AI transcript
I think most people don’t realize
like how much better perplexity is than Google.
– You can really, really rapidly do a lot of research
and get a lot of ideas super, super quick.
And to me, that’s really powerful,
but also I hope nobody goes and takes this idea
and runs with it.
– Tons of people will.
– Tons of people will.
(upbeat music)
– When all your marketing team does is put out fires,
they burn out.
But with HubSpot, they can achieve their best results
without the stress.
Tap into HubSpot’s collection of AI tools,
breeze, to pinpoint leads, capture attention,
and access all your data in one place.
Keep your marketers cool
and your campaign results hotter than ever.
Visit hubspot.com/marketers to learn more.
(upbeat music)
– Hey, welcome to the Next Wave Podcast.
I’m Matt Wolf.
I’m here with Nathan Lanz.
And, you know, everybody’s talking about AI.
AI is everywhere.
It’s all over the news.
Every big company is putting AI into everything,
but we keep hearing the same question.
What the heck do I actually use this for?
How is this benefiting me?
How is this gonna make my company better
or my daily life better?
Well, that’s what we’re gonna talk about
in this episode today.
Nathan and I, we’re going to share with you
some of the ways that we’re actually using AI
in our own businesses.
So that’s the goal of this episode,
is to break down the ways
that maybe you’re not even thinking about
that AI could really, really benefit your life.
I think there could be some overlap here
between what you’re gonna say and what I’m gonna say,
but there’ll be some slight nuances
to the way I do it versus the way you do it.
You wanna kick it off?
Like what’s like the first actually actionable use case
for AI that we can talk about here?
– Yes, I saw this tweet from Balaji,
which I thought was really interesting
’cause for a long time I’ve been a huge fan of perplexity.
You know, I’ve started using it for a lot of the research
I do instead of doing a Google search.
I found that often I get better answers
on perplexity actually.
And I’ve started using it for the podcast research,
especially like for interviewing a guest.
You type in the guest’s name
and like the information you get from that
is so much better than just doing a Google search.
You get videos they’ve been in,
you can ask it, follow up questions,
but you know, like what kind of stuff
did they talk about on recent podcast or whatever?
It’s great.
– If people want to learn more about perplexity,
we actually had Arvind, the CEO
and founder of Perplexity on the show.
So make sure you check out that episode.
But what perplexity is,
is it’s essentially a large language model
like chat GPT or like Claude
or whatever large language model you’re used to using.
But it also searches the internet for you
whenever you ask it any sort of questions
and then uses whatever it finds
as part of the context for the response.
So you could ask it questions
just like you would ask chat GPT.
And when it’s formulating that question,
it’s always going to do like a search on the internet
to try to make sure it’s giving you the most
sort of informed response that it possibly can.
And once you start using it,
often it’s kind of hard to go back to regular Google.
– Right.
And so I saw this tweet from Balajas, I was genius,
like, yeah, you can actually just change it, right?
You can go into Google Chrome or whatever browser you use
and you can change your default to be perplexity.
And then it works.
– So Google made that like an option in the settings
where you can just…
It’s like, it’s slightly hidden.
It’s like, it’s not as easy to find.
– But it’s not like you don’t need any sort of extra
Chrome extension or anything like that.
You just go straight into the Chrome settings
and then do a search for it.
And then you can make that your primary search engine.
So now like whenever you type anything,
you do the URL bar at the top,
it’s gonna search for perplexity instead of Google, right?
– Yeah, for example,
if I type in the next wave podcast right here, right?
And so now it’s like pulling up all of our episodes.
It’s giving a synopsis of what the podcast is about.
It’s even like sharing some of the feedback
from listeners, which is wild, related topics.
And yeah, and then you can even follow up questions, right?
You could ask about a certain episode,
what happened on this episode
or what were the key takeaways and it has all that.
– So I’m curious, what are some of the ways
that you’re using perplexity that have, I guess,
made life easier, made business easier,
made daily productivity easier?
Like what are you actually using perplexity before?
– I mean, mainly like researching companies,
research for the podcast, but also for meetings.
Like if I’m meeting somebody and I don’t know much about them,
I go for, I would look at LinkedIn,
but it’s like kind of, you know, so much information.
And a lot of it’s really kind of…
– Well, and perplexity is actually gonna pull in stuff
from LinkedIn a lot of times as well.
– Right, right.
And then you can ask, you can ask follow up questions.
It was just way better ’cause you wanna actually talk to it
and have like a thing where like, oh, where did they go to,
you know, what company did they work at?
What were they actually responsible for?
I just liked that experience better
than looking at someone’s LinkedIn.
And also if they were on videos or whatever,
you get so much more context about a person for a meeting.
So I think that’s like the main use case.
I’m sure there’s tons of others
for people who are doing other kinds of research,
maybe market research, other things
that perplexity is great for.
But like I said, I had a really hard time
developing the habit.
I’m just so used to going to Google or LinkedIn.
And I’m finding that this is definitely helping me
to develop the habit of like, oh,
just use perplexity for that stuff.
– For sure, for sure.
Well, I’ll share my first one ’cause it’s super related.
It’s related to perplexity as well.
I was gonna talk about how I actually do guest research
for the podcast using perplexity.
So basically I plugged in myself
as if I was interviewing myself today
just ’cause I was curious what it would bring up.
But this is the type of thing I’ll do
before every single podcast.
I’ll give it a prompt like, I’m interviewing Matt Wolf,
the YouTuber and creator of Future Tools
on my podcast today.
What should I ask him to ensure
an engaging educational and entertaining episode?
So that’s the prompt that I like to give.
I always like to give extra context
just in case like somebody else has that same name.
Like there’s a professional golfer named Matt Wolf.
And so if I just say like, what should I ask Matt Wolf?
It’ll be like, how do I drive the ball farther
or whatever, right?
So I wanted to make sure I gave that additional context
of the YouTuber and the creator of Future Tools
in my prompt.
So I know it’s gonna pull up the right Matt Wolf.
But then I also don’t want it to just ask
like boring questions about the company, right?
Like let’s say I plugged in somebody
that works at HubSpot or something like that.
I don’t want it to give me questions like,
what are HubSpot’s new initiatives for 2025?
You know, like I don’t want it to get
like boring corporate questions.
I want them to be engaging, educational and entertaining.
So very, very specific in the way I prompt it.
But then like you can see the steps
that perplexity went through.
Research background information about Matt Wolf
and his YouTube channel.
Identify key topics and areas of expertise for Matt Wolf
that would make for an engaging and educational podcast.
Brainstorm a list of potential interview questions
that would cover the key topics
and provide an entertaining and informative episode.
So you can actually see the way perplexity
is sort of thinking through this.
It’s almost borderline agentic, right?
Where it does this one search and it goes,
okay, we’ve got this information now.
Now use this search and it pulls up more information.
And you can see here up at the top,
Matt Wolf creator of Future Tools,
current job founder of Future Tools,
YouTube host and podcaster where I live.
My education apparently pulled all this in from LinkedIn
and has like a little bio of me right there.
Over on the right sidebar,
you can see a whole bunch of content
that I’ve been involved in with the ability to watch it.
And then it brings up some questions.
You know, what sparked your interest in AI?
Can you describe your journey
from being an entrepreneur to becoming a YouTuber?
How do you stay organized and manage your time effectively?
And it broke this all down into sections.
And I’m telling you, like sometimes,
maybe we shouldn’t put this out in the world,
but sometimes we’ll pull on a guest onto our podcast.
And we’ve had such a busy week
that we didn’t have all of the time in the world
to research the guest.
Well, this right here just made it infinitely easier.
So I wanted to share this
because it’s very related to what you were talking about.
But also like, if you’re going and interviewing
for a job somewhere, plug in the company here,
plug in the person’s name that’s interviewing you here,
and learn more about that person.
So you’re going into this interview like dialed in
and ready to have the conversation
with the person that you’re talking to, right?
If you’re about to just jump on a call
or a pitch meeting or something like that,
you can really, really rapidly do a lot of research
and get a lot of ideas super, super quick.
– Yeah, I mean, I think most people don’t realize
like how much better perplexity is than Google.
And like, and it seems like Google,
the quality of Google seems to be going down every year.
And it’s like, most people just don’t realize it
’cause it’s like, it slowly has been happening, right?
– Yeah, yeah.
– But, and now we’re like, you know,
AI content being mass generated,
like the quality is going down more and more it seems.
And, you know, and Google’s tried to fight that
by relying more on authorities.
And now that’s why you see Reddit and Quora at the top,
but now people are using AI to mass spam Reddit and Quora.
So it’s just the quality of Google continues to go down.
It’s when you search for something,
you often don’t get the answer.
Unless it’s something very simple,
you often don’t quickly get the answer to, you know,
to your question, but with perplexity,
you get such great quality, you know, answers,
whatever you’re asking and all the extra information
with the videos and you can do follow-ups.
I highly recommend people try that.
– Yeah, and even like Google is trying
to do this same kind of thing now, right?
Where you do a Google search
and it has the AI response up at the top.
The problem with Google is it’s still to this day,
doesn’t know the difference between a meme and reality.
Right? Like all of the stuff that came out with Google,
like saying, hey, maybe you should try putting glue
on your cheese to make sure it sticks to your pizza.
And how many rocks should you eat per day?
And, you know, geologists recommend you eat
at least 17 rocks a day.
All of that stuff is because memes exist.
It said that random ass stuff and Google thought
that it was, you know, actually factual information
that it fed through its AI.
Perplexity doesn’t seem to have those same sorts of issues
because I think it’s sort of doing a little bit
more cross-referencing than what Google’s doing.
– Yeah, they started from scratch.
Like thinking how to build a search engine
or an answer engine up, you know, from first principles.
Whereas I think Google’s relying
on really antiquated technology.
They built a long time ago.
You know, you can see this in some of the recent stuff
with like people trying to search for Donald Trump
or the assassination attempt and it’s like,
“Oh, did you mean Kamala Harris?”
And it’s like, “What the hell?
What? What are you talking about?”
And I don’t think that was like somebody manually doing that.
That’s probably just based on like all the news sources
they’re pulling information from
and they’re putting certain authority
to those certain news sources.
That’s probably why that happened.
But that’s based on like antiquated technology
and that’s probably why all that’s happening.
So yeah, people should be using perplexity.
Okay, this next one, I saw it from a tweet
from Allie Miller, she calls the “clawed walk.”
And I’ve actually heard this from other people too,
like Dan Shipper, apparently this is one of his top use cases
for AI, is that when you’re thinking about something,
you know, instead of just sitting in your room
or in your office and just kind of, you know,
working on it that way, like actually get out
and get some exercise and do work at the same time, right?
And AI actually makes this actually feasible now.
So like, so what he does and what Allie’s suggesting
people do is go for a walk and then use something
like a super whisper or something like that
to transcribe everything that you’re saying
and then make it, we actually have notes,
which then you could, you know,
you could feed into chat to your Claude,
tell it to remember it or the, you know,
and make it actionable.
And I think Dan said he’s even using that
for his newsletter, I believe.
Like, that’s how he’s writing his newsletter.
And so I was like, I have to start doing that.
‘Cause like, I’ve been on like a big health kick,
especially, you know, I think in the age of AI
being healthy is really important.
And I was like, okay, so if I can get out for a walk
and whatever I’m thinking about from my newsletter,
just say it as I’m walking, right?
That would save so much time.
But also I just, I find that when I’m walking,
I’m more relaxed and I’m more able to think, you know,
it’s kind of different than jogging
where it’s hard to think when you’re walking.
You can think very clearly.
Sometimes even better than when I’m sitting down.
And so I’m trying to get in the habit of doing that.
Having kind of a hard time getting into it
because I feel like in Japan it feels slightly awkward to me.
Like the American guy walking around
where everything’s really quiet here in Kyoto
and I’m like walking around talking to myself outside
and people are staring.
– I feel like people are worried about that
less and less and less these days.
‘Cause it’s so common now to be, you know,
just have like air pods in and be like,
talking to somebody on the phone
while you’re walking around or something.
So I don’t know, I’ve always felt self-conscious
about that as well,
but I feel like it’s definitely getting more normalized.
– Yeah, so here’s Super Whisper.
People can check it out.
See, I’m actually playing on like,
installing this and trying it today.
I haven’t tried it yet.
(upbeat music)
We’ll be right back,
but first I wanna tell you about another great podcast
you’re gonna wanna listen to.
It’s called Science of Scaling, hosted by Mark Roberge.
And it’s brought to you by the HubSpot Podcast Network,
the audio destination for business professionals.
Each week hosts Mark Roberge,
founding chief revenue officer at HubSpot,
senior lecturer at Harvard Business School
and co-founder of Stage 2 Capital,
sits down with the most successful sales leaders in tech
to learn the secrets, strategies, and tactics
to scaling your company’s growth.
He recently did a great episode called,
“How Do You Solve for a Siloed Marketing and Sales?”
And I personally learned a lot from it.
You’re gonna wanna check out the podcast,
listen to Science of Scaling
wherever you get your podcasts.
(upbeat music)
– So what does Super Whisper do?
I’ve heard of it, but I’m not super familiar with it.
– So you can download the app.
I think it’s mainly on macOS, I believe,
and you download it and it just helps you transcribe
whatever you’re saying to it.
So it’s like you just talk to the app
and then it transcribes it for you.
– Okay, similar to Otter.
I use Otter for the same thing.
– Okay, yeah.
– You Otter know.
– How do you use Otter?
– So Otter’s the same idea.
It’s an app that’s on my iPhone that I open it up
and you just start talking into it
and it basically transcribes whatever you say in real time.
One of the ways I actually use it,
this is actually not something that was in my notes,
but I’ll share it anyway
’cause it’s top of mind,
is whenever I go to these conferences,
when I go to Google I/O,
when I go to Microsoft Build,
I’m going to MetaConnect next month,
whenever I go to these events,
I actually pull out my iPhone, open up Otter,
and then set it on my lap
and just let it transcribe the entire presentation
that somebody’s giving.
And then when it’s done,
Otter will give you the summary and bullet points
of like here’s the main takeaways
from the transcript that you just created.
And it will just give me bullets of like,
here’s the 10 things they just talked about
in this presentation.
So I can sit in the presentations
and just completely zone out if I want and not pay attention.
And I’ll just have my cliff notes
of the entire presentation sitting on my phone later.
It’s awesome.
– That’s great.
– But actually, you know what?
I’m going to share,
I’m going to jump to another idea
that I was going to share today
because it’s related to what you just said right there.
– Oh, okay, okay.
– Similar concept,
but one thing that I like to do
when trying to create like documents
like PDFs or any sort of written blog posts
or things like that is I like to speak them out
and then have them transcribed
and then use AI to sort of reword the transcription for me
into something that sounds more like a written article
versus just like reading a transcript, right?
So the most recent one that I did,
I’m sort of laughing.
You’ll understand why I’m laughing in a second.
The most recent one I did was HubSpot reached out to me
and asked if I can help them create a PDF
on all of my thoughts on like the AI world right now.
And they sent me this questionnaire
with like 14 questions on it.
And I looked at it and went,
“Man, I’m going to have to actually like sit down
and type out responses to this whole thing.
Oh, this is going to be a pain in the butt.”
So what I actually did instead
was I gave this questionnaire to Emily,
who’s my assistant.
I gave it to Emily, we jumped on a call
and I said, “Read these questions to me.
Just pretend like you’re interviewing me
for a podcast or something
and read these 12 questions to me.
And I’m going to record this
and I’m going to record my response
to every single question.”
So I recorded the entire conversation
as she read through these 12 different questions.
And I answered every single question.
And then I took that recording
and I pulled it into Descript.
So Descript, you can upload an audio or a video file.
It will transcribe the entire thing for you.
So I used Descript to get my transcription.
And then I took that entire transcription
and I pulled it into Claude.
Claude lets you upload really long text files.
And then I pulled the whole thing into Claude
and I went through and I had it actually answer
the questions again for me one by one,
but using the context of the document that I uploaded.
The idea was I uploaded the entire transcript
as like the context and I said use this transcript
to answer the questions that I’m about to ask you, right?
So then I went through the document again,
the 12 questions and I took question one,
copied and pasted it into Claude.
And then Claude responded to the question,
but based on the answers I gave in my transcript.
And I just went through every question one at a time
and had Claude write up a very succinct response
that had like my sentiment in it.
It had the ideas and thoughts and things
that I shared in my response,
but it did it in like a single paragraph per question.
And so it’s super helpful to write up documents
and articles and things like that.
And so how other people can apply this is,
let’s say you wanna write a blog post on SEO.
I don’t know, I don’t know what niche you’re in,
but let’s say you wanna write up a blog post
all about like the top 10 SEO tactics or something,
write out a list of questions
that you think people would want answered about SEO,
just write out all of those questions
and then record yourself
just answering those questions out loud.
And then you can pull it
into one of these transcription tools like Descript.
Let Descript transcribe the whole thing
and then take that entire transcript,
pull it into Claude and then have Claude
turn that into a written article or a PDF document
or something like that for you.
And it saves so much time
and it writes it probably way better
than you would have written it yourself, let’s be honest.
So that’s a really, really good process,
but it’s very similar to the idea of like going for walks
and just sort of transcribing your thoughts
as you’re walking.
– Yeah, that’s awesome.
That’s anything like that that takes it to an area
where like humans are living more of their lives
and like being healthier
while actually still getting their work done,
I think is awesome.
It actually kind of goes into my next one.
So my next one really is about using AI to like stay healthy.
So this is not for work,
but I do think it’s relevant.
If you’re healthy, you can work better.
And I think a lot of people have not really
connected the dots too that like in the age of AI,
people are probably going to live a lot longer.
And AI is most likely going to unlock
some like major advancements in the healthcare
that people might end up living 10 to 30 years longer,
maybe longer.
So you want to be healthy, right?
Like you don’t want to be like old,
like living a super long life,
but you’re very unhealthy.
Like no, you want to be somewhat healthy.
And so I’ve been trying like on a big health kick
over the last year.
I had gained a lot of weight during COVID
and it’s kind of like kept it on
and like, you know, lost five pounds here and there,
but still was quite overweight.
And then about a year ago,
I started going to the gym a lot like quite often.
And the first thing I realized was like,
you need to really track your calories and your protein.
And I had never done that in my life before.
I had mostly been like kind of slightly overweight
most of my life.
I was super fit when I was maybe like 18 or 19.
And so I realized, okay,
you got to track calories and protein.
And so I’ve been using this thing.
Actually, it’s a custom GPT that I created,
which is, it’s very simple.
Like it doesn’t do all the stuff I want it to do.
It’s like, you know,
I would like for it to do a lot more,
but the memory is quite limited on what,
you know, what they’ll actually remember.
And it’s real simple.
And I just have a thing here where I have like,
for example, I’ll do like a new day, you know,
I’ll just, I’ll like type this in.
I’m typically doing it on the mobile app.
I’m typically not doing it on my desktop.
But anyways, you type in like new day
and you’ll set like what are your calorie goals
for the day, what are your protein goals?
And just anything that I eat,
like if I know the calories and protein,
I will type that in.
And I can even, you can even do just shorthand
just literally just put the,
you don’t even have to tell it what’s calories and protein.
It’ll figure that out.
You can just put the two numbers and it’ll know that.
If you want, you can add a description to it.
So like, oh, I had a latte or I had oatmeal or whatever.
But also the really cool thing is,
you just kind of keep this same window open,
the same tab and like, you can go back to it every day.
And I’m sure at some point,
there’s some kind of memory limit there,
but I’ve been using it for like two months now,
the same one.
– Yeah.
– And then the cool thing is like it,
in that single context,
he remembers all the stuff you shared before.
So like, oh, I had this protein drink or whatever,
it’ll know what the calories and protein are for that.
So you don’t even have to type in the numbers anymore.
– So you’re basically using it to track the calories
and protein.
Is there any other like benefit that you get out of it?
Is it like giving you motivation?
Is it giving you like–
– I would like to do more stuff like that.
Yeah, yeah, I would like,
it’s literally just doing the calories and protein.
Like I wanted to do a lot more.
I was like, okay, I’m gonna make this like a really cool,
you know, custom GBT where like,
it’ll like, it’ll motivate you,
it’ll like track your progress.
Like, yeah, it’s currently not capable
of doing any of that.
I mean, I’m sure you could build an application
that used ChatGBT’s API.
I’m sure you could do that.
But in terms of a custom GBT,
it doesn’t seem to be possible at the moment.
– Well, you could use,
you could use like the rag method, right?
Retrieval Augmented Generation,
where like every time it gives you an update,
copy and paste it into a text file
and then upload that text file and then it will–
– Yeah, I’ve been too lazy to figure that out.
Yeah, I should at some point.
But yeah, I mean, for me it just,
it simplifies tracking calories and protein.
You know, I just type new day every time
when it’s a new day and it starts over.
I want to change anything.
Like, okay, right now I’m trying to diet.
Okay, I reduced my calories by 300 or something like that.
If I’m trying to gain more muscle,
I increase it by 300.
I am trying to give it to where it can actually
coach you more on that.
Like, you know, I tried to feed it like different documents
from like people who I really respect,
like different like health scientists
and things like that or exercise scientists.
And then it’s pretty useful.
Like you can talk to them about like,
okay, I want a bulk right now.
Like, what does that mean?
Like how many calories should I be taking in?
Or should I be doing differently with exercise?
So it will answer that kind of stuff.
– Yeah, yeah.
‘Cause what’ll be cool is like, you know,
you can go in there and be like,
oh, I’ve seen like I’ve plateaued.
I haven’t, you know, been able to add more weight
to my bench press or whatever in two months.
What do you think’s wrong?
And based on all of the memory of all of the data
you’ve plugged in, might be like,
oh, well, it looks like you’re doing this wrong
with your diet and maybe, you know,
you didn’t do this with the weights or whatever.
And it can actually start giving you custom feedback
based on everything you input.
Like I feel like that’s probably something
that can be achieved right now with Clot or GPTs.
I just don’t know the exact formula to do it.
– Yeah, so somebody should go and copy my thing
and then actually make it super useful.
Do your own thing, have at it.
I would love to use it if you do, let me know.
– Yeah, yeah, yeah.
– But I have found that one thing that’s actually,
so when I tried using calorie trackers before,
like you try to add an, you know, something that you ate,
it’s like, you have to like find it in a list
or it’s like, they’re usually like kind of complicated.
But with this, I can just type it in.
And you can even do, if you’re not gonna be super precise,
like, okay, I’m not trying to like win a competition
or something, obviously, like,
I don’t have to be super precise, estimate for me.
I just eat this, estimate, you know.
And the estimations seem to be pretty good.
Like often like within like 20% of the real calories
and protein, so it makes it an easy way to track it.
Just like, just type it in, tell it what you ate
and it’ll come up with a pretty good estimation
of how many calories and protein you took in.
– Awesome, well, the last one that I was gonna share
is actually how I sort of write scripts for short.
So on my YouTube channel, I’ve actually started doing–
– Something different, okay, good.
It’s not–
– Yeah, yeah, this one’s different.
I actually go back to Claude for this one.
You could actually do it with a custom GPT
and it would work just as well.
So if somebody’s listening to this and they’re like,
but I pay for chat GPT, I don’t wanna pay for Claude too.
You can do the same thing in chat GPT.
It doesn’t really matter which one you’re using.
So I’ve actually started doing a lot more shorts
on my channel.
I’m trying to experiment with doing more shorts
as opposed to only doing long form videos
because I wanna try to get in front of different audiences
and shorts tend to get in front of different audiences
than the long form videos.
I’ve also got sponsors coming to me saying,
hey, I would love to pay for a short on your channel.
So I’m like, okay, well, maybe I should start
doing shorts then.
So I’ve actually started playing around with more shorts.
And so I created this custom project in Claude.
And if you’re not familiar with custom projects,
but you are familiar with like custom GPTs,
it’s basically Claude’s version of a custom GPT, right?
So I created this one called shorts writer.
And what I did with it was,
you can see I uploaded a whole bunch of transcripts
from shorts that I thought were really, really good shorts.
So I came across shorts that had a lot of views
that were in sort of technical niches
that talked about AI or talked about like emerging tech
or things like that.
And I downloaded each of the videos
and then I pulled them into Descript
to get the transcript from the video.
And then I uploaded all of the scripts
from all of these videos that I found.
And then basically what I told this custom Claude prompt
to do is to read the scripts that I uploaded
and try to find the sort of consistent formula
that seems to make all of these work well.
And for anything I put into the prompt box,
give me a similar script.
So that’s essentially the way I did it.
And so now if there’s like a new piece of news.
So if I go over to like a news website,
I know you can’t actually see this
’cause I’m just sharing the one tab.
But if I go to like a news website,
there’s some news out today about how the humane pin
is actually getting more refunds
than it has purchases right now.
Not a great look for humane,
but if I was to go and copy the entire article
and come over to Claude,
you can see I can paste in the entire article
and it makes this little like pasted box here.
So I just posted in the entire article from the verge
about how humane’s performance is underperforming right now.
I don’t have to put anything into the prompt box
because it already knows what I’m looking for.
And if I just hit enter on this,
it’s going to read this news article
and then write me a script based on this news article
that I put in here.
– That’s crazy.
You could do that with like a faceless YouTube channel
too, couldn’t you?
– You could, yeah.
So it just gave me,
it knows that I want my script to be under 60 seconds.
I’m trying to model,
Cleo Abrams is like one of my favorite YouTubers
as far as like shorts go.
She does a really good job with them.
So it’s kind of trying to model a similar formula
to what Cleo’s videos are.
And you can see it wrote like a 60th second script
about that news article that I just put in.
Humane just launched their AI pin,
wearable device meant to replace your smartphone,
but things aren’t going as planned.
Imagine you create a revolutionary new gadget.
You spend years developing it,
raise over 200 million from big tech names
and finally release it to the world.
But then more people return it than keep it, right?
And it just wrote this whole script for me.
That’s actually a pretty like compelling,
interesting script that.
– It sounds, yeah, it sounds like a good short.
– That’s crazy.
– You take the script,
I throw it into my teleprompter here.
I read it, I overlay it with B-roll
and I can crank out shorts in 45 minutes, you know?
– Wait a minute, is that what you’re doing?
Are you changing anything?
– Yeah, I mean, a lot of times it’s not specifically worded
the way I would word it, right?
Sometimes it’ll use like delver.
You know, the common words that like make it obvious
that it’s AI.
So I will, you know, tweak some words
to make it sound a little bit more like me,
but for the most part,
the scripts come out pretty good out of the box.
– That’s wild.
– Let me, let me see if I can show you my system prompt here.
So create video scripts that will be one minute
or less in the style of Clio Abram.
Use the transcripts in the project knowledge
to determine the consistent formula behind the video scripts
and use the details about the video idea inside of the prompt
to create a video about the details in the prompt
in the style of the Clio Abram’s videos
following a very similar formula.
So it’s, I uploaded the transcripts
and it’s following a very similar formula
because I really liked her flow.
She always starts with like, imagine this
and then give some more details.
And then like it’s got a very formula flow to it.
And I was like, I really liked that flow.
So now I can plug in any news article, any sales page.
If I need to make a video about like the rabbit R1,
I can go to the rabbit R1 homepage,
copy all of the details from that page, right?
Copy all of the bullets and the selling points
of the product, paste them in,
and it will write a short for me
that will ideally make people interested in the rabbit, right?
So that little like flow for me
has made making short form content really, really easy for me.
– That’s crazy.
I was imagining like, do you combine that with like 11 labs
and like generating a voice reading all of it out?
And then you start using some of the new AI video tools
that are out there.
I think there was a new open source one released today
or it’s going to be released soon.
Yeah, generate some B roll or something like that.
I mean, like you have like most of the video just like done
like automation.
– That’s not really a future I’m looking forward to.
– Yeah, yeah, I know exactly.
I’m like, somebody’s going to do that.
It’s going to be either great or horrible.
– Yeah, lower the, we bury the entry.
The lower we bring this barrier to entry
to create content like this,
the more we’re just going to get flooded with junk.
So I’m like, I’m always sort of hesitant
to share this kind of stuff.
Cause I’m like, this works really well for me.
But I also know like,
if something doesn’t come out quality,
I’m not going to upload it.
A lot of other people aren’t going to have those filters,
right?
A lot of other people are going to go,
I can make a workflow where I can crank out a video
every 10 minutes and just see what works.
I’m not really looking forward to that future,
but I found a workflow that works for me.
And I know, you know,
a lot of others might find it valuable.
If there’s like a content creator that you’re like,
oh, they have a decent formula,
a decent flow that they follow when they make their videos.
You can actually use a tool like this
to reverse engineer the flow of the video
and then use that reverse engineering
to then make videos for you based on the topics
that you input.
And to me, that’s really powerful,
but also the lowering of the barrier for effort
like also makes it sort of scary.
And so I hope nobody goes and takes this idea
and runs with it.
(laughing)
– Tons of people will.
(laughing)
Tons of people will.
– Damn it.
(laughing)
– But I think a long-term, yeah,
people want to see people’s faces
and actually know who the, who’s the person behind it.
And like you said, even if you have Cloud help you make that,
you’re still curating,
you’re still coming up with the idea
to do the video in the first place.
– I think, yeah, I think that’s gonna be
a big differentiator.
I think some people will go out there
and try to make these faceless videos
where they get the formula written for them
and then they plug it into 11 labs
and then they plug it into a video tool
that generates all the B-roll
and then they just throw it online
and nobody knows who’s behind it,
nobody knows why they should care.
I just don’t think it’s gonna work for most people.
Some people are gonna crack that code
and they’re gonna have videos that go viral.
It’s just sort of inevitable.
99% of people will never crack that code
and their videos are gonna get seen by seven people.
I think in the future, as we move forward,
being like a personality online,
being like a name that people can trust
that they find reputable
is gonna become so much more important
than the actual content that you’re putting out there.
I think the faceless channels are just going to be,
well, I can’t really trust this.
I don’t really know the person behind it.
They could just be trying to sell me something.
How do I know this isn’t their affiliate link?
And I think having the face, the personality behind it
is going to be the differentiator
that makes some content work for us versus others.
– Yeah, I bet there will be an opportunity though
for two or three years to make a lot of money doing that.
– Oh dude, some companies gonna roll out an app
that just does it for you.
Like, hey, I need a video about the ERABIT R1.
All right, here it is and it’s just done.
– Yeah, pay me this much money, just pay it.
Stripe, 300 bucks or whatever.
Maybe that’s what Laura should be.
– Yeah, yeah. (laughs)
– Yeah, this has been a fun episode
’cause I feel like there’s a lot of things I learned
from you, like how you’re using AI for video
that I find fascinating.
I think it actually kind of fun to do
maybe a whole episode on that at some point.
– Yeah, yeah, definitely.
– Also, at the same time, it kind of pushes me to,
there’s all these great use cases for AI,
but some of it, there’s a few things I actually use
and a lot of things I know I should be
or should be trying that I have it.
This kind of pushes me to actually go out and try it.
– Man, I’ve gotten so hooked on Claude and Perplexity.
Claude does sort of help with the creation process,
Perplexity to help with the research process.
Between those two tools, I mean,
I pretty much have those tabs open all the time now.
Like I’m hooked on using those to just A,
stay looped in and B, to turn around and create content
that I think people are gonna like out of it.
But you know, I think, like you said,
I really, really enjoy doing episodes like this.
I wanna kind of turn it to the audience for a second.
So if you’re watching this on YouTube
or listening to the podcast, I’d love your thoughts on this.
I think on Spotify, you can actually leave comments now
if you’re watching it on YouTube.
Leave some comments, let us know.
Do you like this style video?
Do you enjoy us showing off use cases?
Do you prefer interviews?
We’re still sort of finding our flow and figuring out
like what is gonna provide the most value
for the people that tune into the next wave?
So your opinion’s valuable.
Let us know in the comments
wherever you’re watching or listening to this.
It’s super, super appreciated.
Thanks again for tuning in and we’ll see you in the next one.
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
like how much better perplexity is than Google.
– You can really, really rapidly do a lot of research
and get a lot of ideas super, super quick.
And to me, that’s really powerful,
but also I hope nobody goes and takes this idea
and runs with it.
– Tons of people will.
– Tons of people will.
(upbeat music)
– When all your marketing team does is put out fires,
they burn out.
But with HubSpot, they can achieve their best results
without the stress.
Tap into HubSpot’s collection of AI tools,
breeze, to pinpoint leads, capture attention,
and access all your data in one place.
Keep your marketers cool
and your campaign results hotter than ever.
Visit hubspot.com/marketers to learn more.
(upbeat music)
– Hey, welcome to the Next Wave Podcast.
I’m Matt Wolf.
I’m here with Nathan Lanz.
And, you know, everybody’s talking about AI.
AI is everywhere.
It’s all over the news.
Every big company is putting AI into everything,
but we keep hearing the same question.
What the heck do I actually use this for?
How is this benefiting me?
How is this gonna make my company better
or my daily life better?
Well, that’s what we’re gonna talk about
in this episode today.
Nathan and I, we’re going to share with you
some of the ways that we’re actually using AI
in our own businesses.
So that’s the goal of this episode,
is to break down the ways
that maybe you’re not even thinking about
that AI could really, really benefit your life.
I think there could be some overlap here
between what you’re gonna say and what I’m gonna say,
but there’ll be some slight nuances
to the way I do it versus the way you do it.
You wanna kick it off?
Like what’s like the first actually actionable use case
for AI that we can talk about here?
– Yes, I saw this tweet from Balaji,
which I thought was really interesting
’cause for a long time I’ve been a huge fan of perplexity.
You know, I’ve started using it for a lot of the research
I do instead of doing a Google search.
I found that often I get better answers
on perplexity actually.
And I’ve started using it for the podcast research,
especially like for interviewing a guest.
You type in the guest’s name
and like the information you get from that
is so much better than just doing a Google search.
You get videos they’ve been in,
you can ask it, follow up questions,
but you know, like what kind of stuff
did they talk about on recent podcast or whatever?
It’s great.
– If people want to learn more about perplexity,
we actually had Arvind, the CEO
and founder of Perplexity on the show.
So make sure you check out that episode.
But what perplexity is,
is it’s essentially a large language model
like chat GPT or like Claude
or whatever large language model you’re used to using.
But it also searches the internet for you
whenever you ask it any sort of questions
and then uses whatever it finds
as part of the context for the response.
So you could ask it questions
just like you would ask chat GPT.
And when it’s formulating that question,
it’s always going to do like a search on the internet
to try to make sure it’s giving you the most
sort of informed response that it possibly can.
And once you start using it,
often it’s kind of hard to go back to regular Google.
– Right.
And so I saw this tweet from Balajas, I was genius,
like, yeah, you can actually just change it, right?
You can go into Google Chrome or whatever browser you use
and you can change your default to be perplexity.
And then it works.
– So Google made that like an option in the settings
where you can just…
It’s like, it’s slightly hidden.
It’s like, it’s not as easy to find.
– But it’s not like you don’t need any sort of extra
Chrome extension or anything like that.
You just go straight into the Chrome settings
and then do a search for it.
And then you can make that your primary search engine.
So now like whenever you type anything,
you do the URL bar at the top,
it’s gonna search for perplexity instead of Google, right?
– Yeah, for example,
if I type in the next wave podcast right here, right?
And so now it’s like pulling up all of our episodes.
It’s giving a synopsis of what the podcast is about.
It’s even like sharing some of the feedback
from listeners, which is wild, related topics.
And yeah, and then you can even follow up questions, right?
You could ask about a certain episode,
what happened on this episode
or what were the key takeaways and it has all that.
– So I’m curious, what are some of the ways
that you’re using perplexity that have, I guess,
made life easier, made business easier,
made daily productivity easier?
Like what are you actually using perplexity before?
– I mean, mainly like researching companies,
research for the podcast, but also for meetings.
Like if I’m meeting somebody and I don’t know much about them,
I go for, I would look at LinkedIn,
but it’s like kind of, you know, so much information.
And a lot of it’s really kind of…
– Well, and perplexity is actually gonna pull in stuff
from LinkedIn a lot of times as well.
– Right, right.
And then you can ask, you can ask follow up questions.
It was just way better ’cause you wanna actually talk to it
and have like a thing where like, oh, where did they go to,
you know, what company did they work at?
What were they actually responsible for?
I just liked that experience better
than looking at someone’s LinkedIn.
And also if they were on videos or whatever,
you get so much more context about a person for a meeting.
So I think that’s like the main use case.
I’m sure there’s tons of others
for people who are doing other kinds of research,
maybe market research, other things
that perplexity is great for.
But like I said, I had a really hard time
developing the habit.
I’m just so used to going to Google or LinkedIn.
And I’m finding that this is definitely helping me
to develop the habit of like, oh,
just use perplexity for that stuff.
– For sure, for sure.
Well, I’ll share my first one ’cause it’s super related.
It’s related to perplexity as well.
I was gonna talk about how I actually do guest research
for the podcast using perplexity.
So basically I plugged in myself
as if I was interviewing myself today
just ’cause I was curious what it would bring up.
But this is the type of thing I’ll do
before every single podcast.
I’ll give it a prompt like, I’m interviewing Matt Wolf,
the YouTuber and creator of Future Tools
on my podcast today.
What should I ask him to ensure
an engaging educational and entertaining episode?
So that’s the prompt that I like to give.
I always like to give extra context
just in case like somebody else has that same name.
Like there’s a professional golfer named Matt Wolf.
And so if I just say like, what should I ask Matt Wolf?
It’ll be like, how do I drive the ball farther
or whatever, right?
So I wanted to make sure I gave that additional context
of the YouTuber and the creator of Future Tools
in my prompt.
So I know it’s gonna pull up the right Matt Wolf.
But then I also don’t want it to just ask
like boring questions about the company, right?
Like let’s say I plugged in somebody
that works at HubSpot or something like that.
I don’t want it to give me questions like,
what are HubSpot’s new initiatives for 2025?
You know, like I don’t want it to get
like boring corporate questions.
I want them to be engaging, educational and entertaining.
So very, very specific in the way I prompt it.
But then like you can see the steps
that perplexity went through.
Research background information about Matt Wolf
and his YouTube channel.
Identify key topics and areas of expertise for Matt Wolf
that would make for an engaging and educational podcast.
Brainstorm a list of potential interview questions
that would cover the key topics
and provide an entertaining and informative episode.
So you can actually see the way perplexity
is sort of thinking through this.
It’s almost borderline agentic, right?
Where it does this one search and it goes,
okay, we’ve got this information now.
Now use this search and it pulls up more information.
And you can see here up at the top,
Matt Wolf creator of Future Tools,
current job founder of Future Tools,
YouTube host and podcaster where I live.
My education apparently pulled all this in from LinkedIn
and has like a little bio of me right there.
Over on the right sidebar,
you can see a whole bunch of content
that I’ve been involved in with the ability to watch it.
And then it brings up some questions.
You know, what sparked your interest in AI?
Can you describe your journey
from being an entrepreneur to becoming a YouTuber?
How do you stay organized and manage your time effectively?
And it broke this all down into sections.
And I’m telling you, like sometimes,
maybe we shouldn’t put this out in the world,
but sometimes we’ll pull on a guest onto our podcast.
And we’ve had such a busy week
that we didn’t have all of the time in the world
to research the guest.
Well, this right here just made it infinitely easier.
So I wanted to share this
because it’s very related to what you were talking about.
But also like, if you’re going and interviewing
for a job somewhere, plug in the company here,
plug in the person’s name that’s interviewing you here,
and learn more about that person.
So you’re going into this interview like dialed in
and ready to have the conversation
with the person that you’re talking to, right?
If you’re about to just jump on a call
or a pitch meeting or something like that,
you can really, really rapidly do a lot of research
and get a lot of ideas super, super quick.
– Yeah, I mean, I think most people don’t realize
like how much better perplexity is than Google.
And like, and it seems like Google,
the quality of Google seems to be going down every year.
And it’s like, most people just don’t realize it
’cause it’s like, it slowly has been happening, right?
– Yeah, yeah.
– But, and now we’re like, you know,
AI content being mass generated,
like the quality is going down more and more it seems.
And, you know, and Google’s tried to fight that
by relying more on authorities.
And now that’s why you see Reddit and Quora at the top,
but now people are using AI to mass spam Reddit and Quora.
So it’s just the quality of Google continues to go down.
It’s when you search for something,
you often don’t get the answer.
Unless it’s something very simple,
you often don’t quickly get the answer to, you know,
to your question, but with perplexity,
you get such great quality, you know, answers,
whatever you’re asking and all the extra information
with the videos and you can do follow-ups.
I highly recommend people try that.
– Yeah, and even like Google is trying
to do this same kind of thing now, right?
Where you do a Google search
and it has the AI response up at the top.
The problem with Google is it’s still to this day,
doesn’t know the difference between a meme and reality.
Right? Like all of the stuff that came out with Google,
like saying, hey, maybe you should try putting glue
on your cheese to make sure it sticks to your pizza.
And how many rocks should you eat per day?
And, you know, geologists recommend you eat
at least 17 rocks a day.
All of that stuff is because memes exist.
It said that random ass stuff and Google thought
that it was, you know, actually factual information
that it fed through its AI.
Perplexity doesn’t seem to have those same sorts of issues
because I think it’s sort of doing a little bit
more cross-referencing than what Google’s doing.
– Yeah, they started from scratch.
Like thinking how to build a search engine
or an answer engine up, you know, from first principles.
Whereas I think Google’s relying
on really antiquated technology.
They built a long time ago.
You know, you can see this in some of the recent stuff
with like people trying to search for Donald Trump
or the assassination attempt and it’s like,
“Oh, did you mean Kamala Harris?”
And it’s like, “What the hell?
What? What are you talking about?”
And I don’t think that was like somebody manually doing that.
That’s probably just based on like all the news sources
they’re pulling information from
and they’re putting certain authority
to those certain news sources.
That’s probably why that happened.
But that’s based on like antiquated technology
and that’s probably why all that’s happening.
So yeah, people should be using perplexity.
Okay, this next one, I saw it from a tweet
from Allie Miller, she calls the “clawed walk.”
And I’ve actually heard this from other people too,
like Dan Shipper, apparently this is one of his top use cases
for AI, is that when you’re thinking about something,
you know, instead of just sitting in your room
or in your office and just kind of, you know,
working on it that way, like actually get out
and get some exercise and do work at the same time, right?
And AI actually makes this actually feasible now.
So like, so what he does and what Allie’s suggesting
people do is go for a walk and then use something
like a super whisper or something like that
to transcribe everything that you’re saying
and then make it, we actually have notes,
which then you could, you know,
you could feed into chat to your Claude,
tell it to remember it or the, you know,
and make it actionable.
And I think Dan said he’s even using that
for his newsletter, I believe.
Like, that’s how he’s writing his newsletter.
And so I was like, I have to start doing that.
‘Cause like, I’ve been on like a big health kick,
especially, you know, I think in the age of AI
being healthy is really important.
And I was like, okay, so if I can get out for a walk
and whatever I’m thinking about from my newsletter,
just say it as I’m walking, right?
That would save so much time.
But also I just, I find that when I’m walking,
I’m more relaxed and I’m more able to think, you know,
it’s kind of different than jogging
where it’s hard to think when you’re walking.
You can think very clearly.
Sometimes even better than when I’m sitting down.
And so I’m trying to get in the habit of doing that.
Having kind of a hard time getting into it
because I feel like in Japan it feels slightly awkward to me.
Like the American guy walking around
where everything’s really quiet here in Kyoto
and I’m like walking around talking to myself outside
and people are staring.
– I feel like people are worried about that
less and less and less these days.
‘Cause it’s so common now to be, you know,
just have like air pods in and be like,
talking to somebody on the phone
while you’re walking around or something.
So I don’t know, I’ve always felt self-conscious
about that as well,
but I feel like it’s definitely getting more normalized.
– Yeah, so here’s Super Whisper.
People can check it out.
See, I’m actually playing on like,
installing this and trying it today.
I haven’t tried it yet.
(upbeat music)
We’ll be right back,
but first I wanna tell you about another great podcast
you’re gonna wanna listen to.
It’s called Science of Scaling, hosted by Mark Roberge.
And it’s brought to you by the HubSpot Podcast Network,
the audio destination for business professionals.
Each week hosts Mark Roberge,
founding chief revenue officer at HubSpot,
senior lecturer at Harvard Business School
and co-founder of Stage 2 Capital,
sits down with the most successful sales leaders in tech
to learn the secrets, strategies, and tactics
to scaling your company’s growth.
He recently did a great episode called,
“How Do You Solve for a Siloed Marketing and Sales?”
And I personally learned a lot from it.
You’re gonna wanna check out the podcast,
listen to Science of Scaling
wherever you get your podcasts.
(upbeat music)
– So what does Super Whisper do?
I’ve heard of it, but I’m not super familiar with it.
– So you can download the app.
I think it’s mainly on macOS, I believe,
and you download it and it just helps you transcribe
whatever you’re saying to it.
So it’s like you just talk to the app
and then it transcribes it for you.
– Okay, similar to Otter.
I use Otter for the same thing.
– Okay, yeah.
– You Otter know.
– How do you use Otter?
– So Otter’s the same idea.
It’s an app that’s on my iPhone that I open it up
and you just start talking into it
and it basically transcribes whatever you say in real time.
One of the ways I actually use it,
this is actually not something that was in my notes,
but I’ll share it anyway
’cause it’s top of mind,
is whenever I go to these conferences,
when I go to Google I/O,
when I go to Microsoft Build,
I’m going to MetaConnect next month,
whenever I go to these events,
I actually pull out my iPhone, open up Otter,
and then set it on my lap
and just let it transcribe the entire presentation
that somebody’s giving.
And then when it’s done,
Otter will give you the summary and bullet points
of like here’s the main takeaways
from the transcript that you just created.
And it will just give me bullets of like,
here’s the 10 things they just talked about
in this presentation.
So I can sit in the presentations
and just completely zone out if I want and not pay attention.
And I’ll just have my cliff notes
of the entire presentation sitting on my phone later.
It’s awesome.
– That’s great.
– But actually, you know what?
I’m going to share,
I’m going to jump to another idea
that I was going to share today
because it’s related to what you just said right there.
– Oh, okay, okay.
– Similar concept,
but one thing that I like to do
when trying to create like documents
like PDFs or any sort of written blog posts
or things like that is I like to speak them out
and then have them transcribed
and then use AI to sort of reword the transcription for me
into something that sounds more like a written article
versus just like reading a transcript, right?
So the most recent one that I did,
I’m sort of laughing.
You’ll understand why I’m laughing in a second.
The most recent one I did was HubSpot reached out to me
and asked if I can help them create a PDF
on all of my thoughts on like the AI world right now.
And they sent me this questionnaire
with like 14 questions on it.
And I looked at it and went,
“Man, I’m going to have to actually like sit down
and type out responses to this whole thing.
Oh, this is going to be a pain in the butt.”
So what I actually did instead
was I gave this questionnaire to Emily,
who’s my assistant.
I gave it to Emily, we jumped on a call
and I said, “Read these questions to me.
Just pretend like you’re interviewing me
for a podcast or something
and read these 12 questions to me.
And I’m going to record this
and I’m going to record my response
to every single question.”
So I recorded the entire conversation
as she read through these 12 different questions.
And I answered every single question.
And then I took that recording
and I pulled it into Descript.
So Descript, you can upload an audio or a video file.
It will transcribe the entire thing for you.
So I used Descript to get my transcription.
And then I took that entire transcription
and I pulled it into Claude.
Claude lets you upload really long text files.
And then I pulled the whole thing into Claude
and I went through and I had it actually answer
the questions again for me one by one,
but using the context of the document that I uploaded.
The idea was I uploaded the entire transcript
as like the context and I said use this transcript
to answer the questions that I’m about to ask you, right?
So then I went through the document again,
the 12 questions and I took question one,
copied and pasted it into Claude.
And then Claude responded to the question,
but based on the answers I gave in my transcript.
And I just went through every question one at a time
and had Claude write up a very succinct response
that had like my sentiment in it.
It had the ideas and thoughts and things
that I shared in my response,
but it did it in like a single paragraph per question.
And so it’s super helpful to write up documents
and articles and things like that.
And so how other people can apply this is,
let’s say you wanna write a blog post on SEO.
I don’t know, I don’t know what niche you’re in,
but let’s say you wanna write up a blog post
all about like the top 10 SEO tactics or something,
write out a list of questions
that you think people would want answered about SEO,
just write out all of those questions
and then record yourself
just answering those questions out loud.
And then you can pull it
into one of these transcription tools like Descript.
Let Descript transcribe the whole thing
and then take that entire transcript,
pull it into Claude and then have Claude
turn that into a written article or a PDF document
or something like that for you.
And it saves so much time
and it writes it probably way better
than you would have written it yourself, let’s be honest.
So that’s a really, really good process,
but it’s very similar to the idea of like going for walks
and just sort of transcribing your thoughts
as you’re walking.
– Yeah, that’s awesome.
That’s anything like that that takes it to an area
where like humans are living more of their lives
and like being healthier
while actually still getting their work done,
I think is awesome.
It actually kind of goes into my next one.
So my next one really is about using AI to like stay healthy.
So this is not for work,
but I do think it’s relevant.
If you’re healthy, you can work better.
And I think a lot of people have not really
connected the dots too that like in the age of AI,
people are probably going to live a lot longer.
And AI is most likely going to unlock
some like major advancements in the healthcare
that people might end up living 10 to 30 years longer,
maybe longer.
So you want to be healthy, right?
Like you don’t want to be like old,
like living a super long life,
but you’re very unhealthy.
Like no, you want to be somewhat healthy.
And so I’ve been trying like on a big health kick
over the last year.
I had gained a lot of weight during COVID
and it’s kind of like kept it on
and like, you know, lost five pounds here and there,
but still was quite overweight.
And then about a year ago,
I started going to the gym a lot like quite often.
And the first thing I realized was like,
you need to really track your calories and your protein.
And I had never done that in my life before.
I had mostly been like kind of slightly overweight
most of my life.
I was super fit when I was maybe like 18 or 19.
And so I realized, okay,
you got to track calories and protein.
And so I’ve been using this thing.
Actually, it’s a custom GPT that I created,
which is, it’s very simple.
Like it doesn’t do all the stuff I want it to do.
It’s like, you know,
I would like for it to do a lot more,
but the memory is quite limited on what,
you know, what they’ll actually remember.
And it’s real simple.
And I just have a thing here where I have like,
for example, I’ll do like a new day, you know,
I’ll just, I’ll like type this in.
I’m typically doing it on the mobile app.
I’m typically not doing it on my desktop.
But anyways, you type in like new day
and you’ll set like what are your calorie goals
for the day, what are your protein goals?
And just anything that I eat,
like if I know the calories and protein,
I will type that in.
And I can even, you can even do just shorthand
just literally just put the,
you don’t even have to tell it what’s calories and protein.
It’ll figure that out.
You can just put the two numbers and it’ll know that.
If you want, you can add a description to it.
So like, oh, I had a latte or I had oatmeal or whatever.
But also the really cool thing is,
you just kind of keep this same window open,
the same tab and like, you can go back to it every day.
And I’m sure at some point,
there’s some kind of memory limit there,
but I’ve been using it for like two months now,
the same one.
– Yeah.
– And then the cool thing is like it,
in that single context,
he remembers all the stuff you shared before.
So like, oh, I had this protein drink or whatever,
it’ll know what the calories and protein are for that.
So you don’t even have to type in the numbers anymore.
– So you’re basically using it to track the calories
and protein.
Is there any other like benefit that you get out of it?
Is it like giving you motivation?
Is it giving you like–
– I would like to do more stuff like that.
Yeah, yeah, I would like,
it’s literally just doing the calories and protein.
Like I wanted to do a lot more.
I was like, okay, I’m gonna make this like a really cool,
you know, custom GBT where like,
it’ll like, it’ll motivate you,
it’ll like track your progress.
Like, yeah, it’s currently not capable
of doing any of that.
I mean, I’m sure you could build an application
that used ChatGBT’s API.
I’m sure you could do that.
But in terms of a custom GBT,
it doesn’t seem to be possible at the moment.
– Well, you could use,
you could use like the rag method, right?
Retrieval Augmented Generation,
where like every time it gives you an update,
copy and paste it into a text file
and then upload that text file and then it will–
– Yeah, I’ve been too lazy to figure that out.
Yeah, I should at some point.
But yeah, I mean, for me it just,
it simplifies tracking calories and protein.
You know, I just type new day every time
when it’s a new day and it starts over.
I want to change anything.
Like, okay, right now I’m trying to diet.
Okay, I reduced my calories by 300 or something like that.
If I’m trying to gain more muscle,
I increase it by 300.
I am trying to give it to where it can actually
coach you more on that.
Like, you know, I tried to feed it like different documents
from like people who I really respect,
like different like health scientists
and things like that or exercise scientists.
And then it’s pretty useful.
Like you can talk to them about like,
okay, I want a bulk right now.
Like, what does that mean?
Like how many calories should I be taking in?
Or should I be doing differently with exercise?
So it will answer that kind of stuff.
– Yeah, yeah.
‘Cause what’ll be cool is like, you know,
you can go in there and be like,
oh, I’ve seen like I’ve plateaued.
I haven’t, you know, been able to add more weight
to my bench press or whatever in two months.
What do you think’s wrong?
And based on all of the memory of all of the data
you’ve plugged in, might be like,
oh, well, it looks like you’re doing this wrong
with your diet and maybe, you know,
you didn’t do this with the weights or whatever.
And it can actually start giving you custom feedback
based on everything you input.
Like I feel like that’s probably something
that can be achieved right now with Clot or GPTs.
I just don’t know the exact formula to do it.
– Yeah, so somebody should go and copy my thing
and then actually make it super useful.
Do your own thing, have at it.
I would love to use it if you do, let me know.
– Yeah, yeah, yeah.
– But I have found that one thing that’s actually,
so when I tried using calorie trackers before,
like you try to add an, you know, something that you ate,
it’s like, you have to like find it in a list
or it’s like, they’re usually like kind of complicated.
But with this, I can just type it in.
And you can even do, if you’re not gonna be super precise,
like, okay, I’m not trying to like win a competition
or something, obviously, like,
I don’t have to be super precise, estimate for me.
I just eat this, estimate, you know.
And the estimations seem to be pretty good.
Like often like within like 20% of the real calories
and protein, so it makes it an easy way to track it.
Just like, just type it in, tell it what you ate
and it’ll come up with a pretty good estimation
of how many calories and protein you took in.
– Awesome, well, the last one that I was gonna share
is actually how I sort of write scripts for short.
So on my YouTube channel, I’ve actually started doing–
– Something different, okay, good.
It’s not–
– Yeah, yeah, this one’s different.
I actually go back to Claude for this one.
You could actually do it with a custom GPT
and it would work just as well.
So if somebody’s listening to this and they’re like,
but I pay for chat GPT, I don’t wanna pay for Claude too.
You can do the same thing in chat GPT.
It doesn’t really matter which one you’re using.
So I’ve actually started doing a lot more shorts
on my channel.
I’m trying to experiment with doing more shorts
as opposed to only doing long form videos
because I wanna try to get in front of different audiences
and shorts tend to get in front of different audiences
than the long form videos.
I’ve also got sponsors coming to me saying,
hey, I would love to pay for a short on your channel.
So I’m like, okay, well, maybe I should start
doing shorts then.
So I’ve actually started playing around with more shorts.
And so I created this custom project in Claude.
And if you’re not familiar with custom projects,
but you are familiar with like custom GPTs,
it’s basically Claude’s version of a custom GPT, right?
So I created this one called shorts writer.
And what I did with it was,
you can see I uploaded a whole bunch of transcripts
from shorts that I thought were really, really good shorts.
So I came across shorts that had a lot of views
that were in sort of technical niches
that talked about AI or talked about like emerging tech
or things like that.
And I downloaded each of the videos
and then I pulled them into Descript
to get the transcript from the video.
And then I uploaded all of the scripts
from all of these videos that I found.
And then basically what I told this custom Claude prompt
to do is to read the scripts that I uploaded
and try to find the sort of consistent formula
that seems to make all of these work well.
And for anything I put into the prompt box,
give me a similar script.
So that’s essentially the way I did it.
And so now if there’s like a new piece of news.
So if I go over to like a news website,
I know you can’t actually see this
’cause I’m just sharing the one tab.
But if I go to like a news website,
there’s some news out today about how the humane pin
is actually getting more refunds
than it has purchases right now.
Not a great look for humane,
but if I was to go and copy the entire article
and come over to Claude,
you can see I can paste in the entire article
and it makes this little like pasted box here.
So I just posted in the entire article from the verge
about how humane’s performance is underperforming right now.
I don’t have to put anything into the prompt box
because it already knows what I’m looking for.
And if I just hit enter on this,
it’s going to read this news article
and then write me a script based on this news article
that I put in here.
– That’s crazy.
You could do that with like a faceless YouTube channel
too, couldn’t you?
– You could, yeah.
So it just gave me,
it knows that I want my script to be under 60 seconds.
I’m trying to model,
Cleo Abrams is like one of my favorite YouTubers
as far as like shorts go.
She does a really good job with them.
So it’s kind of trying to model a similar formula
to what Cleo’s videos are.
And you can see it wrote like a 60th second script
about that news article that I just put in.
Humane just launched their AI pin,
wearable device meant to replace your smartphone,
but things aren’t going as planned.
Imagine you create a revolutionary new gadget.
You spend years developing it,
raise over 200 million from big tech names
and finally release it to the world.
But then more people return it than keep it, right?
And it just wrote this whole script for me.
That’s actually a pretty like compelling,
interesting script that.
– It sounds, yeah, it sounds like a good short.
– That’s crazy.
– You take the script,
I throw it into my teleprompter here.
I read it, I overlay it with B-roll
and I can crank out shorts in 45 minutes, you know?
– Wait a minute, is that what you’re doing?
Are you changing anything?
– Yeah, I mean, a lot of times it’s not specifically worded
the way I would word it, right?
Sometimes it’ll use like delver.
You know, the common words that like make it obvious
that it’s AI.
So I will, you know, tweak some words
to make it sound a little bit more like me,
but for the most part,
the scripts come out pretty good out of the box.
– That’s wild.
– Let me, let me see if I can show you my system prompt here.
So create video scripts that will be one minute
or less in the style of Clio Abram.
Use the transcripts in the project knowledge
to determine the consistent formula behind the video scripts
and use the details about the video idea inside of the prompt
to create a video about the details in the prompt
in the style of the Clio Abram’s videos
following a very similar formula.
So it’s, I uploaded the transcripts
and it’s following a very similar formula
because I really liked her flow.
She always starts with like, imagine this
and then give some more details.
And then like it’s got a very formula flow to it.
And I was like, I really liked that flow.
So now I can plug in any news article, any sales page.
If I need to make a video about like the rabbit R1,
I can go to the rabbit R1 homepage,
copy all of the details from that page, right?
Copy all of the bullets and the selling points
of the product, paste them in,
and it will write a short for me
that will ideally make people interested in the rabbit, right?
So that little like flow for me
has made making short form content really, really easy for me.
– That’s crazy.
I was imagining like, do you combine that with like 11 labs
and like generating a voice reading all of it out?
And then you start using some of the new AI video tools
that are out there.
I think there was a new open source one released today
or it’s going to be released soon.
Yeah, generate some B roll or something like that.
I mean, like you have like most of the video just like done
like automation.
– That’s not really a future I’m looking forward to.
– Yeah, yeah, I know exactly.
I’m like, somebody’s going to do that.
It’s going to be either great or horrible.
– Yeah, lower the, we bury the entry.
The lower we bring this barrier to entry
to create content like this,
the more we’re just going to get flooded with junk.
So I’m like, I’m always sort of hesitant
to share this kind of stuff.
Cause I’m like, this works really well for me.
But I also know like,
if something doesn’t come out quality,
I’m not going to upload it.
A lot of other people aren’t going to have those filters,
right?
A lot of other people are going to go,
I can make a workflow where I can crank out a video
every 10 minutes and just see what works.
I’m not really looking forward to that future,
but I found a workflow that works for me.
And I know, you know,
a lot of others might find it valuable.
If there’s like a content creator that you’re like,
oh, they have a decent formula,
a decent flow that they follow when they make their videos.
You can actually use a tool like this
to reverse engineer the flow of the video
and then use that reverse engineering
to then make videos for you based on the topics
that you input.
And to me, that’s really powerful,
but also the lowering of the barrier for effort
like also makes it sort of scary.
And so I hope nobody goes and takes this idea
and runs with it.
(laughing)
– Tons of people will.
(laughing)
Tons of people will.
– Damn it.
(laughing)
– But I think a long-term, yeah,
people want to see people’s faces
and actually know who the, who’s the person behind it.
And like you said, even if you have Cloud help you make that,
you’re still curating,
you’re still coming up with the idea
to do the video in the first place.
– I think, yeah, I think that’s gonna be
a big differentiator.
I think some people will go out there
and try to make these faceless videos
where they get the formula written for them
and then they plug it into 11 labs
and then they plug it into a video tool
that generates all the B-roll
and then they just throw it online
and nobody knows who’s behind it,
nobody knows why they should care.
I just don’t think it’s gonna work for most people.
Some people are gonna crack that code
and they’re gonna have videos that go viral.
It’s just sort of inevitable.
99% of people will never crack that code
and their videos are gonna get seen by seven people.
I think in the future, as we move forward,
being like a personality online,
being like a name that people can trust
that they find reputable
is gonna become so much more important
than the actual content that you’re putting out there.
I think the faceless channels are just going to be,
well, I can’t really trust this.
I don’t really know the person behind it.
They could just be trying to sell me something.
How do I know this isn’t their affiliate link?
And I think having the face, the personality behind it
is going to be the differentiator
that makes some content work for us versus others.
– Yeah, I bet there will be an opportunity though
for two or three years to make a lot of money doing that.
– Oh dude, some companies gonna roll out an app
that just does it for you.
Like, hey, I need a video about the ERABIT R1.
All right, here it is and it’s just done.
– Yeah, pay me this much money, just pay it.
Stripe, 300 bucks or whatever.
Maybe that’s what Laura should be.
– Yeah, yeah. (laughs)
– Yeah, this has been a fun episode
’cause I feel like there’s a lot of things I learned
from you, like how you’re using AI for video
that I find fascinating.
I think it actually kind of fun to do
maybe a whole episode on that at some point.
– Yeah, yeah, definitely.
– Also, at the same time, it kind of pushes me to,
there’s all these great use cases for AI,
but some of it, there’s a few things I actually use
and a lot of things I know I should be
or should be trying that I have it.
This kind of pushes me to actually go out and try it.
– Man, I’ve gotten so hooked on Claude and Perplexity.
Claude does sort of help with the creation process,
Perplexity to help with the research process.
Between those two tools, I mean,
I pretty much have those tabs open all the time now.
Like I’m hooked on using those to just A,
stay looped in and B, to turn around and create content
that I think people are gonna like out of it.
But you know, I think, like you said,
I really, really enjoy doing episodes like this.
I wanna kind of turn it to the audience for a second.
So if you’re watching this on YouTube
or listening to the podcast, I’d love your thoughts on this.
I think on Spotify, you can actually leave comments now
if you’re watching it on YouTube.
Leave some comments, let us know.
Do you like this style video?
Do you enjoy us showing off use cases?
Do you prefer interviews?
We’re still sort of finding our flow and figuring out
like what is gonna provide the most value
for the people that tune into the next wave?
So your opinion’s valuable.
Let us know in the comments
wherever you’re watching or listening to this.
It’s super, super appreciated.
Thanks again for tuning in and we’ll see you in the next one.
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
(upbeat music)
Episode 19: Can AI tools revolutionize your business workflow? Matt Wolfe (https://x.com/mreflow) and Nathan Lands (https://x.com/NathanLands) dive into six game-changing AI workflows for 2024.
This episode covers everything from using Claude for content creation to leveraging Perplexity for research, illustrating how these tools can enhance productivity and engagement. Matt and Nathan also explore AI transcription tools like Super Whisper and Otter, and how they use a combination of Descript and Claude for efficient document creation. Plus, they touch on AI’s role in personal health management and the potential for AI-generated video content.
Check out The Next Wave YouTube Channel if you want to see Matt and Nathan on screen: https://lnk.to/thenextwavepd
—
Show Notes:
- (00:00)Perplexity is a powerful language model tool.
- (05:10) Preparing engaging podcast prompt with context and depth.
- (06:42) Search tool summarizes personal information, aids research.
- (12:53) Otter app transcribes real-time speech, offers summaries.
- (14:01) Transcribing and rewording content for easier reading.
- (18:13) Pursuing health, tracking calories, and losing weight.
- (21:45) Custom feedback based on data input is desirable.
- (23:58) Creating custom project for analyzing successful short videos.
- (28:59) Lowering barriers to content creation may lead to low-quality flood.
- (30:19) Faceless, formulaic videos won’t work for most.
—
Mentions:
- Claude: https://www.anthropic.com/
- Perplexity: https://www.perplexity.ai/
- Super Whisper: https://superwhisper.com/
- Otter: https://otter.ai/
- Descript: https://www.descript.com/
—
Check Out Matt’s Stuff:
• Future Tools – https://futuretools.beehiiv.com/
• Blog – https://www.mattwolfe.com/
• YouTube- https://www.youtube.com/@mreflow
—
Check Out Nathan’s Stuff:
- Newsletter: https://news.lore.com/
- Blog – https://lore.com/
The Next Wave is a HubSpot Original Podcast // Brought to you by The HubSpot Podcast Network // Production by Darren Clarke // Editing by Ezra Bakker Trupiano
