How We’re Using AI to Dominate YouTube and X in 2024

0:00:02 (upbeat music)
0:00:09 – Hey everyone, welcome to the Next Wave Podcast.
0:00:11 I’m Matt Wolf, I’m here with Nathan Lanz,
0:00:14 and today we’ve got an exciting episode for you,
0:00:17 live from HubSpot Inbound 2024.
0:00:19 So a little bit of a different, special episode.
0:00:22 We don’t normally GIs in front of an audience.
0:00:26 But we’re gonna talk about the AI-powered growth
0:00:29 that we’ve used for our own businesses in both Twitter,
0:00:31 our podcasts, our newsletter,
0:00:33 the various platforms that we have.
0:00:34 We’re gonna share some tools.
0:00:36 We’re gonna share some workflows.
0:00:37 Should be a fun session.
0:00:39 It’s just gonna be me and Nathan sort of having
0:00:41 a off-the-cuff conversation.
0:00:43 (upbeat music)
0:00:48 When all your marketing team does is put out fires,
0:00:49 they burn out.
0:00:51 But with HubSpot, they can achieve their best results
0:00:53 without the stress.
0:00:56 Tapping to HubSpot’s collection of AI tools,
0:00:59 breeze to pinpoint leads, capture attention,
0:01:02 and access all your data in one place.
0:01:03 Keep your marketers cool,
0:01:05 and your campaign results hotter than ever.
0:01:08 Visit hubspot.com/marketers to learn more.
0:01:11 (upbeat music)
0:01:17 But yeah, let’s go ahead and start by talking a little bit
0:01:18 about YouTube growth.
0:01:20 So I do have a YouTube channel.
0:01:24 Right now I’m hovering around 640,000 subscribers.
0:01:27 The channel is really focused on AI,
0:01:31 but surprise, surprise, I actually used AI quite a bit
0:01:33 to actually grow the YouTube channel.
0:01:36 So some of the ways that I’ve leveraged AI
0:01:38 to grow my YouTube channel,
0:01:41 number one is probably the most obvious way
0:01:43 that I’ve used AI to grow my YouTube channel
0:01:45 is the thumbnails.
0:01:48 So all of my thumbnails are created with AI right now.
0:01:51 And I’ll try to quickly walk through the workflow
0:01:53 of how I make some of those thumbnails.
0:01:55 – And they’re awesome, by the way.
0:01:56 That’s like the first way I noticed you
0:01:58 was like these crazy AI art thumbnails.
0:02:01 – So the way I’ve actually done that is,
0:02:05 I use a tool called stable diffusion for the AI art.
0:02:07 And stable diffusion has all sorts of like plugins
0:02:10 and add-ons that you can use with it.
0:02:13 And one of the add-ons is a tool called dream booth.
0:02:16 And dream booth actually lets you upload
0:02:19 about 20 images of your face, it learns your face,
0:02:20 and then you can go into stable diffusion
0:02:24 and say generate an image of Matt Wolf
0:02:26 in a field of strawberries.
0:02:28 And it will generate that image, right?
0:02:30 And it’ll generate an AI version of my face.
0:02:33 I’d say about 75% of the time they look horrible.
0:02:36 25% of the time they actually look decent.
0:02:40 And that’s kind of how I generate my thumbnails.
0:02:41 One of the ways that I like to do it now
0:02:44 to get a little bit more dialed in with those thumbnails
0:02:47 is I like to go and use either a tool like mid-journey
0:02:48 or Leonardo.
0:02:52 They make a little bit higher quality images in my opinion.
0:02:55 So I’ll generate an image of like guy standing
0:02:58 in a field of strawberries to go back to that same example.
0:03:00 And I’ll get that image generated
0:03:02 and then I’ll pull it into stable diffusion
0:03:04 and I’ll sort of mask out my face
0:03:06 and say, “Reflece this face with my face.”
0:03:08 So I get the main image generated
0:03:10 in a different AI image generator.
0:03:11 – That’s awesome.
0:03:12 – And then mask out the face.
0:03:13 – I was gonna ask you because in mid-journey,
0:03:15 obviously you can’t put your face in there.
0:03:16 So that’s how you do it, awesome.
0:03:17 – Yeah, yeah.
0:03:20 So that’s been one of the biggest sort of like growth hacks
0:03:22 I’ve had from my YouTube channel
0:03:24 is I’d like to make these bright thumbnails.
0:03:28 And a lot of times the prompts are like super bright,
0:03:32 super colorful, man-statting in a field of strawberries,
0:03:35 strawberries, vibrant colors, brilliant greens,
0:03:36 you know, things like that.
0:03:38 I try to make it as colorful as possible.
0:03:40 I want these thumbnails to really, really pop
0:03:42 when people see them on YouTube.
0:03:45 I want it to stand out among all the rest of them.
0:03:48 And mid-journey and Leonardo are the two best at that
0:03:49 in my opinion.
0:03:52 And so I use that sort of face swap technique.
0:03:54 Now, the other ways that I use AI
0:03:58 is I really, really love using Claude from Anthopic
0:04:02 for helping with like title generation.
0:04:03 And one of the things I like to do
0:04:05 is I’ll create my whole video.
0:04:09 So I’ll make a video that’s explaining
0:04:12 how to create a chatbot trained on your own information.
0:04:13 Something like that, right?
0:04:15 I’ll create a tutorial on how to do that.
0:04:17 When I’m done making that tutorial,
0:04:18 I’ll pull it into a tool like Descript
0:04:21 and get the whole transcription of that video.
0:04:23 I’ll take that entire transcription,
0:04:25 plug it into Claude, and then say,
0:04:28 help me come up with 10 titles
0:04:30 that will grab a lot of attention
0:04:31 based on this transcript.
0:04:32 – That works pretty well?
0:04:33 – It works pretty well.
0:04:36 It’ll give me, it’ll give me close, right?
0:04:37 Sometimes I’ll change the words a little bit.
0:04:39 Like the titles will be,
0:04:41 I don’t know if I’d actually say that in real life,
0:04:43 but for the most part, it gets me close.
0:04:46 It gets me a lot of ideas to use for those titles.
0:04:49 And I’ll use that as like this sort of rough draft,
0:04:52 the beginning of that title ideation phase.
0:04:55 And so that’s worked really, really well as well.
0:04:56 Another thing you can do with Claude
0:04:58 is they have this tool called Projects.
0:05:01 And Projects lets you upload additional information into it
0:05:03 that it can use as context.
0:05:06 And so what I’ll do is I’ll go find other YouTube titles
0:05:09 that I really like and I’ll make a text file.
0:05:12 And I will list out 30 different titles
0:05:13 that I came across and I’m like,
0:05:14 this title works really well.
0:05:15 This one got me to click.
0:05:16 This is a really good title.
0:05:19 I pull all those titles into a text document
0:05:22 and then upload those into a Claude project
0:05:25 and then say, use these as examples of titles
0:05:27 that have worked well.
0:05:30 And then Claude will sort of do a decent job
0:05:32 of finding titles that are sort of similar
0:05:34 to the ones that I gave it as example.
0:05:36 – I’ve done the same kind of thing with Twitter.
0:05:38 So you showed examples of things you like.
0:05:40 Do you show examples of things you don’t like or?
0:05:41 – Just things I like.
0:05:42 I mean, I couldn’t do that as well.
0:05:44 You could just like in the same text file,
0:05:46 here’s the titles that I like.
0:05:48 Here’s the titles that I really don’t like.
0:05:51 Use this information when creating new titles.
0:05:53 And that kind of stuff works really, really well.
0:05:56 I also use Claude for scripting.
0:05:58 One of the ways that I use Claude for scripting
0:06:00 that has been really effective
0:06:03 is almost the same idea as the title generation idea
0:06:07 where I will go and find maybe 10 to 15 different
0:06:11 YouTube videos that I really like the flow of the video.
0:06:14 They have like a beginning, a middle and end
0:06:16 and it’s just a perfect flow.
0:06:18 I watched the video, it grabbed my attention,
0:06:20 the whole video, I didn’t want to click away.
0:06:23 So what I’ll do is on YouTube,
0:06:25 you can grab the transcripts from those videos.
0:06:28 I’ll grab the transcripts from those like 15 different videos,
0:06:31 pull each one individually into a Claude project
0:06:35 and then tell it, this is the style of script
0:06:37 I want to make for my video,
0:06:38 make a video in the similar style.
0:06:40 And then I’ll all give it the topic.
0:06:44 I’ll say, I want to talk about the new open AI,
0:06:47 01 model, write me a script about open AI 01 model.
0:06:49 Here’s some information about it,
0:06:51 write it in the style of the scripts that I uploaded.
0:06:54 – Do you modify it or do you just do what it says?
0:06:56 – No, I modify them heavily.
0:06:58 What it gives me, it gives me that flow, right?
0:07:00 What I’m really looking for is the flow.
0:07:02 I don’t need a word for word script.
0:07:04 I don’t actually read a script when I make my videos.
0:07:07 It’s very off the cuff, but–
0:07:08 – I like the bullet points of like,
0:07:10 here’s the kind of stuff I’ll give up.
0:07:11 – There’s always like a, you know,
0:07:14 a hook in the beginning to grab the attention.
0:07:15 And then once you grab the attention,
0:07:17 then you want to move on to, you know,
0:07:20 showing what the end result is going to be.
0:07:22 And then you want to move on to getting into the tutorial.
0:07:24 And then you want to move into, you know,
0:07:27 and it gives you that flow of where you want the video to
0:07:28 start and where you want it to end.
0:07:30 And, you know, the little bits in the middle,
0:07:33 and it sort of helps me find that arc of the video.
0:07:35 And then I’ll use that as the sort of rough draft
0:07:39 outline for my video, because I will upload information
0:07:42 about the topic I want the video to be about.
0:07:45 And it will sort of give me the beats to follow
0:07:46 based on the scripts.
0:07:48 – Yeah, I wonder how many YouTubers know about
0:07:50 all of these strategies, because it seems like
0:07:51 your team is way leaner.
0:07:53 Like for the size of your channel,
0:07:55 like I know other YouTubers who have like
0:07:57 way larger teams for like the same quality
0:07:58 output that you’re doing.
0:08:00 – I have taught one other person in my process
0:08:02 who was making thumbnails for me for a little while.
0:08:06 So it is very sort of teachable, duplicatable.
0:08:07 Anybody can follow that flow.
0:08:09 The shorts is the same flow, right?
0:08:12 So I will go and grab the transcripts of shorts.
0:08:14 – Shorts is a brand new thing I just started doing.
0:08:16 I’ve done like 10 total shorts on my YouTube.
0:08:17 – I don’t know, I think you said with shorts
0:08:19 you’re allowing it, you’re like kind of following
0:08:23 what it says a little bit more on the shorts.
0:08:25 And the reason I do that for shorts
0:08:27 is ’cause you’ve got to hit 59 seconds or less, right?
0:08:30 And you’ve got to hit it like right on that time.
0:08:34 And so if I try to add limb too much, I tend to ramble.
0:08:36 I mean, I’ve already been talking about my YouTube stuff
0:08:37 for 10 minutes here, right?
0:08:38 – You’re still going on it.
0:08:39 – Yeah, I’ll keep going.
0:08:42 So I kind of need a script to keep that 59 seconds
0:08:44 or less on my YouTube shorts.
0:08:47 But if you go and find a short on YouTube
0:08:50 that you really like, if you go and you see
0:08:53 the youtube.com/shorts/, you know,
0:08:56 a long string of letters and numbers, right?
0:08:59 If you replace the word shorts with video,
0:09:01 it’ll actually take you to the video landing page
0:09:02 instead of the short landing page.
0:09:04 And you can grab the transcript from that page.
0:09:06 So for shorts, I’ll do that.
0:09:08 I’ll go and pull in transcripts for inspiration
0:09:10 so that I know I’m hitting all the sort of beats
0:09:12 that I want to hit throughout my video.
0:09:13 – Right.
0:09:15 Have you tried to get the faceless kind of YouTube stuff
0:09:16 or do you think it’s good?
0:09:18 – I haven’t yet, but I do want to explore it.
0:09:21 I was watching the Hagen session yesterday.
0:09:21 – Yeah.
0:09:24 – And that really made me want to play around
0:09:25 with some of that.
0:09:26 See if I, I want to test it,
0:09:28 put a video up on my channel where it’s,
0:09:30 I want to do the Hagen where it’s my face.
0:09:34 It sounds like my voice and then put some like B roll over it
0:09:35 and then see if anybody notices.
0:09:37 I don’t think I’ll do that for like all of my videos,
0:09:37 but I kind of want to test it.
0:09:39 Just be like, is there any way to call me out
0:09:41 and tell if it’s an AI generated version?
0:09:43 – Yeah. I think you know, I’m considering starting
0:09:47 an AI video agency with Lord.com and the two guys,
0:09:48 I’m working with it on,
0:09:50 they showed me a workflow yesterday using make.com
0:09:53 where we could like script out an entire YouTube video
0:09:57 and then use AI video to generate all of this stuff.
0:09:58 And it’s amazing.
0:09:59 I mean, like you probably could like grow
0:10:00 a pretty large channel.
0:10:01 – Oh yeah.
0:10:02 Once you get into some of those make.com workflows,
0:10:04 it gets really, really fun.
0:10:07 But yeah, those are some of my main YouTube strategies
0:10:09 that I’ve used AI to grow on YouTube.
0:10:10 – Yep.
0:10:11 – But let’s talk a little bit about Twitter
0:10:14 because you’ve grown a pretty big following
0:10:15 on Twitter as well.
0:10:18 – Yeah. I mean, so, I mean, I was like a Twitter lurker
0:10:19 for like five or six years.
0:10:22 Like I would use Twitter, like mostly just like read Twitter,
0:10:23 you know, now X, whatever.
0:10:25 And I never really tweeted often.
0:10:27 And then I still had like 5,000 followers,
0:10:30 just got did startups in Silicon Valley and rate up cap,
0:10:31 you know, raise VC funding and stuff.
0:10:34 So I had like 5,000 to start with.
0:10:36 And then I started getting really excited about AI.
0:10:37 Obviously like writing my newsletter.
0:10:39 And I was like, well, how do I grow the newsletter?
0:10:40 – Yeah.
0:10:41 – I was like, okay, well, I guess you use Twitter
0:10:43 and like share whatever you’re learning about.
0:10:45 And so I started doing that.
0:10:46 And then quickly I learned that I could use AI
0:10:48 in so many different ways to grow the Twitter.
0:10:50 And a lot of it’s similar to what you said.
0:10:52 Like I would teach the chat to BT,
0:10:56 like what are the kind of Twitter threads that work?
0:10:57 Like what are the kinds I like
0:10:59 and what are the kinds that I don’t like?
0:11:01 And it probably, it would save me like two or three hours
0:11:02 like doing threads.
0:11:04 And I’m not as active right now.
0:11:06 I’ve been more focusing on getting our podcast going
0:11:07 for the last five months.
0:11:09 But when I was like doing it every day,
0:11:13 like I could consistently get at least 300,000 views
0:11:15 on tweets every single day.
0:11:16 – So how are you using AI to do that?
0:11:18 Like what was the workflow?
0:11:19 – Well, so what I would do,
0:11:21 I use this thing that’s not an AI tool.
0:11:22 It’s called a tweet hunter.
0:11:24 It’s like a Chrome extension.
0:11:24 So I would use that.
0:11:27 And then what tweet hunter would do is you can look at
0:11:29 like who are the top people in your category, right?
0:11:32 So it’s like, okay, AI or creators or whatever.
0:11:33 You can look up those people
0:11:37 and then you can see what are their top tweets ever, right?
0:11:39 So I would do that and I would make a list.
0:11:41 Like, here’s the people I really want to,
0:11:43 not copy, but I like their style.
0:11:44 And so I would look at,
0:11:46 and I would get their top one or two tweets
0:11:48 and I would put that into a chat to BT
0:11:51 and say, here’s the kind of tweets I like.
0:11:53 I’d use AI as an editor, not as like the writer.
0:11:55 ‘Cause like the writer, I felt like there’s only
0:11:58 a few niche use cases where you can actually use it
0:11:59 to like write the content.
0:12:01 I think it works a little bit better in LinkedIn.
0:12:03 I’m not sure why, maybe like the quality of the content
0:12:05 is a little bit different on LinkedIn.
0:12:07 On Twitter, I felt like it writing the content
0:12:08 for me didn’t do that well.
0:12:10 So I would write the content myself
0:12:11 and then use it as an editor.
0:12:14 And I felt like they need probably like two hours or so.
0:12:17 ‘Cause like editing a Twitter thread
0:12:19 to make it like really go viral can take hours.
0:12:21 Like I’d be sitting at a coffee shop in Kyoto,
0:12:25 like just sitting there like, okay, look at it again.
0:12:26 Let me look at the hook again.
0:12:28 And then having the AI where I could just like
0:12:30 show it all that and get feedback.
0:12:31 It’s taking me so much time.
0:12:33 And actually I felt like when I was doing it myself,
0:12:34 I was getting a lot less views.
0:12:36 And as soon as I started using AI as like an editor,
0:12:38 all of a sudden like, oh, every time I tweet,
0:12:39 it goes viral.
0:12:40 Well, that’s great.
0:12:41 – Yeah, yeah.
0:12:43 (upbeat music)
0:12:44 – We’ll be right back.
0:12:46 But first I wanna tell you about another great podcast
0:12:47 you’re gonna wanna listen to.
0:12:51 It’s called Science of Scaling, hosted by Mark Roberge.
0:12:54 And it’s brought to you by the HubSpot Podcast Network,
0:12:57 the audio destination for business professionals.
0:12:59 Each week hosts Mark Roberge,
0:13:02 founding chief revenue officer at HubSpot,
0:13:04 senior lecturer at Harvard Business School
0:13:06 and co-founder of Stage 2 Capital,
0:13:09 sits down with the most successful sales leaders in tech
0:13:12 to learn the secrets, strategies, and tactics
0:13:14 to scaling your company’s growth.
0:13:16 He recently did a great episode called,
0:13:19 “How Do You Solve for a Siloed Marketing in Sales?”
0:13:21 And I personally learned a lot from it.
0:13:23 You’re gonna wanna check out the podcast,
0:13:27 listen to Science of Scaling wherever you get your podcasts.
0:13:30 (upbeat music)
0:13:33 Now you were telling me a story about Naval Ravicon, right?
0:13:34 – Yeah, a lot of things came from me growing on Twitter.
0:13:36 I mean, like, in a way, like,
0:13:38 ’cause like very quickly I went for like 5,000 followers,
0:13:40 like 50,000, I think in like three months.
0:13:41 – Right.
0:13:42 – But, you know, a lot of stuff came from that.
0:13:44 Like I went on TV two or three times
0:13:47 and talked about AI, Elon Musk started responding to me.
0:13:48 – Right.
0:13:49 – You know, Jeff Bezos followed both of us.
0:13:50 – Right.
0:13:51 – And that’s kinda how we connected is
0:13:53 ’cause Jeff Bezos saw one of my tweets
0:13:54 and followed both of us.
0:13:56 – Well, I think it was a tweet that you put that was like,
0:13:58 here’s the people to follow in AI
0:13:59 and you listen to a bunch of people
0:14:00 and my name was one of them.
0:14:01 – Yeah.
0:14:02 – Jeff Bezos just followed that whole list.
0:14:05 – The idea for that thread came from AI.
0:14:06 So I was like, here’s the kind of stuff
0:14:08 that’s the other people doing in different categories.
0:14:09 – Right.
0:14:10 – And I do that in the AI category.
0:14:11 – Yeah.
0:14:12 – Like, oh, here’s the top people to follow or, you know,
0:14:14 and those kind of, you know,
0:14:15 threads don’t get as many views,
0:14:16 but they would like,
0:14:18 a lot of people I would mention, they would all follow me.
0:14:18 – Yeah.
0:14:20 – Right, so in terms of the Naval,
0:14:22 that was the one use case where I found with the Twitter
0:14:24 that you could actually use AI
0:14:26 to actually create the content.
0:14:27 To give people context,
0:14:29 Naval Ravik, the founder of Angel List,
0:14:30 in Silicon Valley,
0:14:32 he’s one of the most famous people in Silicon Valley.
0:14:34 He did a great podcast episode with Joe Rogan.
0:14:36 It’s one of my favorite podcast episodes ever
0:14:37 ’cause it talks all about like,
0:14:39 everything from business to like life, right?
0:14:40 How to live a good life.
0:14:43 And so I literally just like fed the transcript
0:14:46 of that episode into chat to BT.
0:14:47 And so he, and then he already knew like,
0:14:49 what kind of threads I like and all that sort of,
0:14:50 had that context.
0:14:51 And I just fed it to transcript and said,
0:14:55 make me a thread, put that into Twitter, you know, Twitter.
0:14:58 And yeah, it got like maybe like 500,000 views.
0:15:00 Naval retweeted it.
0:15:02 I didn’t write any of it at all.
0:15:05 I literally just like fed it to chat to BT and everything.
0:15:07 – Do you remember what sort of prompt you gave
0:15:09 to just to get that kind of output for it?
0:15:11 – It was literally just like me giving you the contact.
0:15:13 So here’s the threads I like and the ones I don’t like.
0:15:14 – Okay.
0:15:15 – Right, ’cause like, yeah,
0:15:16 I want to have like a good hook and all that,
0:15:17 but I don’t want to be like too cheesy.
0:15:19 I don’t want to go too like click maybe.
0:15:21 Oh, there’s like a fine line there, right?
0:15:23 Like a little bit, but not too much, right?
0:15:25 And it’s literally just providing the context,
0:15:27 kind of like the same thing you do with YouTube, right?
0:15:29 Like giving it like, here’s the kind of titles I like,
0:15:31 or, you know, the flow as I like.
0:15:32 Same kind of thing works for Twitter,
0:15:34 probably works better for LinkedIn, honestly.
0:15:35 – Yeah, yeah.
0:15:36 And I mean, something like that works
0:15:38 if you’re trying to grow a podcast
0:15:40 or grow a YouTube channel or even grow a newsletter, right?
0:15:45 You can go in, plug the content from the YouTube video
0:15:46 or the transcript from the podcast
0:15:49 or the entire newsletter that you just wrote up,
0:15:52 plug in the chat GPT and say, hey, summarize this
0:15:54 into something that would work well on it as a Twitter thread.
0:15:56 And at the end, add your little call to action.
0:15:58 And yeah, I think there’s a lot of ways
0:16:00 we probably could be using AI more, right?
0:16:01 Like we both have AI newsletters, you know,
0:16:04 he’s got future tools, I’ve got lord.com.
0:16:06 And I used to do more like free call to actions
0:16:08 and post call to actions to promote my newsletters.
0:16:09 Like, okay.
0:16:10 – So you need to find that?
0:16:11 What’s a free call to action?
0:16:12 – A free call to action is like, okay,
0:16:14 my newsletter is coming out tomorrow.
0:16:17 Here’s what you might learn in this newsletter issue, right?
0:16:18 – Okay.
0:16:21 – And then the post is yesterday, so-and-so, you know,
0:16:22 20,000 people got this newsletter,
0:16:25 you know, the Lord newsletter, here’s what I learned.
0:16:26 Here’s what you missed out on, basically.
0:16:27 Got it.
0:16:28 And subscribe.
0:16:29 And so those typically don’t get tons of views.
0:16:33 They may get like 10,000 views versus like one of my other
0:16:34 tweets getting like over a hundred thousand views,
0:16:36 but they convert really well.
0:16:37 And so, yeah, I think you probably,
0:16:39 we probably should be automating that, right?
0:16:41 Like we probably should just like be like feeding our
0:16:43 newsletter into like a system and then having it like,
0:16:45 just generate those tweets and like schedule them
0:16:46 through the Twitter API or something.
0:16:48 – Yeah, yeah.
0:16:48 I like that idea.
0:16:50 I think when it comes to Twitter personally,
0:16:53 I still really like it to be me, right?
0:16:54 It’s going to be my voice.
0:16:56 But if we started like a next wave podcast Twitter,
0:16:59 I could see like trying to automate that with, you know,
0:17:00 here’s what we talked about in this episode.
0:17:02 Every episode gets fed and like you can use something
0:17:04 like make.com, right?
0:17:05 And I’m just sort of spitball like ideas here.
0:17:07 You could use something like make.com,
0:17:09 where whenever you make a new podcast episode,
0:17:13 it watches that RSS feed, pulls it into like a descript
0:17:16 or some sort of transcription tool, right?
0:17:17 – Or we can make a newsletter.
0:17:17 – Transcribe it.
0:17:21 And then it automatically uses the cloud or GPT API
0:17:23 to then turn it into a Twitter thread
0:17:26 and then automatically post on a Twitter account.
0:17:27 That actually is something we probably should do
0:17:28 with the next wave.
0:17:30 – Yeah, we can make a newsletter, right?
0:17:34 We could like, we’re like the key takeaways of every issue.
0:17:36 And then you could also automate the promotion of as well.
0:17:37 Yeah.
0:17:38 – Yeah, well, I mean, let’s talk a little bit more
0:17:41 about some of these newsletter strategies
0:17:42 to like grow a newsletter, right?
0:17:45 We’ve both grown fairly substantial sized
0:17:46 newsletters in the AI space.
0:17:49 You’ve got Lord.com, I’ve got the future tools newsletter.
0:17:53 So we’ve used obviously AI to help grow those
0:17:55 and manage them and maintain them.
0:17:57 One of the ways I’ve been experimenting with growing it
0:18:01 is actually using AI to create the sort of lead magnet
0:18:04 to bring somebody onto the list in the first place.
0:18:07 And with these AI tools, it’s so easy to create
0:18:10 really high value lead magnets right now, right?
0:18:14 Like I can go and take some of my YouTube tutorials
0:18:17 that teach really valuable concepts
0:18:20 that people really wanna learn, have those transcribed,
0:18:22 pull those transcriptions into something
0:18:25 like Claude or Chad GPT and say,
0:18:28 this is a transcription, make it read more like an ebook
0:18:30 or make it read more like a blog post
0:18:31 or you know, that sort of thing.
0:18:33 So that it feels more readable
0:18:36 than just reading a transcription, right?
0:18:38 And now you’ve got a PDF that you can offer
0:18:41 as an opt-in to get people to join your newsletter.
0:18:44 Another thing that’s gotten really, really easy now
0:18:47 is even easier now with the OpenAI 01 platform
0:18:52 is going and creating a simple software app, right?
0:18:54 Like so some of the things that I’ve seen,
0:18:56 I’ve seen like in the real estate industry,
0:19:00 people making like fairly simple mortgage calculators
0:19:03 and things like that, make really simple tools
0:19:06 where you ask for the email in order to use the tool
0:19:07 and you put a little check box below
0:19:08 and saying, would you also like to sign up
0:19:12 to my newsletter when you use this tool?
0:19:14 That reminds me of the stuff that Greg Isenberg
0:19:15 was talking about when you came on our show,
0:19:18 like creating widgets for like SEO and stuff, right?
0:19:20 Like you see widgets, you have to like hire a team
0:19:22 and go build that and like now,
0:19:24 especially like with 01 that just came out,
0:19:27 like OpenAI recently released 01.
0:19:28 It’s not clear if it’s a new model or what it is,
0:19:30 but it’s way better at coding
0:19:32 and like creating apps rather than before.
0:19:34 And so I’m pretty sure most of these widgets now,
0:19:37 like a single creator can create like 10 useful widgets
0:19:39 and like get tons of SEO traffic.
0:19:41 There’s so many things like it’s hard to figure out
0:19:43 like what to do ’cause like there’s plenty of opportunities
0:19:43 now.
0:19:45 You can really dial it in towards whatever niche
0:19:46 you’re operating in.
0:19:50 So you can create something like really, really customized
0:19:52 just for the people that you’re speaking to.
0:19:54 And there’s another tool that’s been fairly popular lately
0:19:58 called Cursor, which is like a fork of visual studio code.
0:20:02 But it’s got like AI scripting directly built in.
0:20:04 I use Cursor, it’s amazing.
0:20:05 Like I’m like an amateur coder,
0:20:07 like I’ve been in Silicon Valley for a long time
0:20:09 so I used to code more and now I don’t code a lot,
0:20:11 but now it’s like easier to go back to coding
0:20:14 ’cause like it can look at your entire code base
0:20:15 and like suggest changes or like,
0:20:16 oh, here’s a bug in the code
0:20:18 and like it knows the entire code base.
0:20:19 It’s right.
0:20:20 Yeah.
0:20:23 You know, in another way that I use AI for my newsletter
0:20:25 is I let it to be my proofreader for me.
0:20:27 It’s essentially become like my editor, right?
0:20:28 I actually have a team that helps me
0:20:30 with the newsletter as well now,
0:20:32 but I still write a bunch of it.
0:20:33 My team still writes a bunch of it,
0:20:36 but the final pass, we let AI look at it
0:20:38 and we usually plug it into,
0:20:39 I was using chat GBT in the beginning,
0:20:41 now I’m using Claude a little bit more,
0:20:42 but I’ll plug in the whole newsletter
0:20:45 and I’ll say proofread this for grammar,
0:20:47 spelling and readability and it will look through it
0:20:49 and it’ll tell me all the grammar and spelling
0:20:52 looks pretty good, but you might wanna, you know,
0:20:53 move this sentence down a little bit,
0:20:55 move this sentence up a little bit.
0:20:56 It just makes it easier to follow
0:20:58 what you’re trying to say, right?
0:21:01 And it’s really good as sort of rearranging some,
0:21:04 you know, little nuanced stuff in the email
0:21:09 to make it more readable for the newsletter subscribers.
0:21:10 – Yeah, I found the same thing,
0:21:13 like I tried to have AI write part of my newsletter
0:21:16 and like it’s really clear when like you’re using AI
0:21:17 to actually write the thing, right?
0:21:19 Like I tried that one issue
0:21:22 and like people like responded back, like, what is this?
0:21:23 – Yeah, yeah.
0:21:24 – Like, okay, that’s obviously not Nathan, like, okay.
0:21:27 And so I use Claude to like edit my newsletter.
0:21:28 – Yeah, for sure, for sure.
0:21:29 And then we also talked a little bit
0:21:31 about the Twitter threads, right?
0:21:34 Like, I’m assuming is Twitter the way
0:21:35 that you’ve grown your list the most?
0:21:37 Is it mostly because of the–
0:21:39 – Yeah, yeah, it’s mostly been Twitter.
0:21:41 Like, you see every time I do a viral tweet, you know,
0:21:45 you get like 100 to 200 subscribers every single time.
0:21:46 – Yeah.
0:21:47 – You know, of course some people unfollow,
0:21:48 you might lose 10 or 20 people.
0:21:50 They’re like, yeah, and it’s pretty consistent.
0:21:53 – Yeah, yeah, I mean, my newsletter’s mostly grown
0:21:54 off the back of the Future Tools website
0:21:56 a little bit from the YouTube channel.
0:21:58 I found that a lot of people, like I always have a call
0:21:59 to action at the end of my YouTube video
0:22:01 that goes to subscribe to the Future Tools newsletter.
0:22:04 I find that most of the traffic that goes
0:22:06 to the newsletter does not actually come from YouTube.
0:22:08 It mostly comes from the Future Tools website,
0:22:10 which is a tool that I built to help sort of organize.
0:22:12 – Yeah, it’s hard to convert like different mediums, right?
0:22:15 From like video to text and vice versa.
0:22:18 – Yeah, that’s really the way that I’ve grown my newsletter.
0:22:21 I’m at about 160,000 subscribers on mine now,
0:22:24 but it’s mostly come from basically building a tool
0:22:26 over at Future Tools, a tool to help you find tools.
0:22:28 – Yeah, you were talking about widgets earlier.
0:22:29 So have you actually created the widgets yet?
0:22:32 Or is that like an idea you’re thinking about using AI for?
0:22:33 Or you already have done it?
0:22:35 – So I’ve messed around creating a few widgets,
0:22:37 but I actually haven’t used them as an opt-in yet.
0:22:41 So it’s something that I want to play around with more.
0:22:43 I mean, if you look at Future Tools as a tool
0:22:44 that helps you find tools,
0:22:47 that’s the biggest growth for my newsletter, right?
0:22:49 But I mess with VS Code a lot.
0:22:51 I’ve been using Claude mostly for coding.
0:22:53 I’m gonna probably start using OpenAI01
0:22:54 for coding a little bit.
0:22:56 And I’ve made little widgets and stuff
0:22:58 that I use internally for myself,
0:23:00 but I haven’t actually given many away yet.
0:23:01 – Yeah, I think you should,
0:23:02 ’cause like you said that Future Tools,
0:23:04 so Matt has Future Tools,
0:23:06 it’s basically the top AI tools directory.
0:23:10 And my understanding is like when people search for AI tools,
0:23:12 you’re often like the top two Google results
0:23:13 at the time, right?
0:23:14 – Yeah, yeah, with Future Tools,
0:23:16 yeah, it’s really well SEOed.
0:23:19 And all of the descriptions of every single tool
0:23:21 on Future Tools was generated with AI.
0:23:22 – That’s great.
0:23:26 – So basically the way that works is perplexity goes
0:23:29 and looks at the tool, gives me a description of the tool.
0:23:31 All of this is through a make.com automation, right?
0:23:33 – Yeah, we’re not sponsored by make.com.
0:23:34 – We’re not either.
0:23:37 But basically perplexity will go and look at the tool,
0:23:38 give me a description of the tool,
0:23:40 send that description over to Claude.
0:23:42 Claude will write a single paragraph of it
0:23:44 and then feed it into Webflow,
0:23:47 which is what the whole database is built on top of.
0:23:50 So that workflow is all AI.
0:23:53 Every single description on that entire site is AI.
0:23:55 And if anybody says that, you know,
0:23:57 you can’t rank for AI-generated content,
0:23:59 I’m proof that that’s false,
0:24:01 because it’s usually the tool that’s number one.
0:24:03 And then the Future Tools listing for that tool,
0:24:06 that’s number two for a lot of AI tools.
0:24:06 – Yeah, I know.
0:24:09 I mean, so we had Greg Eisberg on like several times.
0:24:12 And I know he has like an AI SEO agency
0:24:13 and he said is like killing it.
0:24:15 Like they’re doing it really well.
0:24:16 And I think they’re mainly,
0:24:19 they’ll do stuff where they have like AI generate
0:24:22 like thousands of pages of content and they rank.
0:24:25 And then what they do is then they’ll see,
0:24:27 ’cause obviously the quality will be a little bit lower
0:24:28 with the AI content.
0:24:29 And so what they’ll do is they’ll generate
0:24:31 like thousands of pages of content
0:24:32 and they’ll see what it ranks.
0:24:34 And once they find out what’s ranking,
0:24:36 then they’ll go in with humans and edit the ones
0:24:38 that are ranking to make them really, really good.
0:24:39 – Smart.
0:24:40 – Right?
0:24:41 Yeah, so that’s, if you have like one key takeaway,
0:24:43 that’s one, probably everyone who’s doing SEO
0:24:45 should be probably doing that right now.
0:24:47 – Yeah, one thing I want to share before we wrap up,
0:24:48 ’cause we’ve got a couple minutes left here,
0:24:51 is if you don’t have a newsletter,
0:24:54 it’s probably easier than ever to have a newsletter
0:24:58 because you can use AI to sort of help you out,
0:25:00 at least at the very least outline the newsletter
0:25:01 if you still want to write it yourself, right?
0:25:03 So if you’re in like a specific niche,
0:25:05 like we’re both making newsletters in the AI niche,
0:25:08 I can go and find, you know,
0:25:10 the seven biggest news stories in the AI world
0:25:13 for the week, take all seven of those stories,
0:25:16 plug them into something like cloud or chat GPT,
0:25:18 and then say, you know, either write up
0:25:20 or outline a newsletter for me
0:25:23 with all of these news pieces worked into it.
0:25:25 And you can even feed it some of your past content
0:25:28 so that it’ll try to write sort of in your style.
0:25:32 And now you’ve got a quick and easy newsletter,
0:25:33 at least at the very least,
0:25:34 you’ve got an outline or a rough draft
0:25:35 that you can go back in
0:25:37 and sort of add your own voice to,
0:25:39 but it makes it really, really easy
0:25:42 to consistently put out a new newsletter.
0:25:43 Just pick your niche,
0:25:45 make sure you’re keeping an eye on the news in that niche,
0:25:48 and then feed that in once a week or twice a week
0:25:50 or whatever the cadence of your newsletter is.
0:25:52 – Yeah, and I think it works for like curation
0:25:53 focused newsletters.
0:25:55 I do wonder though, like in the next few years,
0:25:56 if that’s gonna die off,
0:25:58 because the AI tools are gonna get so good
0:25:59 that you’ll just be able to like have your own
0:26:01 custom newsletter of like curated information.
0:26:02 And then–
0:26:04 – Oh, it’s newsletter, well, I think.
0:26:05 – Yeah, so like, that’s why I’m trying to be like
0:26:06 more editorial too,
0:26:08 like actually writing what I think
0:26:10 about stuff emergency is just like, here it is, you know.
0:26:12 – Yeah, yeah, no, I definitely think
0:26:14 that’s where it’s gonna go.
0:26:15 Unfortunately, that’s probably gonna sort of
0:26:16 devalue your newsletters,
0:26:18 but we still have a window of opportunity right now.
0:26:19 – We still have a podcast.
0:26:23 So, you know, I feel like people should always capture
0:26:26 that window of opportunity while you’ve got it and–
0:26:28 – I mean, it’s helped us, so we started our podcast
0:26:30 with HubSpot maybe five months ago.
0:26:33 We got, I don’t know, three something thousand subscribers
0:26:34 in the first month, and now we’re–
0:26:35 – Where are the episodes?
0:26:36 Was it first episode?
0:26:37 – Oh yeah, that’s right.
0:26:38 – Yeah.
0:26:39 – And then now we’re like 10.3, I mean, in five months,
0:26:41 and a lot of that’s from our newsletters.
0:26:43 So, yeah, newsletters work.
0:26:45 – Yeah, yeah, well, I think that’s about
0:26:47 all of the time we have.
0:26:50 This has been a fun, interesting episode.
0:26:52 Fun story, this is actually the first time
0:26:54 Nathan and I have met live in person.
0:26:55 We’ve been doing a, we’ve known each other
0:26:56 for a couple of years now.
0:26:57 We’ve been doing a podcast.
0:26:59 – I told him I was six months old, you know.
0:27:01 – But this is actually our first time meeting in person.
0:27:03 He lives in Japan, I live in San Diego.
0:27:05 And we’ve finally converged here at HubSpot
0:27:07 to record a few of these episodes live.
0:27:10 So, we really appreciate everybody hanging out with us
0:27:12 and enjoying the podcast.
0:27:15 If you want to hear more content like what we did here,
0:27:17 our podcast is on YouTube.
0:27:18 It’s called the Next Wave Podcast.
0:27:20 And you can find the audio version
0:27:21 wherever you subscribe to podcasts.
0:27:24 So, thanks again for hanging out with us
0:27:26 and tuning into this episode.
0:27:26 – Thank y’all.
0:27:29 (upbeat music)
0:27:32 (upbeat music)
0:27:34 (upbeat music)
0:27:37 (upbeat music)
0:27:40 (upbeat music)
0:27:43 [MUSIC PLAYING]