give it specific instructions,you give it knowledge.
And then it, you go into that spacewhenever you want to do something.
(18:53):
So for example, I've got quite a fewdifferent projects where I go in.
So I've got ones that do very simplethings, like I've got one called Ned
who's a newsletter sniper, basically.
So I have a newsletter,I write my newsletter.
I. And occasionally I'll make them myarticles in my newsletter a little bit
too long, and I want it to be snipped.
So rather than me typing out a very longprompt every time, which basically says, I
(19:17):
want you to, I want you to change this toa different word count, but I don't need
to take out my jokes, all my content, allmy, you know, I, I've got this whole like
long spiel of what I don't want it to do.
So instead of putting that intoa prompt every single time, I
just wanna go and snip a bit.
My newsletter.
I've created a project, so justas I would with a custom GPT.
So I've given it that there's details.
So when I put something, give yousomething, can you, you know, edit it
(19:39):
down, ask me what the word count it needsto be, and then I want you to, you know,
keep as close to the content as possible.
Don't take out my jokes.
Not that they're any good in thefirst place, but you know what I mean.
Stay as exactly on tone, but just cut it.
That's all I want you to do.
And so it's just really easy now 'causeI just literally drop in an article
that I've written and it will just tellme, ask me what co workout has done.
(20:00):
So it's just really, really easy.
So I've got, I've got ones like that.
I've got projects that are, I'vegot like my customer personas.
So one of my customerpersonas is founders.
So founders of businesses.
And I've given it this really,really detailed persona.
So marketing persona of a of a founder.
And then sometimes I'll go in andI'll just wanna say, you know,
(20:22):
what do you think of this idea?
Or if you know, as a founder,what, how do you feel about this?
Or I might take in my copy and say alanding page that I've just written
or a new sales page for a coursethat I'm creating and say what
resonates about this and what doesn't?
And so things like, that's quite nice.
So.
This idea of, I really like theidea of creating this kind of
interactive persona because.
(20:43):
Generally as a, as a marketer, as a, awriter or any kind of creative, you get
given these personas and they're basicallyjust a made up person on a piece of paper.
And I, I love the fact that now you canget something, create a really detailed
persona and put it in, and then you,you can actually interact with it.
And I know it's not the same astalking to an actual customer.
Then in reality, the onesyou get on a piece of paper
aren't a real customer either.
(21:03):
It's, you know, astrategist has made that.
And so I put it in.
So I've got, I've got aproject that does that.
Again, it just means I don't have to sayto it, you are this role, this is what
you believe, this is who you are, thisis your, what demographic you're in.
It's just all there in the project.
So I can just say, whatdo you think of this?
So really, really quick I'm tryingto think what else I've got in there.
I've got ones that helped me writeLinkedIn post, so I've given it really,
really detailed granular instructions onhow to write a blog style LinkedIn post.
(21:28):
Just go in with what my current insightis and it will help me write one.
Um, I've also got a tone of voiceone, so slightly more detailed than
the, the one that you get if youjust go down off the prompt box.
But, you know, really detailed viewof how I work, my tone of voice and
what it sounds like and how I writeeverything from the syntax to the the
sentence lengths and all the rest of it.
(21:48):
So it's all in there.
And again.
If I've written something that's not,maybe not exactly as I wanted it to
sound, it's like, I want it to soundlike my normal, like newsletter voice,
then I can just drop it in there.
So, and the same, I suppose ifyou can create tone voices for
all different things, differentbrands, different businesses.
I've got one that's reallybold and irreverent.
Sometimes I'm just like, ah, Ijust wanna create something that's
(22:09):
a little, feels a little bit.
A bit more cheeky today.
So I'll lit. I've literally trainedon this, like really irreverent tone
of voice and I just go in and justsay, make, just make this, give this
some life, and then I'll just bringit and bang it in and off it goes.
So those are just some examples.
There's loads where there's, I've alsogot business mentor as well, where
I've given it loads of details aroundmy business, my belief systems, the
(22:29):
things that I love, my, my methodologiesand my ethos around ai, which is.
Hopefully as you gather, very mucha kind of human and AI combination.
So I've given it all thatinformation about me.
So then if I'm working through somethinglike something in my business, I can just
go to that project because it alreadyknows all the information about it.
I don't have to give ita massively long prompt.
(22:49):
So.
Basically anywhere that youwould use a custom GPT, you can,
you can use called projects.
And to be fair, I've got the,they, I've got ones in chatt, ones
in and Claude and yeah, I've gotsome that are exactly the same.
I've just taken instructions fromone and brought it over to the other.
So it, and it's notdifficult to make, so, yeah.
Daniel Rowles (23:07):
was just gonna say, some of the ideas you were coming up with
there are just absolutely brilliantand actually apply to all of the AI
systems as well that you could use.
So I think there's somebrilliant ideas in there.
So, and then this kind of approach toprompt engineering, how do you do that
with different types of marketing content?
Do you approach it in adifferent kind of way?
Kerry Harrison (23:22):
Yeah, I, I guess so.
I think for me, I've talked a lotabout the AI sandwich and a lot
around what goes into a prompt.
So I think a lot of that still stands.
It, it's just being as,as detailed as you can.
I think a couple of other thingsthat we do in terms of prompting is.
To give the model examples, and Ithink that doesn't matter whether it's,
(23:43):
it's called or charity deal, any modelthat you are using, if you show it
what good looks like, I really thinkthat makes a big difference in terms
of guiding in the right direction.
So recently, for example, I wasworking on some case studies.
My case studies follow apretty similar structure, so
it's like, what's the problem?
What was the solution andwhat were the results?
So a three-part process, essentially, Igave it two examples of case studies that
(24:07):
had already written and then a load ofbullet points about the new one that I
was just, that I was just about to write.
Um, and then gave, and then ask itto do, to have a look, to use the two
examples as a guide to tone and style.
And then I was, I was actually really.
Really surprised at what it generatedas in like it stayed really.
'cause obviously AI lovesa pattern, doesn't it?
It's the whole thing is, you know,statistical prediction models.
(24:28):
So it loves a pattern and I, and itreally stuck to that pattern very well.
So it get, it made the new case study forme, it, it structured it in the problems.
It solution results.
It get like headlines and subs.
I was a bit like, wow, thisis, this is pretty good.
So I think giving it examplesmakes a, a really big difference.
If you're, if you're writingblogs, give it a couple examples
of blogs you've written.
(24:49):
If you're writing headlines, ratherthan just trusting it to do some great
headlines, get, show it examples ofone's headlines that you've done that
are really high performing or that haveworked really, really well for you.
So it, again, it has an understandingof your, your tone and your style.
So I think examplesmakes a big difference.
I guess the only thing, justto be aware of examples is.
Yes, it makes a massive differenceputting examples, but just, especially
(25:11):
if you're on the free models, just becareful what you put into the models.
I'm always very much like, don'tput anything private, confidential,
proprietary, into the free models.
So just be aware of that.
And also if you're writing press releasesand you wanna give it some examples,
just make sure you press releases onand bargained and stuff like that.
Write writing reports.
Just make sure that you're happywith, with what you're putting in.
Nothing confidential or private, but.
(25:32):
Generally speaking, if you,if it's, if you're okay to
put it in, put it, put it in.
And I think that, that make,that will make a significant
difference to the output.
Daniel Rowles (25:40):
And then what about going beyond copy, um,
and kind of content creation?
What, what else is, is Claudeparticularly good at, do you think?
Kerry Harrison (25:48):
Yeah.
I think I mentioned earlier, but I, I useit for a lot of marketing side of things.
As I mentioned the, thecustomer personas thing.
I really like that, and I can askyou about copy, but I could also
talk to it generally about anykind of strategy or, or marketing.
, as I was talking about positioning,so, so any kind of like strategy,
content, strategy, so not justactually the writing, but it helping
(26:11):
to inform your writing as well.
So it might be at looking at customers,looking at, their pain points.
Looking at I've got this prompt where Ilook at a customer's fears, frustration,
streams and desires, and, you know,create this whole scenario of them and
what they, what they might think abouta product or a service, which can then
help me to again, look at those fearsand frustrations, streams and desires
(26:32):
and go, okay, so maybe I could createsome content around that or around that.
So I think strategic stuff isreally good for my own business.
I've been doing, working a lot onpositioning, as I mentioned before.
So it's been, I've been working at.
Through that, I think it makesa really good thought partner.
So not necessarily giving me the answers,and sometimes I think it's quite useful to
(26:55):
not ask for answers, but to ask for ideas.
So.
Just as a different way of prompting.
'cause I think we often go tothe models and look for, look for
answers, but actually sometimes justgetting ideas back is really helpful.
So I like the fact that I, it mightcome back with something and I'll be
like, oh, that's really interesting.
And it might take my thinking offin, in a new direction or what I've
done with positioning because I've,I've had quite long conversations
(27:17):
with court about my positioning.
That sometimes it'll bring somethingback or an an idea back to me and then
I'll look at it and go, it is definitelynot that I don't want to do that.
And it's made me question my, oh,it's been really good at just helping
me even tap into my own intuition.
'cause it'll present something back tome or a target audience and I'd be like
no, that I, I wouldn't wanna do that.
I don't wanna target these peoplethat would not make me feel very happy
(27:39):
or wouldn't make me feel fulfilled.
So I, I've really lovedthat side of it as well.
'cause if it just gave me an answer.
I don't think I would necessarilygo through those thought presses
and, and think, what do I actuallythink about this right now?
Or like, am I happy to do that?
Would that make, you know,would that fill me up or not?
So I've, I've used it for that.
I think brainstorming, it's reallygood for not just for copy, but for
(28:02):
content, just for life, to be honest.
Just anything where I've got ideas.
Anytime that I just feel like there was apoint in time where, you know, often it'd
be like, oh, I just don't have any ideas,or, I'm really, really tired and it's
five o'clock and I just need some ideas.
And I, you know, the fact that I can getideas within a matter, you know, 10 ideas,
20 ideas in a matter of seconds, andthey're not gonna be perfect by any means.
(28:23):
And they're, if you're doinganything highly conceptual.
As much as I want AI to help me to dothat, I don't think it, it can do it.
You know, it's things like strap lines.
It's just really poor.
It just generates aload of generic outputs.
And so I sometimes start with it,but I always have to use my own
brain for those kind of things.
But for, for general sort of ideastarters, I think it's really helpful and.
(28:44):
Normally I'm just, it justgets over the blank page thing.
So I'm like, I don't panic anymore.
It's like, okay, wellI've got something down.
That's the start.
And then I, and then I'll be like,okay, so that's quite interesting.
Or I quite like that angle.
I hadn't really thought about that before.
So it might take mythinking in a new direction.
It might be that it comes backwith two ideas if you blend
them together with pretty well.
(29:04):
Or, so I just I like it forbrainstorming, for getting me
started, but yeah, hopefully that's.
Daniel Rowles (29:10):
Yeah, definitely.
And I, I, one of the things I found withit is that, and I do this at chatt pt,
I'll get it to come up with five differentideas and then compare and contrast them.
And I'm not necessarily gonna useone of those ideas, but there's the
thought process that it goes through.
Kerry Harrison (29:22):
Yes.
Daniel Rowles (29:23):
its conclusions is what's gonna spur my thinking with this as
Kerry Harrison (29:25):
Yeah.
Yeah.
I think that's, that's really true.
And actually with Claude, you'vealso got extended thinking.
So extended thinking is anotherlittle toggle that comes off.
Daniel Rowles (29:33):
research mode,
Kerry Harrison (29:34):
yeah, it's, it's a bit like deep research, but you
just literally toggle it on andit will show you its thinking.
So it's a little bit slower.
Daniel Rowles (29:41):
Fantastic.
Okay.
saw that in deep seekand that was one of the
Kerry Harrison (29:44):
yes.
Daniel Rowles (29:45):
Deep
Kerry Harrison (29:45):
Yeah.
Daniel Rowles (29:46):
reasoning
Kerry Harrison (29:46):
Yeah, so I, I really like that if you just, if you're just
wanting to work through something,you want it to slow down a little
bit because obviously if you use thedefault models, it, it's pretty quick.
If you want it to slow down or youwanna think slightly wider or get it to
do more, consider like considerationsand come back with multiple
considerations or slightly differentcreative angles, things like that.
If you take click on extendedthinking it will then.
(30:10):
It will just give you more andyou'll see its thought processes
and it's like sometimes you'relike, oh, that's really interesting.
'cause I wouldn't have necessarilygone down that route myself.
But if I do go down that route, Iwonder where that would take me.
So I, I do, yeah, I do think as athought partner, it's really useful.
I,
Daniel Rowles (30:24):
then what about things like data and analytics?
Kerry Harrison (30:26):
So it's great for data analytics and I think,
you know, not, I suppose.
Comparable with chart GPT, andit's also really good visuals.
So if you ask it to do data analytics,it will also generate graphs for you or
diagrams for you and things like that.
So in terms of marketing, if you'reanalyzing campaigns or how your
social media's going or thingslike that, it's really useful.
(30:47):
You can, you'll get that results and youcan make them pretty, which is not, and
choose your colors and all the rest of it.
Which, which is great forany kind of reporting.
So I use it for that.
I've also turned on, recently, they'vegot now introduced in the last month or
so, integrations, so you can now integratewith 'cause they've be, they've had
investment from Google um, anthropic.
(31:08):
So they had a lot ofinvestment from Google.
They also had huge amountsof investment from Amazon.
So I think it was
6 billion of investment from Amazon.
So I think they've got these links withGoogle, so you can link it up literally
with really easily, just literally clickthrough, link it through to your calendar.
Just check if you're gonna link it throughto your calendar and and Gmail just.
Check the privacy policy andmake sure you're happy with that.
'cause I know, you know,have to be careful.
(31:29):
It's like I haven't yet linked it upto my, uh, Gmail, my, like main Gmail
because it's also my personal Gmailand it goes back all, you know, so far
now there's probably loads of personalstuff out about all sorts of things.
I'm like, I'm not sure if I wanna giveit that, but I've given it my business.
My business account.
So that's been really useful.
So a couple of things I've donewith that in the last week or so
is to get it to tell me what aremy main priorities for this week.
(31:53):
So I'll have a look at my calendar.
It will look at my, gmail, andit'll say, you need to do these.
And it, you know, sometimes it'll say Youreally need to get back to this person.
Because this seems like it'sgonna be really important for you.
So, so I like that.
So it'll prioritize.
Or even my day.
So today I just said to it, whatare my priorities for today?
So obviously I knew it got thepodcast, so that was on there.
(32:14):
So that's, that's a good one.
It can analyze that.
Um, I've also asked it to analyze mytime, like how am I spending my time?
And that's really interesting froma data analytics point of view.
And, and, and it even came back'cause then I was saying, okay,
so now you've analyzed my time.
Tell me how I can be more efficient.
Really interestingly, it said somethinglike, you'll have, you are doing too
(32:36):
many 20 minute introductory calls thataren't necessarily leading to full work.
So how wide, how about creating a,you know, a questionnaire so that
when people come through to you fora call, you've already vetted them.
And I was like.
God, that's, you know, andobviously I would've thought,
could've thought of that.
In fact, my friend suggested to methe other day, but I just thought it's
(32:56):
amazing that Claude has just been ableto look at my, the way I spend my time.
It said to me, you probablyneed to do a bit more exercise.
So it's like, I'm not, you know, oftenI'd have like yoga classes in definitely
the last few weeks I've been so busy.
There's not been any yogaclasses in my, my calendar.
So it's just, it's pretty incredible whatit can do and just the different angles
that it can look at your time from,you know, not 'cause I asked it how I
(33:19):
can be more efficient, but I could askit, how could I improve my wellbeing?
And it may well say, actually youneed to put 10% more time into.
Going to yoga or going runningor whatever it might be.
So that was pretty life changing.
So I, I found that really good.
So yeah, how I spend my time.
And then another thing, I tried it for theother day, I get so many AI newsletters
(33:39):
and I love reading them, but the lastfew weeks I've been so busy and I just
said to me can it, can you scan my AInewsletters that I've received in my
inbox and give me a summary of each one.
So it just said, youknow, this is this one
Daniel Rowles (33:52):
that.
That's
Kerry Harrison (33:52):
that, yeah, so it just meant that because I haven't had time,
it's just summarized the newslettersthat I've had through in the way that
it thinks is most relevant to me.
And then what I would do then, becauseI've been so strapped for time, I
probably, now I've got this summaries,I'm gonna go, actually, do you know what?
I'm gonna read that one inthat one from top to bottom.
'cause I think that's gonnabe super useful for me.
(34:13):
So I would feel like.
Those integrations, which havegone in very recently could be
so useful for work, for planningfor your own personal wellbeing.
So many different things from a businesspoint of view that it can guide you on.
So that's, yeah, for me, that'sbeen a real game trip flight.
A total game changer, actuallythe last couple of weeks.
Daniel Rowles (34:32):
interesting 'cause chat, DBT are going down that route as
Kerry Harrison (34:34):
Yeah.
Daniel Rowles (34:34):
it's
Kerry Harrison (34:34):
Yeah.
Daniel Rowles (34:35):
see which tools that are gonna integrate into
what and how that changes things.
So, look, really great introductionto Claude but actually loads of
ideas in there that apply to ai.
Full stop.
So
Kerry Harrison (34:43):
Yes.
Daniel Rowles (34:44):
much for those Kerry and, and the time you've given us.
Tell people a little bit aboutwhat you do on a day by day, the
services you offer and how peoplecan get hold of you if they want to.
Kerry Harrison (34:51):
So I spend most of my time training people in how
to use AI for their marketing.
So everything fromcontent to data analytics.
Sometimes I do very specificcopywriting, you know, all day
copywriting for copywriting teams.
So it depends what you need.
Uh, I also speak at conferences and um.
Yeah, sit on panels,all that kind of thing.
I've got a newsletter, it'scalled Minds and Machines.
(35:12):
It's on Substack.
, I go in there and I give tips andkeep you up today with the latest news
and normally share some of the thingsthat I've learned and I'm exploring.
I'm on LinkedIn, Kerry Harrison.
So you'll find me, uh, I talk a loton LinkedIn, not just about AI and
all the incredible things it cando, but very much about the human.
How we can keep thinking, how we canstay creative, how we can nurture
(35:33):
everything that makes us human.
I just feel like, as you saidearlier, I feel like that's
gonna be incredibly important.
And outside of the ai.
I'm also a yoga teacher and Irun women's circles in the woods.
So this next week I'm actually runninga retreat down in the Woodlands,
near Salisbury with a, um, reallysuper experienced a retreat host.
So off, pretty much off grid, workingwith analog creative practices, writing
(35:59):
with pen and paper sitting around a fire.
, So I do all of those things too.
And so somehow my whole life isall just trying to smash these
two worlds together of high tech.
But also just what it means to justsit on a log around, a fire under
the stars and to just go, right,how can we bring these two together?
'cause for me, they feel like they'retwo sides to the same coin almost.
Daniel Rowles (36:19):
Brilliant.
Well, we'll put all of thatinto the show notes, thank you
Kerry, so much for your time.
Some brilliant tips there and I'm pleasedo let us know on what you think and
your takes on all of this as well.
Kerry Harrison (36:26):
Cool.
Thanks.
For more episodes, resources toleave a review or to get in contact,
go to target internet.com/podcast.
Welcome back to the DigitalMarketing Podcast, brought to
you by target internet.com.
My name's Daniel Roll, and in this episodewe have an expert guide to copywriting
using AI and a hands-on guide to Claude.
(00:20):
So if you're anything like me, you've beenusing chatt pt, maybe Google Gemini for a
while, you've been doing some copywriting.
You think you're pretty good at writinggood prompts for copywriting and getting
good quality copy out of these toolsyou may not have played with Claude.
You may think there's a littlebit more you can explore, and this
(00:41):
is the interview that is gonnareally help you with this as well.
So I have got an interviewwith Kerry Harrison.
Kerry is an expert in using Claude.
She's worked out the real differencesbetween Claude and the other AI and
how you get the most out of them.
And she's also.
Originally by trade A copywriter.
So writing is their craft.
(01:02):
And I thought it was really interestingto see someone as a copywriter not
feeling threatened by ais, but actuallyreally embracing 'em and working out
how you get the most out of them.
And what Kerry's gonna take us throughis a step-by-step guide to approaching
copywriting for any given scenario,but also while we're doing that,
we look at how you use Claude thedifference between it and the other
ai and what you can get out from it.
(01:24):
I really learned some greatpractical tips in this one.
Uh, so over to the interview.
Daniel Rowles (01:30):
Okay, so I am here with Kerry.
Just kind of explained to peoplethat I'm not familiar with Claude.
It's been off my radarfor a while and chat.
GBT is the default for a lot of people.
So why do you recommend Claude andwhat kind of separates it, do you think
from chat GBT and Gemini and like.
Kerry Harrison (01:45):
Yeah, it's a, it is a good question.
It is so interesting 'cause Chat GTis the default for so many people
in most of my training that I run.
Literally every, if you had anyonetried any tool at Chat, gt and I
totally get it, it was the firstthat came out November, 2022.
So chat PT was the first there, andso people have just sort of grown with
it, which, and it's a brilliant tool.
It's not that I don't use it, I'm justa really big fan of Claude as well.
(02:06):
So I think it's a really great, youknow, not that dissimilar to chat
pt. It's not quite as innovative.
I feel like Cha t is always likethe first to to be, to be ahead.
They're a really innovative company, but.
I'm a copywriter by trade, soI write a lot and Claude is
really good for copywriting.
I just think it's got areally nice default home, so
(02:27):
I tend to use it for that.
In terms of how it separates, and this issomething I talk about in my training, is
that I. It's one of the things I reallylike is they're very, so it's, it's, it's
run by a company called Philanthropic.
Philanthropic is the, is the bigcompany and includes their product and
they have trained their model to alignwith a constitutional AI document.
So this document talks about thingslike opposition to inhumane treatment.
(02:50):
It talks about freedom, ittalks about, um, safety.
And so essentially the model'sbeing trained to kind of align
with those values, which Ithink is really interesting.
If you go onto the philanthropic website,the first thing you'll see is this.
Their interest in safety they're sortof slower to move, so you haven't
got, you know, all the fancy bells andwhistles that you get with our GPT, but.
(03:14):
They take their time to releasethings, they really wanna make sure
that they're doing things properly.
They talk a lot about thelong-term implications.
They talk a lot about doingthe right thing for humanity.
So I really like that kind of ethicalside and that approach that they take.
So that's one reason to use it.
In terms of writing, there's a littletool in it that I, that I use for
writing, which is helpful for anyonewho's doing any kind of writing,
(03:36):
whether it's for social media orblog writing, whatever it might be.
There's something called Usestyle, which is just accessible
from the main prompt window.
And if you press that, you, firstof all, you can decide whether how
you want Claude to respond to you.
So it might be that you want,so you can do a normal one,
which is just the default.
Or you can choose something like precise,where it'll give you something that's
(03:59):
very, um, a much more concise answer.
Or you can go explanatory.
So if you're learning something,then you click on the explanatory.
Response and it will give you somethingthat's a bit more like a teacher.
So will reply in that kind of way.
But what you can also do with that isyou can train it on your own style.
So you can upload a document, itwill scan it and have a look at the
way that you write, and then it willname that style, it'll absorb that
(04:21):
style, and then that's also thenbecomes one of the dropdowns as well.
So if you wanna train it on somethingto sound a bit like you, you can just
then go to that drop down and it reply.
It's not perfect.
It's not ever gonnasound exactly like you.
But it just gets out of the default tone.
And I re I just think that's reallyimportant with writing, considering
more and more people are now usingthese tools for their writing.
It's just like, how do we notsound like chat GPT basically.
(04:45):
So I like the fact that I cantrain it on different styles.
So I've, I've trained one for my sortof newsletter, so where I'm very much
me and I just sound exactly like I am.
And then I've got one whereI'm a bit more professional.
So I've trained on different ones,so I can just literally click and go.
I want you to reply like this.
So, yeah, things like that Ijust think are really helpful.
Daniel Rowles (05:03):
And, and that default writing style you've said is probably
better than the kind of standard chatGBT for writing as well, do you think?
Kerry Harrison (05:08):
Yeah, I, it's interesting because I've gone through
this real journey with Claude.
So there's been two model releases.
Literally in all my courses, I waslike, if you're a copywriter and you're
writing B2C copy, always go to Claude.
It's just really, really good.
And pretty much everyone I taught,PR agencies, copywriters, normal
agencies, they would just be like,oh yeah, now I've seen Claude.
This is amazing.
(05:30):
And then they did a an upgrade to3.7, so they've just upgraded to four.
And when they went into 3.7.
I started playing with it and Iwas really disappointed in it.
I think it was the first time whereI'd explored a new model and it had not
been better than it was before, justfor copywriting for everything else.
I know coders had said like, oh mygosh, the New York grade is amazing.
(05:51):
And then for me as a copywriter,I was like, oh no, it's just, no,
it's just not as good as it was.
So I was actually dropping downto the former model, and then I
reverted back to chat GPT, and Iwas talking to fellow copywriters.
Who were doing the same 'causethey were like, oh, we are really
disappointed with the Claude one as well.
Have you found the same?
I'm like, yeah.
So I felt a bit sad 'cause Ijust really like Claude for all
the reasons we talked about.
(06:11):
And then I was like, oh no.
But then they've just upgraded to four, sothey've upgraded the models again and then
it's gone back in and I've kind of got,yeah, I've like, I'm back into CLA again.
I really like Claude again.
So I was really glad that.
Daniel Rowles (06:22):
I remember we had this scheduled in and you kind of messed me.
We went, do you mind if wepush this back a little bit?
Because I'm a little bitconcerned about the new model.
And it's really interestingbecause the different versions
of chatt T can be the same and
Kerry Harrison (06:31):
Yeah.
Daniel Rowles (06:32):
and it will give you a different, slightly
different tone of voice becausethe way the model's been trained.
So it's a
One to pick out that if you'rewriting, you need to know the
model and the response you get.
And don't expect it tobe the same every time.
I
Kerry Harrison (06:42):
yeah.
Yeah.
I think that's true and that was thefirst time where I'd actually was
disappointed because I think with chatGPTI often found chat GPT was a little
bit stiff and actually I think theirnew upgrades as they've gone along, it's
actually got better for B2C copywriting.
So, you know, there's probably, it's.
There was a point where I thought,actually now Chapter GP is actually
better than Claude, and now we'rekind of somewhere in between.
(07:04):
I just, I just quite like Claude's kindof approach and this kind of slower
approach to releasing their tools.
They, they just seem to be taking thetime and thinking things through and I
really like, like, I think another thingwith Claude that I quite like, where is, I
feel like chat GPT and I know there's beenquite a lot of talk about this lately.
That it's quite syco antic, youknow, it'll tell you that even your
like, really quite mediocre ideasare genius and things like this,
(07:26):
it will, it loves to flatter you.
I feel like Claude's not quite like that.
It's a bit, it doesn'tsort of flatter you.
'cause sometimes it's like, it's lookslike occasionally you'll be like,
yeah, I know my ideas are pretty good.
Right?
Um, and then occasionally just like,do you know what JGBT just tell me
straight 'cause I don't want you tojust keep me nice to me all the time.
Um, I feel like Claude's a bit morerealistic and I quite like that.
Daniel Rowles (07:47):
A critical friend, which is good.
Kerry Harrison (07:48):
Yeah.
Daniel Rowles (07:49):
So they've upgraded the model recently,
Kerry Harrison (07:50):
Yeah.
Daniel Rowles (07:51):
you think about it?
And I think there's like a defaultand a reasoning model, isn't it?
Kerry Harrison (07:54):
Yeah.
Yeah, that's right.
So the new model you have now, ifyou're in the, if you pay, if you're
free, if you're in the fee version,and, and really do use the free version
if you've never tried call before.
'cause it, it, you know, it'sa great place to start if
you are in the paid model.
You also get the reasoning option as well.
So you've got four Sonet, which you getfor free, but you just don't get it.
(08:18):
You just have a limited use of it.
And then when you're on a page, you canobviously use it for much longer, but
you also get four Opus, which is likethe most advanced model that they've
got, which is quite high powered.
It's really good forthinking things through.
Critical thinking and, and you tendto think with chapter pt, with our
reasoning models, they tend to bemore suited to things like stem and
(08:39):
maths and coding and things like that.
I feel like with the Claude one,it's also pretty good for content
creation, for things like strategy.
So I do use it for, if I'mthinking more strategically.
And more recently I've been lookingat my own business and my positioning
and how I might move forward.
And actually, Claude's been a really good.
Thought partner for workingthrough those kind of things
(09:01):
as well with this larger model.
Just it's, it, it's just much more,yeah, it's kind of critical thinking.
It's reasoning, it's, it movesthings through step by step.
So I found that really power,a really powerful model.
Daniel Rowles (09:14):
And there, there's a lot of debate about AI content and
chat creating stuff that's quite,you know, generic sometimes as well.
How can you use Claw to makesure that that copywriting is
kind of authentic, compelling?
It's gonna be original as well.
I.
Kerry Harrison (09:26):
Yeah, it's, it's a good question.
I think there is a danger with chat GPTbecause it's the default model for a lot
of people, and I, I think we're, we'refinding now that because so many people
are using it for copywriting or content.
Actually people are getting pretty savvyand they know that it's generated with
our GPTI, I think you don't have to useit for that long to, to recognize it.
(09:47):
The certain phrases that it picks up,like every time I set create any kind
of, uh, blog or if I ask it to thinkabout anything that's related to AI
and and writing or AI content or AImarketing, it always starts the line.
Something like, in this fast baseddigital world or something like
that, like every single time.
Daniel Rowles (10:05):
is the one I've seen a lot.
Yeah.
Kerry Harrison (10:06):
And you see it a lot.
And then I see it in blogs and emailsand stuff like this that it's like
you've, you've generated that with GPT.
But I was talking to and someone theother day, she's a recruitment consultant
and she was saying literally nine overlike 90% of the covering letters she's
getting at is are made with G pt.
And she can just tell straightaway now, and she's just kind
of working out, you know.
(10:27):
Maybe I need to look at ones thataren't generated by charity, GBT.
'cause I've actually made the effort todo it, which is, which is interesting.
And so I, I think with chatGBT, and I think it's the same
with all the models, really.
We have to be coming out of thedefault tone as a de a default.
You know, we have to make sure that we'renot just sticking with what's there.
I suppose with Claude, just becauseit's a bit more of an underdog, I
feel like less people are using it.
(10:48):
So you might be able to get awaywith just going with the default
tone for a little bit longer.
People pull before people go like, oh,it's, it's, she's doing it with Claude.
Which I don't think youwould necessarily pick out.
So there's that.
But I think also just the fact that it'sgot that little abuse style thing where
you can train it on different styles.
But I think a lot of authentic, creatingauthentic content with AI comes down
(11:10):
to also bringing in the human element.
And that's regardless of whether you'reusing Claude or Chacha, bt, or Gemini, it,
the same rules apply really, you have to.
Put in the work as in you have to use,you know, we've got amazing brains.
We should keep using them.
One of the key things that I worry aboutwith AI is this idea that of de-skilling,
you know, we'll just kind of go, oh, I'llget to do it, or I'll get cla to do it.
(11:32):
And, and we just, I, I reallyfear that we'll just stop.
Doing our thinking, our critical thinking.
So, I have this methodologythat I talk about in all of my
presentations at conferences and stuff.
Uh, it is called the AI sandwich, whichI know is not the most sophisticated
name for a methodology, but it, it worksreally well and it people seem to remember
it and go, oh, I love your sandwich.
(11:53):
So for me to get the best writingout of AI models, and I guess it,
it expands wider than writing is.
It's a hu a human AI human process.
It's a three part process.
I. And it started with this idea,'cause I wrote this creative
methodology in, in January, 2020.
Where are we now?
2024. So I wrote it.
(12:14):
So back end of 2023.
I was thinking actually, I think thebest way to get great writing out of
something like charge BT or code is touse AI to generate a first draft and
then for me as a human, to go in andedit it and craft it with my own skills.
And then what I realized towards theend of 2023 was actually the prompting.
(12:34):
Prob is equally, if not moreimportant than what we do afterwards.
So that's where things reallychanged me back at, back at 2023.
So almost like a year into chapter GBT.
It's like, hang on a minute.
The prompting is so blooming important.
So I thought actually we need to dowork before we even come to the tool.
We need to have thought about,want to come to tool with.
At the very least an objective.
(12:56):
'cause a lot of people now justgo, oh, I'll just go to chat GPT.
But they haven't really thought throughwhat they wanna create or why, or how
that links to their wider marketingstrategy or their content strategy.
You know, just kind of go,oh, I'll just do it with that.
So for me, I think we need tocome with an objective, but also
to think about other things.
So how does it link to our strategy?
What kind of what?
Of my own ideas might Iwanna incorporate into this.
(13:18):
You know, if you're writinga blog, what's the angle?
Tell the model what your angleis so it moves in that direction.
So doing all of that thinking,taking time on your prompts to
get as much detail as you can.
I tend to say in my, in my courses, thatif you think about it as an intern, so
if you are asking an intern to write copyfor you, if you're asking 'em to write
(13:39):
a blog or a social post, you are notjust gonna say to them, go write me a.
Blog post on, I dunno,workplace wellbeing.
You're gonna say, I need to writea blog on workplace wellbeing.
This is who it's going to, this is whereit's, it's gonna go in this publication,
or it's gonna go on our website.
It needs to be, you know,500 to 800 words long.
You probably would alsosay, I've got these ideas.
(14:00):
I'm thinking it probably needsto include this, this, and this.
So maybe start there.
Uh, have you got any questions?
And basically that's what you woulddo in your prompt, you know, that,
think about it as an inter, whatinformation would you give it?
Think about context, bit of background.
How long do you want it to be,where it's going, who's your target
audience or all of that stuff.
And so for me, putting, being reallydetailed in the prompts then guides
(14:23):
the model in the right direction.
So.
And I think what also really kind ofstill slightly intrigues me is the fact
that it takes me a lot longer to writea prompt that does for AI to generate.
So it will take me awhile to get my prompt.
Perfect.
And then what It will generate theresult in like a matter seconds.
It's like that's wild.
But I, yeah.
(14:43):
But I do think that itmakes a, a huge difference.
And also to bear in mind that a lotof people are coming to the models
and just giving them a one sentence.
Very unthought through prompt.
And so what you're gonna get is prettyvanilla, pretty generic, pretty,
oh, it's definitely been writtenby SHA G pt. Um, so I think doing
that e early thinking, so that's thesort of human bit of the sandwich.
(15:05):
And then in the middle, whichis the filling, is the ai.
So then you get the AI to do onceyou've done your thinking, get
the AI to go off and do its thing.
And it's brilliant and it's fastand it's quick and it will give you
loads of it, give you 20 ideas in afew seconds, give you overviews or
insights or data analysis, and thenyou come back to the human again.
And when it comes out and you'vegot your res, your response from.
(15:26):
Whatever model you've used, the firstthing we need to do is, is fact check it.
So as you know and anyone knows who'susing these tools, hallucinations
are a thing and we, it's not,it doesn't hallucinate all the
time, but it may have done.
And so we have to check it.
So fact checking everything, anykind of stats or facts that come
back, just make sure they're truebecause they sound so viable, but
you know, it might not even exist.
(15:48):
And then we also sense checking itis another thing that I do because
I just think that it's really easy.
It's so easy to get wowed by the,the AI thing, the fact that you
generated a blog in what, 20 seconds?
You kind of go, oh great, I've got a blog.
Brilliant.
And I also think on first read,sometimes you can go, oh, that's
(16:08):
great, but then on second readyou're like, oh, do you know what, it
hasn't quite got the depth I wanted.
Or maybe it's a littlebit repetitive here.
So.
I think sense check and just pushing pastthe kind of, wow, and actually looking
at what you've got, and again, thisis where we need to use our analytical
brains like am I happy with that?
Another question I ask myselfis, if I wrote this from scratch
(16:29):
or created this from scratch,would it look anything like this?
And if the answer's no.
Probably need to think, well, youknow, maybe need to go back in and
think about the problem because it,you know, it, the fact is we're,
we're quite lazy naturally, aren't we?
And if, if we've got a blogdone in 10 seconds, it's like,
brilliant, I'll go and walk the dog.
Um, actually you justneed to like analyze it.
Uh, and then the last thing is craft.
So I still feel like in this.
(16:51):
Point in time that we still need peoplethat know their stuff, people who have
taste, people who have experience.
And so for me, adding the human craft ofcopywriting, which I've done for 23 years,
you know, I've been writing for 23 years,I still think it makes a difference.
You know, I know what goodlooks like, so it it, it helps
me to analyze the copy and go.
(17:11):
I think that's good.
Or this, it needs more ofthis, or it needs less of that.
So yeah, I still say to copywriters,there's a lot of copywriters are like
really worried about their work, but I,I still think there's a real need for
people who have deep expertise becauseI also think, you know, there's a lot of
talk around AI at the moment with the,the deep research tools where it says,
oh, you can have a, a PhD research 24 7.
(17:32):
I always think it's a bit weird 'causeif it generates something at PhD level,
but I haven't got a PhD in that thing.
It's like how do I actuallyknow if there's any good or not?
So I still think we need to beapplying or critical thinking.
So that's the kind of sandwich.
So prompting thinking.
Briefing it like an intern, you know, ifyou are really putting in your thought
and actually thinking of your own ideas.
So rather than just relying on the modelto come up with ideas for you, what
(17:55):
would you like to see in your blog thatyou're just about to write or your social
posts that you're just about to create?
Then the AI does its thing and then you goin sense check, fact check ad your craft.
So that's how you get more authentic.
Write writing.
I mean, there's a load more tothe prompting and stuff, but
I basically, if you wanna makesomething sound really beautiful,
you need to add a little bit of you.
Daniel Rowles (18:13):
Brilliant.
I love that.
I mean, I keep coming back to thisphrase of like, lean into your humanity.
When everything's AI andeverything's deep fake, that
becomes really, really important.
What about the pro, theproject feature in Claude?
How can you use that inyour kinda working process?
Kerry Harrison (18:24):
Yeah, so project, the projects is really good.
The projects is a pretty much anequivalent to the custom GP and gpt.
Daniel Rowles (18:31):
Right.
Kerry Harrison (18:32):
For me, that is a reason for me to pay for Claude.
And now also because they've got the, youknow, this more advanced reasoning model.
So projects allows you to create alittle area, or what they call a project.
You give it specific instructions,you give it knowledge.
And then it, you go into that spacewhenever you want to do something.
(18:53):
So for example, I've got quite a fewdifferent projects where I go in.
So I've got ones that do very simplethings, like I've got one called Ned
who's a newsletter sniper, basically.
So I have a newsletter,I write my newsletter.
I. And occasionally I'll make them myarticles in my newsletter a little bit
too long, and I want it to be snipped.
So rather than me typing out a very longprompt every time, which basically says, I
(19:17):
want you to, I want you to change this toa different word count, but I don't need
to take out my jokes, all my content, allmy, you know, I, I've got this whole like
long spiel of what I don't want it to do.
So instead of putting that intoa prompt every single time, I
just wanna go and snip a bit.
My newsletter.
I've created a project, so justas I would with a custom GPT.
So I've given it that there's details.
So when I put something, give yousomething, can you, you know, edit it
(19:39):
down, ask me what the word count it needsto be, and then I want you to, you know,
keep as close to the content as possible.
Don't take out my jokes.
Not that they're any good in thefirst place, but you know what I mean.
Stay as exactly on tone, but just cut it.
That's all I want you to do.
And so it's just really easy now 'causeI just literally drop in an article
that I've written and it will just tellme, ask me what co workout has done.
(20:00):
So it's just really, really easy.
So I've got, I've got ones like that.
I've got projects that are, I'vegot like my customer personas.
So one of my customerpersonas is founders.
So founders of businesses.
And I've given it this really,really detailed persona.
So marketing persona of a of a founder.
And then sometimes I'll go in andI'll just wanna say, you know,
(20:22):
what do you think of this idea?
Or if you know, as a founder,what, how do you feel about this?
Or I might take in my copy and say alanding page that I've just written
or a new sales page for a coursethat I'm creating and say what
resonates about this and what doesn't?
And so things like, that's quite nice.
So.
This idea of, I really like theidea of creating this kind of
interactive persona because.
(20:43):
Generally as a, as a marketer, as a, awriter or any kind of creative, you get
given these personas and they're basicallyjust a made up person on a piece of paper.
And I, I love the fact that now you canget something, create a really detailed
persona and put it in, and then you,you can actually interact with it.
And I know it's not the same astalking to an actual customer.
Then in reality, the onesyou get on a piece of paper
aren't a real customer either.
(21:03):
It's, you know, astrategist has made that.
And so I put it in.
So I've got, I've got aproject that does that.
Again, it just means I don't have to sayto it, you are this role, this is what
you believe, this is who you are, thisis your, what demographic you're in.
It's just all there in the project.
So I can just say, whatdo you think of this?
So really, really quick I'm tryingto think what else I've got in there.
I've got ones that helped me writeLinkedIn post, so I've given it really,
really detailed granular instructions onhow to write a blog style LinkedIn post.
(21:28):
Just go in with what my current insightis and it will help me write one.
Um, I've also got a tone of voiceone, so slightly more detailed than
the, the one that you get if youjust go down off the prompt box.
But, you know, really detailed viewof how I work, my tone of voice and
what it sounds like and how I writeeverything from the syntax to the the
sentence lengths and all the rest of it.
(21:48):
So it's all in there.
And again.
If I've written something that's not,maybe not exactly as I wanted it to
sound, it's like, I want it to soundlike my normal, like newsletter voice,
then I can just drop it in there.
So, and the same, I suppose ifyou can create tone voices for
all different things, differentbrands, different businesses.
I've got one that's reallybold and irreverent.
Sometimes I'm just like, ah, Ijust wanna create something that's
(22:09):
a little, feels a little bit.
A bit more cheeky today.
So I'll lit. I've literally trainedon this, like really irreverent tone
of voice and I just go in and justsay, make, just make this, give this
some life, and then I'll just bringit and bang it in and off it goes.
So those are just some examples.
There's loads where there's, I've alsogot business mentor as well, where
I've given it loads of details aroundmy business, my belief systems, the
(22:29):
things that I love, my, my methodologiesand my ethos around ai, which is.
Hopefully as you gather, very mucha kind of human and AI combination.
So I've given it all thatinformation about me.
So then if I'm working through somethinglike something in my business, I can just
go to that project because it alreadyknows all the information about it.
I don't have to give ita massively long prompt.
(22:49):
So.
Basically anywhere that youwould use a custom GPT, you can,
you can use called projects.
And to be fair, I've got the,they, I've got ones in chatt, ones
in and Claude and yeah, I've gotsome that are exactly the same.
I've just taken instructions fromone and brought it over to the other.
So it, and it's notdifficult to make, so, yeah.
Daniel Rowles (23:07):
was just gonna say, some of the ideas you were coming up with
there are just absolutely brilliantand actually apply to all of the AI
systems as well that you could use.
So I think there's somebrilliant ideas in there.
So, and then this kind of approach toprompt engineering, how do you do that
with different types of marketing content?
Do you approach it in adifferent kind of way?
Kerry Harrison (23:22):
Yeah, I, I guess so.
I think for me, I've talked a lotabout the AI sandwich and a lot
around what goes into a prompt.
So I think a lot of that still stands.
It, it's just being as,as detailed as you can.
I think a couple of other thingsthat we do in terms of prompting is.
To give the model examples, and Ithink that doesn't matter whether it's,
(23:43):
it's called or charity deal, any modelthat you are using, if you show it
what good looks like, I really thinkthat makes a big difference in terms
of guiding in the right direction.
So recently, for example, I wasworking on some case studies.
My case studies follow apretty similar structure, so
it's like, what's the problem?
What was the solution andwhat were the results?
So a three-part process, essentially, Igave it two examples of case studies that
(24:07):
had already written and then a load ofbullet points about the new one that I
was just, that I was just about to write.
Um, and then gave, and then ask itto do, to have a look, to use the two
examples as a guide to tone and style.
And then I was, I was actually really.
Really surprised at what it generatedas in like it stayed really.
'cause obviously AI lovesa pattern, doesn't it?
It's the whole thing is, you know,statistical prediction models.
(24:28):
So it loves a pattern and I, and itreally stuck to that pattern very well.
So it get, it made the new case study forme, it, it structured it in the problems.
It solution results.
It get like headlines and subs.
I was a bit like, wow, thisis, this is pretty good.
So I think giving it examplesmakes a, a really big difference.
If you're, if you're writingblogs, give it a couple examples
of blogs you've written.
(24:49):
If you're writing headlines, ratherthan just trusting it to do some great
headlines, get, show it examples ofone's headlines that you've done that
are really high performing or that haveworked really, really well for you.
So it, again, it has an understandingof your, your tone and your style.
So I think examplesmakes a big difference.
I guess the only thing, justto be aware of examples is.
Yes, it makes a massive differenceputting examples, but just, especially
(25:11):
if you're on the free models, just becareful what you put into the models.
I'm always very much like, don'tput anything private, confidential,
proprietary, into the free models.
So just be aware of that.
And also if you're writing press releasesand you wanna give it some examples,
just make sure you press releases onand bargained and stuff like that.
Write writing reports.
Just make sure that you're happywith, with what you're putting in.
Nothing confidential or private, but.
(25:32):
Generally speaking, if you,if it's, if you're okay to
put it in, put it, put it in.
And I think that, that make,that will make a significant
difference to the output.
Daniel Rowles (25:40):
And then what about going beyond copy, um,
and kind of content creation?
What, what else is, is Claudeparticularly good at, do you think?
Kerry Harrison (25:48):
Yeah.
I think I mentioned earlier, but I, I useit for a lot of marketing side of things.
As I mentioned the, thecustomer personas thing.
I really like that, and I can askyou about copy, but I could also
talk to it generally about anykind of strategy or, or marketing.
, as I was talking about positioning,so, so any kind of like strategy,
content, strategy, so not justactually the writing, but it helping
(26:11):
to inform your writing as well.
So it might be at looking at customers,looking at, their pain points.
Looking at I've got this prompt where Ilook at a customer's fears, frustration,
streams and desires, and, you know,create this whole scenario of them and
what they, what they might think abouta product or a service, which can then
help me to again, look at those fearsand frustrations, streams and desires
(26:32):
and go, okay, so maybe I could createsome content around that or around that.
So I think strategic stuff isreally good for my own business.
I've been doing, working a lot onpositioning, as I mentioned before.
So it's been, I've been working at.
Through that, I think it makesa really good thought partner.
So not necessarily giving me the answers,and sometimes I think it's quite useful to
(26:55):
not ask for answers, but to ask for ideas.
So.
Just as a different way of prompting.
'cause I think we often go tothe models and look for, look for
answers, but actually sometimes justgetting ideas back is really helpful.
So I like the fact that I, it mightcome back with something and I'll be
like, oh, that's really interesting.
And it might take my thinking offin, in a new direction or what I've
done with positioning because I've,I've had quite long conversations
(27:17):
with court about my positioning.
That sometimes it'll bring somethingback or an an idea back to me and then
I'll look at it and go, it is definitelynot that I don't want to do that.
And it's made me question my, oh,it's been really good at just helping
me even tap into my own intuition.
'cause it'll present something back tome or a target audience and I'd be like
no, that I, I wouldn't wanna do that.
I don't wanna target these peoplethat would not make me feel very happy
(27:39):
or wouldn't make me feel fulfilled.
So I, I've really lovedthat side of it as well.
'cause if it just gave me an answer.
I don't think I would necessarilygo through those thought presses
and, and think, what do I actuallythink about this right now?
Or like, am I happy to do that?
Would that make, you know,would that fill me up or not?
So I've, I've used it for that.
I think brainstorming, it's reallygood for not just for copy, but for
(28:02):
content, just for life, to be honest.
Just anything where I've got ideas.
Anytime that I just feel like there was apoint in time where, you know, often it'd
be like, oh, I just don't have any ideas,or, I'm really, really tired and it's
five o'clock and I just need some ideas.
And I, you know, the fact that I can getideas within a matter, you know, 10 ideas,
20 ideas in a matter of seconds, andthey're not gonna be perfect by any means.
(28:23):
And they're, if you're doinganything highly conceptual.
As much as I want AI to help me to dothat, I don't think it, it can do it.
You know, it's things like strap lines.
It's just really poor.
It just generates aload of generic outputs.
And so I sometimes start with it,but I always have to use my own
brain for those kind of things.
But for, for general sort of ideastarters, I think it's really helpful and.
(28:44):
Normally I'm just, it justgets over the blank page thing.
So I'm like, I don't panic anymore.
It's like, okay, wellI've got something down.
That's the start.
And then I, and then I'll be like,okay, so that's quite interesting.
Or I quite like that angle.
I hadn't really thought about that before.
So it might take mythinking in a new direction.
It might be that it comes backwith two ideas if you blend
them together with pretty well.
(29:04):
Or, so I just I like it forbrainstorming, for getting me
started, but yeah, hopefully that's.
Daniel Rowles (29:10):
Yeah, definitely.
And I, I, one of the things I found withit is that, and I do this at chatt pt,
I'll get it to come up with five differentideas and then compare and contrast them.
And I'm not necessarily gonna useone of those ideas, but there's the
thought process that it goes through.
Kerry Harrison (29:22):
Yes.
Daniel Rowles (29:23):
its conclusions is what's gonna spur my thinking with this as
Kerry Harrison (29:25):
Yeah.
Yeah.
I think that's, that's really true.
And actually with Claude, you'vealso got extended thinking.
So extended thinking is anotherlittle toggle that comes off.
Daniel Rowles (29:33):
research mode,
Kerry Harrison (29:34):
yeah, it's, it's a bit like deep research, but you
just literally toggle it on andit will show you its thinking.
So it's a little bit slower.
Daniel Rowles (29:41):
Fantastic.
Okay.
saw that in deep seekand that was one of the
Kerry Harrison (29:44):
yes.
Daniel Rowles (29:45):
Deep
Kerry Harrison (29:45):
Yeah.
Daniel Rowles (29:46):
reasoning
Kerry Harrison (29:46):
Yeah, so I, I really like that if you just, if you're just
wanting to work through something,you want it to slow down a little
bit because obviously if you use thedefault models, it, it's pretty quick.
If you want it to slow down or youwanna think slightly wider or get it to
do more, consider like considerationsand come back with multiple
considerations or slightly differentcreative angles, things like that.
If you take click on extendedthinking it will then.
(30:10):
It will just give you more andyou'll see its thought processes
and it's like sometimes you'relike, oh, that's really interesting.
'cause I wouldn't have necessarilygone down that route myself.
But if I do go down that route, Iwonder where that would take me.
So I, I do, yeah, I do think as athought partner, it's really useful.
I,
Daniel Rowles (30:24):
then what about things like data and analytics?
Kerry Harrison (30:26):
So it's great for data analytics and I think,
you know, not, I suppose.
Comparable with chart GPT, andit's also really good visuals.
So if you ask it to do data analytics,it will also generate graphs for you or
diagrams for you and things like that.
So in terms of marketing, if you'reanalyzing campaigns or how your
social media's going or thingslike that, it's really useful.
(30:47):
You can, you'll get that results and youcan make them pretty, which is not, and
choose your colors and all the rest of it.
Which, which is great forany kind of reporting.
So I use it for that.
I've also turned on, recently, they'vegot now introduced in the last month or
so, integrations, so you can now integratewith 'cause they've be, they've had
investment from Google um, anthropic.
(31:08):
So they had a lot ofinvestment from Google.
They also had huge amountsof investment from Amazon.
So I think it was
6 billion of investment from Amazon.
So I think they've got these links withGoogle, so you can link it up literally
with really easily, just literally clickthrough, link it through to your calendar.
Just check if you're gonna link it throughto your calendar and and Gmail just.
Check the privacy policy andmake sure you're happy with that.
'cause I know, you know,have to be careful.
(31:29):
It's like I haven't yet linked it upto my, uh, Gmail, my, like main Gmail
because it's also my personal Gmailand it goes back all, you know, so far
now there's probably loads of personalstuff out about all sorts of things.
I'm like, I'm not sure if I wanna giveit that, but I've given it my business.
My business account.
So that's been really useful.
So a couple of things I've donewith that in the last week or so
is to get it to tell me what aremy main priorities for this week.
(31:53):
So I'll have a look at my calendar.
It will look at my, gmail, andit'll say, you need to do these.
And it, you know, sometimes it'll say Youreally need to get back to this person.
Because this seems like it'sgonna be really important for you.
So, so I like that.
So it'll prioritize.
Or even my day.
So today I just said to it, whatare my priorities for today?
So obviously I knew it got thepodcast, so that was on there.
(32:14):
So that's, that's a good one.
It can analyze that.
Um, I've also asked it to analyze mytime, like how am I spending my time?
And that's really interesting froma data analytics point of view.
And, and, and it even came back'cause then I was saying, okay,
so now you've analyzed my time.
Tell me how I can be more efficient.
Really interestingly, it said somethinglike, you'll have, you are doing too
(32:36):
many 20 minute introductory calls thataren't necessarily leading to full work.
So how wide, how about creating a,you know, a questionnaire so that
when people come through to you fora call, you've already vetted them.
And I was like.
God, that's, you know, andobviously I would've thought,
could've thought of that.
In fact, my friend suggested to methe other day, but I just thought it's
(32:56):
amazing that Claude has just been ableto look at my, the way I spend my time.
It said to me, you probablyneed to do a bit more exercise.
So it's like, I'm not, you know, oftenI'd have like yoga classes in definitely
the last few weeks I've been so busy.
There's not been any yogaclasses in my, my calendar.
So it's just, it's pretty incredible whatit can do and just the different angles
that it can look at your time from,you know, not 'cause I asked it how I
(33:19):
can be more efficient, but I could askit, how could I improve my wellbeing?
And it may well say, actually youneed to put 10% more time into.
Going to yoga or going runningor whatever it might be.
So that was pretty life changing.
So I, I found that really good.
So yeah, how I spend my time.
And then another thing, I tried it for theother day, I get so many AI newsletters
(33:39):
and I love reading them, but the lastfew weeks I've been so busy and I just
said to me can it, can you scan my AInewsletters that I've received in my
inbox and give me a summary of each one.
So it just said, youknow, this is this one
Daniel Rowles (33:52):
that.
That's
Kerry Harrison (33:52):
that, yeah, so it just meant that because I haven't had time,
it's just summarized the newslettersthat I've had through in the way that
it thinks is most relevant to me.
And then what I would do then, becauseI've been so strapped for time, I
probably, now I've got this summaries,I'm gonna go, actually, do you know what?
I'm gonna read that one inthat one from top to bottom.
'cause I think that's gonnabe super useful for me.
(34:13):
So I would feel like.
Those integrations, which havegone in very recently could be
so useful for work, for planningfor your own personal wellbeing.
So many different things from a businesspoint of view that it can guide you on.
So that's, yeah, for me, that'sbeen a real game trip flight.
A total game changer, actuallythe last couple of weeks.
Daniel Rowles (34:32):
interesting 'cause chat, DBT are going down that route as
Kerry Harrison (34:34):
Yeah.
Daniel Rowles (34:34):
it's
Kerry Harrison (34:34):
Yeah.
Daniel Rowles (34:35):
see which tools that are gonna integrate into
what and how that changes things.
So, look, really great introductionto Claude but actually loads of
ideas in there that apply to ai.
Full stop.
So
Kerry Harrison (34:43):
Yes.
Daniel Rowles (34:44):
much for those Kerry and, and the time you've given us.
Tell people a little bit aboutwhat you do on a day by day, the
services you offer and how peoplecan get hold of you if they want to.
Kerry Harrison (34:51):
So I spend most of my time training people in how
to use AI for their marketing.
So everything fromcontent to data analytics.
Sometimes I do very specificcopywriting, you know, all day
copywriting for copywriting teams.
So it depends what you need.
Uh, I also speak at conferences and um.
Yeah, sit on panels,all that kind of thing.
I've got a newsletter, it'scalled Minds and Machines.
(35:12):
It's on Substack.
, I go in there and I give tips andkeep you up today with the latest news
and normally share some of the thingsthat I've learned and I'm exploring.
I'm on LinkedIn, Kerry Harrison.
So you'll find me, uh, I talk a loton LinkedIn, not just about AI and
all the incredible things it cando, but very much about the human.
How we can keep thinking, how we canstay creative, how we can nurture
(35:33):
everything that makes us human.
I just feel like, as you saidearlier, I feel like that's
gonna be incredibly important.
And outside of the ai.
I'm also a yoga teacher and Irun women's circles in the woods.
So this next week I'm actually runninga retreat down in the Woodlands,
near Salisbury with a, um, reallysuper experienced a retreat host.
So off, pretty much off grid, workingwith analog creative practices, writing
(35:59):
with pen and paper sitting around a fire.
, So I do all of those things too.
And so somehow my whole life isall just trying to smash these
two worlds together of high tech.
But also just what it means to justsit on a log around, a fire under
the stars and to just go, right,how can we bring these two together?
'cause for me, they feel like they'retwo sides to the same coin almost.
Daniel Rowles (36:19):
Brilliant.
Well, we'll put all of thatinto the show notes, thank you
Kerry, so much for your time.
Some brilliant tips there and I'm pleasedo let us know on what you think and
your takes on all of this as well.
Kerry Harrison (36:26):
Cool.
Thanks.
For more episodes, resources toleave a review or to get in contact,
go to target internet.com/podcast.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
My Favorite Murder with Karen Kilgariff and Georgia Hardstark
My Favorite Murder is a true crime comedy podcast hosted by Karen Kilgariff and Georgia Hardstark. Each week, Karen and Georgia share compelling true crimes and hometown stories from friends and listeners. Since MFM launched in January of 2016, Karen and Georgia have shared their lifelong interest in true crime and have covered stories of infamous serial killers like the Night Stalker, mysterious cold cases, captivating cults, incredible survivor stories and important events from history like the Tulsa race massacre of 1921. My Favorite Murder is part of the Exactly Right podcast network that provides a platform for bold, creative voices to bring to life provocative, entertaining and relatable stories for audiences everywhere. The Exactly Right roster of podcasts covers a variety of topics including historic true crime, comedic interviews and news, science, pop culture and more. Podcasts on the network include Buried Bones with Kate Winkler Dawson and Paul Holes, That's Messed Up: An SVU Podcast, This Podcast Will Kill You, Bananas and more.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.