Latest / Content, Briefly / How to collect synthetic feedback (+ more Claude tips)

Transcript

How to collect synthetic feedback (+ more Claude tips) · Content, Briefly

• 0:00
  [ on-hold music] Hey, it's Alex here. This episode of Content Briefly is brought to you by Gauge. Before I joined Superpath, I was running growth marketing at a B2B SaaS company.
• 0:17
  About a year ago, we started noticing more leads self-reporting that they found us through AI search, but all of our lead traffic from AI search was going to just one of our products. That's when we brought in Gauge.
• 0:28
  Our team had a lot on our plate.
• 0:30
  What I love most is that Gauge helped us identify the top priority actions to take, the third-party sites feeding significant citation share, and the page structure tweaks that would make a difference.
• 0:41
  With limited time spent on our end, Gauge helped one of our products move from not showing up at all to the top 15 within a few months. Real MQLs started coming in.
• 0:51
  Full transparency, Gauge was actually one of the first companies I reached out to when I started doing partnerships for Superpath.
• 0:57
  I've already referred half a dozen people to them before this became official because the product just works so well for us. Reality is, search is shifting. Fewer clicks, more AI answers.
• 1:06
  If you're not showing up when someone asks ChatGPT, "What's the best tool for X?" you're invisible to a growing chunk of your audience.
• 1:13
  Gauge is offering Superpath members a free AI visibility report, basically a 30-minute assessment of where you currently stand, plus 50% off your first three months. You can book that in the show notes below.
• 1:25
  Thanks so much and enjoy Content Briefly. Hey, and thanks for tuning in to another episode of Content Briefly. Today, Eric and I chatted about synthetic feedback. At least that's where we started.
• 1:34
  Synthetic feedback is this idea that you could create AI-powered personas that mirror your actual users, buyers, editors, skeptics, et cetera, so that you could collect feedback on your work before sending it out to peers or editors or whoever else might review content before you publish it.
• 1:50
  I've been messing around with this recently and found it to be really, really helpful, kind of in the same spirit as, like, the AI as sparring partner theme, which I feel like a lot of content marketers have embraced because it's a way to improve your work rather than outsource it or...
• 2:05
  I know that many people, including myself, still feel, like, at least somewhat hesitant to let AI just do all the work for us, but having AI make our own work better feels like an awesome use case. So we get into that.
• 2:16
  We talked about some ways to do it, and then that led us down a bit of a rabbit hole of other things we've been doing with Claude.
• 2:21
  So you'll get some tips on synthetic feedback, as well as an audio tutorial of a few other ways to utilize Claude to make your content output better.
• 2:31
  Quick reminder that there is so much cool stuff happening in Superpath right now. We have the one-on-ones and group calls, as we have had for a while. Now we also have monthly show and tells. We have member-led spaces.
• 2:42
  We have three of them running right now. These are channels in Slack where a member creates a space, everyone in Superpath Pro is invited, and it's sort of a niched-down version of even the channels we already have.
• 2:54
  And they're member-led, so they have their own events, they have their own topics, their own... kind of just their own little culture, and it's awesome.
• 3:01
  This is, like, one of the things that was on my to-do list for five years, and Alex spun it up quickly, and I think he nailed it. These have been awesome. We also have AMAs, workshops, a bunch of other things going on.
• 3:12
  So Superpath has never been busier. There's never been more opportunity to engage with your peers, learn things, offer support, get support, and all that good stuff. So superpath.co to learn more about all that.
• 3:23
  Okay, I'll get out of the way now. Enjoy this conversation with Eric and I about AI. [laughs] A lot of AI. Take care. Hey, everybody. Jimmy from Superpath and Reforge, I guess.
• 3:32
  [laughs] I'm not used to saying, not used to saying that still. Anyways, happy to be back. It's been a minute since I've been on the podcast, and good to catch up, Eric. It's nice to- It's been a nice break from you.
• 3:42
  [laughs] You're like, "Why are we doing this again?" Yeah. [laughs] I think we have a cool topic for today.
• 3:48
  We were gonna talk about synthetic feedback, which is something that honestly came across my radar, like, really, like, last week, because at Reforge, we have a suite of software products now, and we just launched a feature called Synthetic Users, where you can build synthetic AI-powered personas.
• 4:03
  The idea is, like, you can match, like, actual users and buyers and then collect feedback on your product before putting it in front of users to validate a feature that you're working on.
• 4:13
  And I had this revelation as I was working on it that, like, oh, you could do this with content too.
• 4:18
  Like, and, and in the product use case, it's like you give it a, a page, and it can actually, like, navigate the page, fill out forms, click buttons, and- Hmm...
• 4:24
  really useful for, like, you know, just kind of finding all the gaps before you waste someone's time collecting human feedback on it.
• 4:31
  But, like, so anyways, I ran some tests on this with content by giving it old articles that I'd written, and created personas that roughly match, like, what I think of as Superpath's ideal customers.
• 4:41
  And it was so fascinating to see, not just, like, catching grammar and match, trying to match, like, tone and style, but, like, giving- Yeah...
• 4:48
  this synthetic user a point of view, and then asking you to evaluate the content from that point of view was so interesting.
• 4:57
  Anyways, and, and we were talking about this in Slack, and then Eric was like, "Oh, yes, you could build a Claude skill. You could do this or that." And I was like, "Yes. Yes. That's, that's what I was talking about."
• 5:04
  It's not hard to get me excited about topics like this, yeah. Yeah. I like immediately, like, pulling up screens and screen sharing. Yeah, which I can't do on- Am, am I late on this?
• 5:13
  Like, is this something you've already been doing? I don't think you're late on this. I think, I mean, we're all kind of early on everything, right? Isn't that funny that we are- Yeah. [laughs]... but I feel late.
• 5:21
  [laughs] Yeah, yeah. Everyone feels behind on everything because there's infinite use cases now. No, I don't think you're behind. I think we've had some show and tells in the...
• 5:30
  We do these monthly AI show and tells in the Superpath community, for those who haven't been inside the community. But we've seen, it was Anton Reus was showing some.
• 5:40
  He made some really cool, they were Claude, or they were custom ChatGPT projects, I believe. But he was doing, like, evaluate this blog post as if you were so and so persona.
• 5:53
  I think his were, like, a little more, like, cinematic or, like, literary in a sense. It was like, as if you were this famous author or as if you were... So I might be[laughs] I might be hallucinating this memory.
• 6:04
  I think this is what Anton showed to use AI language. But yeah, so he had more like... It wasn't like a synthetic user like that closely.
• 6:13
  I'm sure he had that too, but it was like, "Just give me, like, some creative ideas from different vantage points."
• 6:18
  So I've seen like versions of this exist of like, I would like some sort of additional input on when I'm writing, like, creatively or whatever.
• 6:27
  And instead of doing it just by, like, going back and forth with ChatGPT, asking for line by line suggestions, it's like creating some sort of fictional editor that can give you feedback.
• 6:38
  So I think there's been, like, versions of this floated around for the last year I've seen.
• 6:42
  But I think the concept of, like, trying to recreate your end user and having that give you feedback, I think we're still early in those days. Yeah, yeah.
• 6:52
  You know, one thing, as I was testing it that I found interesting was, like, I chatted with our PMs about this, and they were like, "Make personas that aren't just gonna pat you on the back."
• 7:01
  You know, like, "Make one that's really skeptical. Make another one that's like a pretty harsh critic. Make another that's a CEO and has like two seconds to think about this at... from like a very high level."
• 7:10
  And, like, the feedback that it would provide felt very well-rounded, and it made me think, like, back in the Animals days, like, we would write blog posts, we would edit them internally, and, like, working with a great editor is amazing.
• 7:22
  Like, they absolutely make your writing better, but they also can't capture every possible point of view that is going to look at the piece when it's published.
• 7:32
  And so, like, sometimes we would publish stuff, and then we would get emails back from the newsletter being like, "Oh, you missed this," or like, you know, "You didn't really, like, fully think through this idea, and there's kind of like a little gap in the thinking here."
• 7:43
  And I would go back and, like, you know, if it was good feedback and they found something, like, that was worth updating, I would go back and update it.
• 7:51
  And I was thinking, like, how nice would it be to do that beforehand? [laughs] Like- Yeah... before the piece is live and, like, I'm editing, like, a published article based on something I missed, you know? Yeah.
• 8:02
  I think a really... I can imagine so many fun use cases there of you... Like, for example, even in Superpath, like, you'll send me an article and say, "I'm about to post this on the blog. Like, can I get your take?"
• 8:13
  And I always... I go through a few different modes as an editor of reading it. Mm. I'm like, "Okay, if I'm a freelancer reading this," like...
• 8:21
  'Cause, 'cause, you know, you have an in-house perspective, half the community is freelance. Like, would a freelancer think of this? Like, are you predicting the end of their industry or something?
• 8:28
  And, like, we don't wanna... Like, let's make sure we're not, like, cr- starting fires we didn't mean to start with- Yeah... like, one of our sub audiences.
• 8:36
  So I think that's a really good use case where if you have, like, multiple customer personas, you can say, you know, "What would the freelancer think? What would the in-house person think? What would the agency think?"
• 8:46
  And sometimes when you're writing an article, you only have, like, one brain on, and you need to be of, like, m- multiple minds. Another one I can think of there is, like, only reads the headline and only reads...
• 8:58
  Like, who's... Like, what's- Yeah, yeah... the really reactionary, like, if they don't do a deep read, like, what are you accidentally implying that you didn't mean to say? Yeah.
• 9:07
  Like, recently you sent me the article you wrote about the... I think your original title was, like, "Blogging Is Dead And We All Killed It" or something. [laughs] Yeah.
• 9:15
  But, like, your, your actual point in the article is kind of optimistic about... I mean, I'm paraphrasing a very long article.
• 9:21
  But I, I think it was basically like, it's okay that people aren't deep reading all kinds of content, and there's other things you can do, and people should go read that article.
• 9:29
  But I think when I read the headline only, I was like, "I know exactly how people are gonna react to this." Yeah. You know, let's reframe that way.
• 9:35
  So if you could build, like, different checks and balances in for yourself with all these sort of synthetic users or feedback givers. Totally.
• 9:45
  You actually could make an AI Eric persona and just give me the markdown file, and it would save you a bunch of time. Yeah, I wouldn't have to do anything anymore. [laughs] Yeah.
• 9:56
  [laughs] I was thinking also- Just say no. That... It's one line of instruction. It says, "Just say no to all Jim..." No, we don't need another Zap. There's, like, [laughs] one. It's the only line.
• 10:05
  Alex changed the password. I, I can't mess it up anymore. Yeah, that's good. [laughs] The best thing that's ever happened to Superpath, yeah. [laughs] I was thinking too, like, where this type of work fits in.
• 10:15
  Like, the, to me... Or stepping back for one second, I feel like there's so many ideas out there floating around about AI and, like, some of them are exciting, some of them are not so exciting. And, like, what...
• 10:25
  Something, to me, something like this, like, it, it fits into, like, the sparring partner mentality that I know- Mm-hmm... a lot of writers kinda lean into when they're working with AI, like, in a creative way.
• 10:36
  And, like, this is an unlock to me. Like, this makes my life... No, actually, I was gonna say it makes my life easier. It doesn't make my life easier, but it makes my work better, which is worthwhile. Yeah.
• 10:44
  It makes your life harder in the best possible way because you're not just- Yeah... shipping content sort of willy-nilly. You've got, like, proper feedback and checks and balances built in. Totally.
• 10:55
  And, like, I, when I write a draft of an article for Superpath, I send it to you, I send it to Alex, I send it to Keisha, like, and you guys give me really good feedback.
• 11:02
  Like, and I recognize, like, not everyone has three, like, deeply experienced content marketers who, like, understand the field extremely well that can provide that type of feedback, like, very quickly. Yeah.
• 11:13
  And we're the audience, and we're content marketers, right? There's like two bonuses built in there that, like, we can- Yeah...
• 11:19
  we can react well to content, and then we also know the subject matter, which you almost never have those two things available to you as a content marketer. Right. I think now, though, w- this is my thinking.
• 11:31
  Like, where this fits into the process is collecting synthetic feedback first is a really nice thing to do to your editors because, like, the last- Yeah...
• 11:41
  draft I sent you, like, you pointed out some things that were like, "Duh. Why is this in here?" [laughs] Like, this is...
• 11:45
  You know, like, I find for myself, like, I sort of get to a point in a draft where I'm like, "I don't wanna look at this anymore." Like, I just... I know it's not, I know it's not finished.
• 11:53
  Like, I'm sick of looking at it.
• 11:54
  I need someone else to just, like, tell me what to do, and that's, like, the point when I drop it in Slack for you and Alex and Keisha to look at.Be like, it actually is sort of a disservice to have something like synthetic feedback available and then ask you for your time- [laughs]...
• 12:10
  to clean up some of the really basic stuff, you know? Like, it's more useful for me to say like, "Hey, Eric, is the hypothesis of this article strong? Like, is this worth writing?"
• 12:19
  You know, and then later, like, "Could you just help me polish this up a little bit and make sure there's not any, like, fine-tuning that I'm missing?"
• 12:27
  And, like, leave that middle chunk for AI to deal with, because it's somewhat tedious to read through an article, then have three people read through an article, only to point out stuff that's like not, it's not rocket science necessarily.
• 12:39
  Yeah. That makes a lot of sense to me. Yeah. And you're going before the f- full final draft. Yeah, it's almost, it reminds me of, like, when you as a content marketer get edited to ask AI slop.
• 12:49
  [laughs] Like someone writes something and they send it over and they say, "Hey, I wrote this. Can you take a quick look?" And you're like, "Okay, well, you" [laughs]
• 12:55
  like, "You just wrote this with AI quickly and now you're asking me to edit it." It's kind of like that you wouldn't want that same, like, disservice for you, that I can feel that.
• 13:03
  Like, yeah, I can see the benefits of that. I don't have an editor at Doc ever. Oh, interesting. Yeah. I'm just fully flying solo, right? So for me, it's extremely useful.
• 13:12
  I might get in the little bit in the weeds here, but I, like, I built myself something like this that I, I could imagine a lot of people are listening like, "Cool. Yeah, duh, this is obvious.
• 13:21
  How do I actually go about making one that's somewhat useful?" So I'd love to know how you built yours, but I built something at Doc, actually. I got the idea because I saw, like, a state of...
• 13:32
  So our main customer profile is s- a sales enablement leader at Doc. And so I had read, like, a state of sales enablement.
• 13:41
  And it was one of these, like, 100-page documents where it was like, you know, they really care about their content that they make being discovered by sales reps.
• 13:49
  They care about, you know, ROI, like, being able to prove the ROI. Like, it was very research heavy. And so I put that into Claude and said, like, "Pull me," you know, "Do an analysis of this research report.
• 14:02
  Tell me, like, what are the 15 things that a sales enablement leader cares most about?"
• 14:07
  And sort of built my ICP that way, because I felt, like, one of the challenges I have when I'm writing something is like, who will care about this? Like, it's one thing to do kind of thought...
• 14:16
  Or sorry, not who will care about it, but, like, will they actually care about the points I'm trying to make, and, like, does this resonate with their pain points day to day? Hmm. Yeah.
• 14:23
  'Cause I feel like so often, you know, we're trying to push narratives like why you should care more about buyer enablement and, 'cause our product helped with that.
• 14:31
  And, you know, we were trying to sort of push a narrative. And when I was looking at this data, I was like, "Okay, they don't care about what I'm talking about."
• 14:39
  [laughs] And like, "How do I connect what I wanna talk about to the things they care about?" Right?
• 14:43
  And that step is very abstract and high level and hard to do when you're trying to write 10 blog posts a week or whatever it is now that people are trying to do. So- Yeah...
• 14:51
  now what I did is I took this report of, "Okay, here's what people said they do care about." I think I actually found a couple and sort of threw them all together, and then built a persona that way.
• 15:03
  And then what I did is I built a Claude skill for it. So if, if people haven't used Claude skills yet, basically it's like a little skill that you can call in the middle of any chat.
• 15:13
  So it's similar to a custom GPT or a project, but you can just sort of say, "What would my audience think about this?" And it will just pull this skill into the chat that I'm using right now.
• 15:25
  So it's really nice because I can be writing a LinkedIn post, I can be writing a long blog post and then just call this skill, and then it will give me feedback. And I've given it a few ideas on how to give me feedback.
• 15:36
  So that was like I was trying to take a data-based approach.
• 15:39
  I didn't wanna, like, make up what I think the audience would care about and have Claude repeat that back to me, 'cause then I'm just going through the same mental checks and balances that I was already sort of internalized while I was writing it.
• 15:51
  Yeah, yeah. So, and I found that's been extremely effective. It's like, "Oh, remember, like, they really only care about this stuff." [laughs] I'm like, "Oh, okay. Yeah, you're right. Sorry." Yeah, yeah. That's cool.
• 16:00
  I mean, I guess I did kind of a similar thing where I kind of tried to, like, recreate the feature, the synthetic user feature from Reforge in Claude.
• 16:08
  And one thing I find myself doing with Claude a lot lately is instead of like, instead of thinking through all the steps and then asking Claude to do each step, I just say like, "Here's my, like, high level idea.
• 16:17
  You tell me the steps." Mm-hmm. So then the steps are like, "Okay, let's talk about your personas. Who are the people that you want to read this article? Tell me all about their point..."
• 16:26
  So we, like, created three different personas.
• 16:28
  One is like a professional freelancer, someone who, like, does this full time, you know, wants to grow their business, like, wants to like, you know, whatever, stay on top of all the trends in content and understand what their customers want, et cetera.
• 16:40
  The next is in-house content manager.
• 16:42
  Someone who's kinda like in the middle of their career, working in-house, you know, wants the professional development, wants to stay on top of AI stuff, and also wants to, like, further their career in-house.
• 16:50
  And then another persona for, like, VP/CMO, someone who's, like, thinking about content, like, from only, like, a very strategic level, like, not really in the weeds. Mm-hmm.
• 17:00
  And I included that one because I feel like all content should be, not all, but a lot of content I think should be, like, somewhat aspirational. You know what I mean? Yeah.
• 17:07
  Like you sort of like write it for your persona's next job title rather than- Totally... their current, you know? Yeah. And so then we created those. Claude says save these as markdown files.
• 17:17
  I didn't build a skill yet, but I think that's the next step.
• 17:19
  So we save those as three markdown files, and then I ask Claude to, like, help me, like, write a prompt, so that any time I have an article for you to review, we're gonna rely on the same prompt.
• 17:27
  Use these same three personas, and then l- you make a table and tell me, you know, for each person- Mm. Very cool... tell me, like, what do they think, and then, like, their recommended updates to the article.
• 17:37
  So I did all of that for the recent, like, you know, how to publish content without readers or how to have influence without readers or traffic article. Mm-hmm. And the feedback was, like, spot on.
• 17:46
  There is-- I feel like this is the point where this either works or breaks. Like, if you just listen to everything it says, you will make your article worse. But if you- [laughs]...
• 17:55
  then, like, turn your brain back on, which I find, like, when I'm working with AI, my brain is, like, kind of off, and then when I, like, need to, like, have editorial judgment, I have to, like, wake it back up again because, like, now it's my turn.Mm-hmm.
• 18:08
  And so I like picked through the recommendations and like some were really good, some were like, yeah, that probably ought to be in here, and some were like, eh.
• 18:16
  Not like full on hallucinating, but just like m- totally misses the point.
• 18:20
  And so I think then the next step would be like refine the prompt we're using, refine the persona so that the output gets better and better each time. Yeah, absolutely.
• 18:28
  But then I guess what you said was make a skill, so then all you have to do is like you have all that ready to go, and then you say, "Hey, review this article." Yeah.
• 18:35
  The skill just saves you the step of pacing the prompt again. So it's-- like it's just sort of built. You make it built into Claude. Yeah. So I mean, whatever, I'll save you 30 seconds each time you're using it [laughs].
• 18:46
  Yeah, yeah. I think that's important.
• 18:49
  I feel like actually formalizing it as a skill is an important part of building a workflow, refining it to the point where you're confident in it, and then it's a formalization of a process when it becomes a Claude skill.
• 19:00
  Like- Yeah...
• 19:01
  at that point, you feel confident enough to like quote unquote "ship it" so that you can rely on it in the future, versus like if you kind of do what I did and just sort of like leave it a little messy, like, I don't know, I feel like you-you're prob- like in that case, like I'm not quite confident enough in this process to formalize it, and I think the right thing to do would be make it better so that I can do that.
• 19:19
  Yeah. And not for this whole conversation to be a Claude tutorial, but the nice thing about a skill too is that you can sort of chat with Claude quite natively to get it to update the skill.
• 19:29
  So I can kind of pause like midstream of editing an article and say, "Hmm, I didn't quite like the output I got from the skill. I would prefer if it did it like this. Can you kind of stop what you're doing now?"
• 19:41
  Or you can just like open a new chat window and, you know, not have to like context switch, but and then say, "I would like you to update the skill in this way," and it can just pull up the skill, edit the markdown file, and like save it back to Claude.
• 19:54
  Mm. So there- That's slick... you can do, you can like very easily do the iterative feedback as you're going and not have to copy paste prompts and stuff like that. That's pretty slick.
• 20:03
  You know what, while we're talking about Claude, can I share one other cool Claude thing I found recently? Totally.
• 20:09
  So we've been going through this series of product launches for like some pretty core features of the SaaS tools that Reforge is building. And so like for each one, I create a, a new chat in Claude.
• 20:19
  I load it up with a bunch of context very specific to that feature, and then all that lives within a Claude project, which has a bunch of other-- I sort of consider it like global context and then like project context, you know.
• 20:30
  Mm-hmm. All the kind of bigger context about the company and goals and, you know, whatever, voice tone, things like that.
• 20:36
  The annoyance I was having though was like, I was kind of like doing this from scratch each time, where this Claude thread just became like I could not live without it, and also it was just becoming so unwieldy.
• 20:47
  And I had this realization that like, this is dumb. I mean, this is helping me a lot, but it's also like getting to be quite cumbersome. So I was like, I'll just ask Claude like what is the solution? Yeah.
• 20:56
  And Claude's solution was awesome actually. Claude's solution was, let's build a template in Notion, since that's what we're using for all of our like docs and commenting and feedback and stuff like that.
• 21:07
  We'll build in there, so anytime you need to work a new product launch, you tell me, we'll copy that, start a new one.
• 21:13
  Claude will fill it out with all the important information, and then leave a series of questions for me to answer about like target audience, you know, kind of like the hook for the launch.
• 21:24
  Like kind of like what's the thing that's gonna get people most interested?
• 21:26
  It included like a little messaging and positioning exercise for each one that we kind of go back and forth on to like find like what are the key points we wanna highlight from this.
• 21:35
  But then to me, the coolest thing actually is that instead of the output of Claude stuff going back in Claude, it all goes in Notion. Hmm.
• 21:44
  So then once we're working on the template, I say whatever to Claude, and it just says, "Okay, check Notion." And then I go in Notion, and it's there.
• 21:52
  And then like as I make edits, it will recognize that I've made edits, and obviously like, "You got those edits?" And it says, "Yeah." So like, "Here's what you did, and here's what I learned from it."
• 22:01
  So that the thread becomes really short and very easy to kind of scroll back on if I need to, and then like the work is happening in Notion, which is like a, even though I'm not like a giant Notion fan, like at least- Mm-hmm...
• 22:10
  it's a document editor, and it's just like much easier to like get in there and kind of work in a way that I feel much more accustomed to doing. Yeah. That's very cool.
• 22:20
  And I mean, yeah, now we're just fully in AI talk, but I think that bubbles up into advice for pretty much anything you're trying to do with content and AI is like start by asking, instead of trying to write a prompt yourself, like go into Claude or ChatGPT and say, "Here's what I'm trying to do.
• 22:37
  Here's all the inputs I have. Here's the output I'm trying to achieve." Be like, "How should I do this?" And it will give you ideas.
• 22:44
  And sometimes you're like, "Hmm," be like, "Yeah, well, I don't really wanna use Zapier for this one. Like, can we just keep it in all in Claude?" And it'll say, "Okay, here's some ideas."
• 22:52
  And like that will almost always give you a better first draft than trying to do it yourself, is what I found. Yeah. And then- Totally... yeah, and you try it a few times, see what your actual outputs look like.
• 23:01
  Like do some actual projects with it, and then the first five, six times you'll have to go back and tinker with the prompt, and then eventually, like you should pretty quickly arrive at something.
• 23:11
  And again, tell it what you're not happy with the output, and it will write a better prompt than you will. 100%. Actually, that reminds me.
• 23:18
  One of the other things that Claude told me to do was to connect Claude to Linear- Mm-hmm... which is the tool that our PMs use to-- It's like project management for PMs, basically.
• 23:30
  Which I did, so I authenticated it, and now all the contacts that I was like, you know, downloading as PDFs or taking screenshots of Slack conversations, like so much of that now is just me saying like, you know, "Hey, check Linear for all the comments and documentation on this feature to get me up to date really quickly."
• 23:50
  And then- Mm-hmm...
• 23:51
  actually, even beyond that, we installed a Linear Slack bot, so if you need like quick answers, you can always just be like, @Linear in Slack, like, "Tell me about this feature," and it will give me like a quick overview and then tell me like, "Give me links to all the places to go to learn more," which is just like a huge time saver.
• 24:06
  Yeah. Very cool. Very cool.Well, we've gone deep down the Claude rabbit hole. We were talking about [chuckles] synthetic users at some point. Yeah, yeah. Yeah.
• 24:14
  But actually, I was thinking too, there's so many applications of that 'cause there's like different personas, there's different levels of reader.
• 24:22
  You could even have like-- So, and by levels of reader I mean, like you have the skimmer, you have the deep analyzer, like there's different ways that someone might consume your content, and it's interesting to...
• 24:33
  I always think about when I'm writing to write for each of those levels of reader. There's, you could think about like funnel depth. Mm. You know, like- Yeah... like analyze this.
• 24:44
  Like, okay, what if somebody's already interested in our product and they're reading this?
• 24:48
  Or, you know, you could either be trying to write it for multiple levels of the funnel at once, or you're like, "Hey, this was supposed to be a bottom funnel article. Why are we talking about so and so?"
• 24:57
  So I think there's so many different-- there's the synthetic user thing, but I think there's also lots of other kinds of modes of feedback.
• 25:04
  And obviously you don't wanna have 10 editors reading each article the every time. But I think there's, for different-- Like in my case at Doc, like I kinda only have like one main persona in mind.
• 25:16
  So I think for me it would be the sort of synthetic editor reader would have different modes than that. Yeah. Yeah. You know, I'm curious what you think about this.
• 25:25
  I find myself-- Like I'm still kinda relatively new back in an in-house role, and I find myself kinda like doing the job and also learning how to do the job simultaneously. Mm-hmm.
• 25:36
  The learning how to do the job part would have been actually I think a lot easier in the pre-AI era, whereas like now I'm like constantly trying to figure out like how do I get Claude or whatever AI tool to help me do this better, you know, kinda like unlock efficiency, maybe even like create a whole new workflow I wasn't thinking about, while also like just trying to do the thing I'm supposed to be doing.
• 25:57
  And I sense that you're beyond that at this point. Like, that you've kinda like settled into a rhythm where you've like extracted maybe not every ounce of value that AI can offer Doc's marketing, but you know.
• 26:07
  Like to a point where like you don't have to think about the two tracks anymore. Yeah. Are you thinking more as like a subject matter sense, or like in terms of like industry knowledge?
• 26:17
  Or are you asking from the sort of like day-to-day marketing tactical sense?
• 26:22
  I think it's like, maybe to clarify, it's like I feel like I am adopting AI as a marketer at the same time that I am marketing, and so I kinda, like earlier in the episode- Right...
• 26:33
  we sort of talked about like the feeling like you're behind. Like I sorta always feel like I'm behind even though like- Yeah... I think like objectively I'm not.
• 26:39
  Like I feel like I'm like somewhat sophisticated with this stuff. And so I feel like my brain is always like, "Okay, I have this job to do. How is AI going to help me do it better or faster or whatever?"
• 26:49
  You know, so there's like two jobs- Yeah... happening simultaneously. Yeah. Okay, I get it.
• 26:54
  Yeah, I think even right before we s- hit record I was saying I feel like I've really hit a groove now where I'm comfortable understanding what AI can do for my workflows, and like at a point where I've like lost the anxiety about it, where- Mm...
• 27:10
  I think a year ago I was like we had a whole episode [chuckles] where I was like AI's just making me anxious because I feel like I don't understand what the expectations are for me in my role now.
• 27:19
  Like w-what does, uh, what is the right amount of output look like for me as a content marketer- Yeah... with AI? Like how much am I supposed to understand the voice of customer, things like that.
• 27:31
  'Cause I think there are kind of different expectations now.
• 27:34
  I'm sort of sidestepping your question, but just for example, I think when you used to start at a company like you are now, there was this sense that like it's gonna take you six months to understand the product, the customer, all of that.
• 27:48
  Like you'll shadow some customer calls, you'll probably do a few case studies that sort of click in your mind of, you know, the value of your product, but it was sort of accepted that would be like a natural, slow, organic process.
• 28:02
  And versus now it's like, oh yeah, you have to make sure you get all your sales calls, Gong recordings, and you connect them into a big spreadsheet, and you get AI to analyze it, and like you should be infusing all these insights into your brain every single day.
• 28:16
  And like th-that, that feels like overambitious and- Mm. Yeah... not the way things will actually shake out.
• 28:24
  But I feel like if I were starting in a job now, I would feel like I had that pressure to like make everything happen faster with AI. And so I think like the world will shake out somewhere between those two.
• 28:35
  Like before it was, yes, it was like slow, but there was some value to doing things manually and like you can't just inject all this knowledge instantly into your brain with AI now.
• 28:44
  So I think like I can understand if I was in a new role right now trying to figure out my workflows, trying to figure out AI, trying to figure out the customer, all of that, that it would, if-- it would feel like I'm learning too much at once.
• 28:55
  'Cause you're like, yeah, you're trying to learn how to be a marketer again, and you're trying to learn the company, and you're trying to learn the customer. That's a lot at once, so I empathize with you.
• 29:03
  It's a lot at once. Yeah. And then like every two days there's like a new AI thing and I'm like, "Should I do that?" Like should I set up Claude Code? What about like there's like a Claude CoWork now?
• 29:12
  Like are all my integrations set up with the tools that I use? Yeah. And then like next week it's a new thing and I'm like, "Should I just do the work I'm supposed to be doing?"
• 29:20
  'Cause it's like, you know, I need to get it done this week. Or like should I carve out an hour or two and learn? It, it's a little...
• 29:26
  I'm finding it a bit tough to balance, even though like as we've discussed on this pod, I love tinkering with this stuff. It's like I really- Yeah... enjoy that part of it.
• 29:34
  Yeah, I think a good thing to remember is like every other marketer sitting at every other company is going through the same things you are, and as long as you're, you know, keeping up to date on what's possible.
• 29:46
  I think like it's more important to understand what's possible than to feel like you have to master everything at once, right? Like- Yeah...
• 29:53
  when you come across a new task that is, like a new, say, a new marketing channel, like we just relaunched our podcast at Doc, and it's been sort of two years since...
• 30:03
  Like the first time we had it, like AI wasn't as deeply embedded in my marketing workflow, so, you know, I could think through all the tasks that had to be done and I had to bring on people to do them all.
• 30:12
  And now I could kinda look at the whole workflow and say, "Okay, like I can see naturally where AI fits into this now."And then I'll spend, you know, a week trying to figure out how to actually do it.
• 30:21
  But, uh, I think the anxiety before was like I couldn't see... Like, if you're too in the weeds with like how you're gonna build the prompt and how that's gonna connect to this, then it's easy to get stuck.
• 30:33
  But if you can sort of look at the high level process and think like, "Okay, now I know enough about AI that I know where it will be helpful and where it won't be," that I feel less anxious about, like, that tinkering time, if that makes sense.
• 30:43
  Yeah. Yeah. That does make sense.
• 30:45
  You know, actually, as I was building, my- the first Claude skill I made was I set up the linear integration in Claude, and then created a Claude skill to, like, basically go look and find what I needed for the feature that I'm working on.
• 30:56
  And I got a little lost creating it, and I was like, "I know Eric's good at this." And I actually went back and watched the show and tell you did, and like I watched you build one. I was like, "Ah, perfect."
• 31:05
  So thank you. [chuckles] Oh, you're welcome. It's, it's funny like being in this seat and being like- [chuckles]... I'm a content marketer in need of a little help. Oh, Superpath. Superpath's a great resource, you know?
• 31:13
  [chuckles] How the turns have tabled. [both laughing] I don't remember what movie that was from. Cool. I don't know what we're gonna call this episode because I feel like it was a bit wide-ranging.
• 31:21
  I guess I'll just ask Claude to give me a couple of ideas. And I kinda like this format.
• 31:26
  To be completely honest though, like I find this to be a little cathartic, and if other people do too, I could imagine some other episodes in the same vein where it's like hitting on a couple of different topics rather than sticking to just one thing, because there is so much going on right now.
• 31:39
  And like you said, I'm sure a lot of other people are feeling the... Pressure's not the right word. The FOMO.
• 31:44
  Whatever it is, it's kinda like sort of, uh, we're hitting the gas pedal like all the time on, like, probably more than one thing. Yeah.
• 31:51
  Well, this episode was like the, a perfect microcosm of like let's just talk about this one AI use case, and then like- Yeah... yeah, look at it bubble up into this, like, massive high level anxiety about using AI.
• 32:04
  Like I just feel that, like, cycle happen often, and hopefully we'll, we've like addressed it now. Yeah. Now everyone can sleep easy. Well, if you want more. Yeah. Yeah.
• 32:13
  If you want more, if you want support from your peers, professional development, training, show and tells, one-on-one calls, group calls, then you should join Superpath.
• 32:22
  [chuckles] Alex has so much cool stuff going right now.
• 32:24
  Like I actually, I just posted on LinkedIn the other day a screenshot of your announcement post in Slack from last week, because I saw it and I was like, "Holy crap. There's so much happening."
• 32:33
  Like, one of the things he told me as we were going through the acquisition process was like, he was saying like, "I wanna increase programming.
• 32:38
  I just want there to be, like, more opportunities for people to engage, to connect with each other, to learn, et cetera," and like that's the first thing he's done.
• 32:45
  There's webinars, workshops, AMAs, one-on-one calls, group calls happening, like, in most cases, like twice a week. So that's the call to action. It's like come hang out, come...
• 32:56
  It's 30-day free trial too, so you can come check it all out and see if it's right for you. Yeah.
• 32:59
  And I think what's beautiful about what's happened recently too is it kinda used to be me and you running things with some amount of capacity of like how many group calls can I host a week, how many- Yeah...
• 33:11
  webinars could you host a month? And now we've really, like, activated the community to host more events and spaces.
• 33:18
  So yeah, now we have like a great freelancer sort of subspace where there's monthly events specifically for freelancers.
• 33:26
  We've got like a little, like, culture club where people are just talking about, like, non-work things, and it's just a nice space for people to do that.
• 33:32
  And then there's like a content op space, and there's a sort of content leader judgment space for, like, what about these high level scenarios that you might encounter as a content leader, and how do you get better at navigating them?
• 33:46
  So, like, now the community itself is activated into all these little subspaces where it's actually members running these spaces, and it's been super cool. Really cool to see.
• 33:55
  It's like watching a child grow up, go to university, and like doesn't- [laughs]... doesn't need us anymore. [laughs] That's so funny. It's a beautiful thing. It really is.
• 34:03
  Like, that was something that was like on my list literally for five years, and I just like never quite figured out how to do it. And then Alex saw that opportunity and was like, "Let's just do it."
• 34:10
  Yeah, it's been awesome. And he did it. And it's awesome to have members leading because, like, they bring their own perspective, their own points of view, you know? Like, that's great.
• 34:17
  You don't wanna just hear from me and you all the time, you know?
• 34:20
  It's, like, really nice to have people, like, kind of create a space that, like, accurately reflects, like, their own experience and interests and all that. So anyways, superpath.co. Come check it out.
• 34:31
  And otherwise, thanks, Eric. It's good to catch up, and if people like this episode or hate this episode, email us. We still haven't gotten an email, so we're waiting for the first one. Yeah.
• 34:38
  [chuckles] podcast@superpath.co. [laughs] Yeah. Or if you really hate this episode, it's alex@superpath.co. Send it all. Yeah, yeah, complain to Alex. [chuckles] Yeah, complain to him. Yeah. [laughs] Thanks, everyone.
• 34:49
  Awesome. Thanks, everybody. Take care. [outro music]