Building Smarter Systems as a Creator with Andrew Wilder and Miranda Wicker

If you've been listening to the podcast for a while, you know I'm always talking about how creators can better understand where traffic comes from and how people experience their brand. We've been talking about that a lot lately, brand. And the team at Yoast, which is the SEO plugin many of us, I would say most of us already use, actually just rolled out 2 updates that make those things so much easier. First, Yoast SEO Premium, so that's one plugin, the Yoast SEO Premium plugin, it now brings Site Kit by Google Analytics, Right inside of Yoast. So for those who aren't familiar, Site Kit is Google's official WordPress plugin for Analytics and Search Console. And this integration makes your data— this is a wonderful word— simple. There's no extra logins and no bouncing between dashboards. It's just clean, actionable numbers, easily accessible right within Yoast. So that's the Yoast SEO Premium plugin. Wonderful plugin. It's one that we use on all of our sites. But second, And also really important is this concept of brand and specifically brand as it relates to AI. So a new feature with the Yoast SEO AI+ plugin, so this is a different plugin, is the Yoast AI Brand Insights. So this scans how your brand shows up in ChatGPT and Perplexity. You can see where you're mentioned in AI-generated answers. You can track sentiment and also compare your site's visibility with other publishers. To better understand how AI interprets your brand. Wonderful updates and wonderful features in a set of plugins that are really powerful and are needed for us as publishers to both understand our traffic and our analytics and our numbers, but also understand some of these brand considerations on how our site is showing up in AI. So you can upgrade or download and get 10% off using FOODBloggerPRO10%. So that is 1, 0, and the percent sign. And that's to get 10% off of Yoast SEO Premium, WooCommerce SEO for any of the e-commerce folks out there, and Yoast SEO AI+ plugins.

1, 0, 3, 0, 2, 2.

Welcome back to another episode of the Food Blogger Pro podcast. I'm Ann from the Food Blogger Pro team, and in this episode, Bjork is sitting down with Andrew Wilder and Miranda Wicker from NerdPress for a roundtable on staying human in an increasingly automated world. Andrew shares how NerdPress grew from a WordPress support side gig into a full team serving hundreds of publisher sites, while Miranda reflects on her own path from teacher and parent blogger To communications manager, and together they explore how creators can rethink how they work, what they value, and where their voice still matters most. From there, they dig into the difference between using AI as a creator and using it as a tool. Andrew and Miranda share real examples of memory, skills, scheduled tasks, and guardrailed connections, including NerdPress's own guided agent connector, and make the case that these tools can help you work smarter, but only if you stay intentional and keep enough oversight to protect what makes your work yours. And now without further ado, I'll let Björk take it away.

Andrew and Miranda, welcome to the podcast.

Hello.

Hi, thanks for having us.

Yeah, this is fun. It's always fun to have 2 guests. It's a little bit of a roundtable. We're going to be talking about all things AI and content. I would say the 2 things that people want to talk about most on this podcast or want us to talk about most as of late is AI, and then all of us that listen to this have content businesses. And that's what we're going to be talking about today. Before we do that, uh, Andrew, you've been on the podcast multiple times, like probably top 3 guests that we've ever had in terms of appearances on the podcast. Uh, but for those who haven't connected with you, do you want to share a little bit of your story and the background of NerdPress and kind of what your day-to-day looks like right now? Sure.

Um, so I got my start in, in this realm as this career, this version of my career in 2010, when I started a food blog and Um, I had already, I had done website stuff in the, in the past, so I knew what I was doing technically. And when other food bloggers found out I knew what I was doing, they asked me for help. And long story short, I started doing work on food blogs and I started support subscriptions and fast forward 16 years later.

Amazing.

Um, NerdPress, we've grown into NerdPress. We are a WordPress maintenance and support company. Uh, we're 20 people all in the US, uh, supporting hundreds of publishers' sites. Um, most of them are food bloggers, so Of course, it feels like coming home every time I talk to you.

Yeah.

And, you know, I'm also a Food Blogger Pro Expert. And so I'll pop into the forums every so, so often. Um, and, uh, yeah, we've been, uh, just supporting food bloggers and content creators for a decade and a half now. And, um, I think the, the most exciting thing for me, uh, is I picked this path, like in 2010, I was wearing 5 different hats and I picked like supporting websites and tech Um, as like the one thing I wanted to focus on because it was always going to be changing and therefore I'd always be learning and interested. Um, because by then I felt like I had said everything I needed to say about healthy eating. Um, but tech is always evolving, right?

Yeah. And it's never been more quickly evolving than today.

And now it's kind of like, now I'm feeling like, be careful what you wish for.

Sure.

Yeah. You got it. Never changing. Here we are.

You know, 2026 is definitely, I feel like we are in the middle of the S curve of a main, you know, huge technology change. Like, and maybe it's not the middle of the curve, but it feels like we're in the steepest part of that curve right now where everything is just like shifting. And I don't know, my, my spidey sense is this is going to be the year where we look back and go, okay, that was the year everything changed.

Yeah. Um, it's amazing. Yeah. You and I have both been doing this for a really long time. It's, it's fun to look back, you know, I can remember having conversations when you'd say you're a team of 20, having a conversation around your first hire.

Yeah.

We're talking about like, okay, what does it look like to hire? And it's amazing that you fast forward now, you have a team of 20. A huge reason why is because you all do such a wonderful job supporting creators with one of the things that the vast majority of creators, I would say 95% of the people that we talk to, don't want to do, which is like the technical stuff related to running a website. And you guys not only do that, You do it well, um, and have done it for a long time. Uh, Miranda, you are one of those 20 employees that has joined the NerdPress team. You have a background as a creator yourself, uh, started publishing content on a blog before Andrew and I did, which doesn't happen a lot that we have these conversations that, you know, somebody started publishing online pre-2010. It sounds like you started in 2008. Tell us a little bit about your story.

I mean, if we go all the way back to LiveJournal, then it's been a very long time. So we, we don't have to bring that, those days up. But, um, I was a teacher and, uh, taught high school English. You know, I love books and, and, and all of that stuff. So I loved what I did and then I had kids and then I became a mom with a blog. Um, I needed a place to write about parenting and postpartum depression and anxiety specifically.

Yeah.

which I was dealing with after my son was born. And what I found through blogging was this really strong sense of community around— it was like the first time that I think, um, I knew there was like that, that me too moment. You know, you find that, oh, oh, somebody else out there is experiencing this too. And that was like in that, in that blowup of Twitter was new and Instagram hadn't even been invented yet. I tell my kids, I'm like, I predate Google, guys.

Yeah.

Like, they can't believe it. Facebook wasn't a thing. You had to have a college email address.

Yeah.

Um, and so I really found like a, like a really strong sense of community. And from that blog, I left the classroom, started freelance writing, building communities for brands, and kind of pivoted into social media marketing and content marketing with a background in the strength of communities for growing a brand, growing a business. Um, and, and growing one another. Like, how do we collaboratively do this? I'm a big believer that the internet is not like, it's not finite. Resources here are not finite. There's enough to go around. And when we learn how to share that with one another, we can magnify it for everybody. Uh, and so I joined NerdPress in December 2024 as communications manager. And now I'm our communications manager and beekeeper slash beekeeper slash, uh, What does Andrew need me to do today? Um, keeping Andrew on task. That's that, you know, I keep Andrew, uh, on task as much as I can with the marketing and communication stuff. Um, but I am just, I really love working with creators. I really, I really enjoy, uh, specifically as a woman, I love how many women have been able to find a career and that, that people didn't, they've been able to support their families in ways that nobody thought possible. Nobody saw this coming. And then, and then women have been able, I mean, I think.

Yeah.

I think women really created the influencer and creator economy. Um, and, and I'm proud to be a part of that. So I love seeing where it's gone and I can't wait to see where it goes next.

Yeah. You both have such a deep, uh, pool of experience and a lot of knowledge to draw on, not only because of your personal involvement as creators, each of you have this origin story, but now also you have Hundreds, thousands of touchpoints to other creators. And so you have a really good beat on not only what the arc of the industry has looked like, but also what the present pulse of the creator is because you so intimately work with these creators and probably see them in their most like joyous and most fearful moments. And Andrew, I want to hear from you a little bit. As you, if you were to kind of summarize what the current state of the industry is, kind of the average of the conversations that you're having right now, how would you summarize it?

Cautious optimism.

Hmm.

Um, I feel like the, I feel like people aren't operating out of panic anymore. And maybe that's me. I'm not as panicky as I was even 6 months ago about what's going to happen with AI, right? Um, but I think, um, you know, 2, 2 and a half years ago, I was really worried, genuinely worried that websites would not exist, blogs would not exist. They were all going to be replaced by some sort of AI thing where websites were just created dynamically by AI. Um, that was 2 years ago, 2 and a half years ago, I think.

I don't even know if that was 2 years ago.

Like I, But guess what? Uh, rumors of blogging's demise have been greatly exaggerated, right? We're still here. Now, a lot of publishers have not fared well. You know, I don't want to discount that. Um, but a lot of that, especially in the food space, was actually, I think, a correction after the pandemic. It wasn't really about AI. Um, the helpful content update was really separate from AI. So like, it's been turbulent and challenging. Um, cause we kind of reached peak food content. during the pandemic, um, when everybody started a blog and stayed home and cooked, right?

Mm-hmm.

Um, and so, um, that's not been easy. It's been very hard. And a lot of people have given up and shut down or sold their sites and left. Um, but blogging in general and food blogging in general, um, many publishers are still doing very well. I know a lot of publishers whose traffic is up 20% year over year. Um, so, um, there is a lot of hope for There's a lot of reason to be optimistic now. I think also the biggest reason I'm optimistic is because people have realized how crappy AI content creation is.

Sure.

And I don't mean creators have figured that out. I mean, the general public has figured that out. Creators knew it from the instant, you know, but now like there is this hatred and anger towards anything that's not created by human. By, you know, Miranda, how old are your kids? How much do they love AI?

17 and 14 and they hate, well, I mean, it depends on what we're talking about. My son despises it. He absolutely hates it. My daughter understands that like sometimes there might be a use case for this if I am stuck on a math problem and I can't get, you know, like there, she has moments, but mostly, especially for creation purposes, oh, she hates it.

And it's, it's soulless, right? Um, and you can tell, especially written. content. Um, and I think the reason I'm optimistic is that the general public is recognizing that and craving that human connection. And, you know, we've been saying for years, like, be authentic online, right? Um, you know, be uniquely you and have your unique perspective. Like, we've talked about that, I'm sure, in our many, many conversations. And like on the SEO for Bloggers episodes, we've talked about it a lot. And I think, um, that's Now more than ever, so valuable and appreciated. And so that's really like the people who've been doing that well and authentically and showing up as them, that is really more valuable than ever.

Yeah. In one of the things I think a lot about within the context of marketing, a lot of what we are doing as creators is marketing. I know Miranda, you have kind of a marketing hat that you wear within the context of your role as well. One of the things that we talk about is this idea of novelty and novelty as it relates to marketing. And if you've seen something 10,000 times, if something isn't novel, if something is rote, like it won't be as compelling, it won't be as effective. And I think as content scales up in terms of its ability to be created through AI, one of the ways that we can be novel is to create content that doesn't seem like AI. And I think one of the common rebuttals will be, yeah, but AI will get better. AI will be able to sound more natural. AI will be able to sound more interesting. And my rebuttal to that would be, yeah, but I think people will continue to want to know, to try and understand what is AI and what isn't. And we're not going to be perfect at it, but I think there, what has proven to be true in these last 3 years is that people are wanting to know if something was created by AI or not. I don't think it's always bad if something is created by AI. I think there's a lot of use cases where people are consuming AI content and they are aware of it. There might be a video, there might be a song, and people are aware of it. But I think within the context of somebody saying that it's them trying to pretend that it's you trying to pretend that it's not AI, and then it is, there's an opportunity for creators to come and create organic, compelling, interesting content in a way that pushes back against that thing that you talked about, Andrew, which is like, people don't like it. People find a tension with it. So Miranda, one of the things that we're going to talk about in this episode is this idea of Tools versus, uh, AI as a tool versus AI as a creator. Can you talk about that distinction a little bit as it relates to somebody who has a creator business?

Yeah. So one of the things that I think we have to keep in mind, like when you're talking about AI will just continue to get better, one of the things that sticks out to me is AI is only ever going to be an average. It's math. It's really always just a math problem and it can only ever be average. So if creators keep pushing the envelope and AI keeps getting better, I think that's because creators keep pushing the AI. Like the creators keep pushing the envelope. They keep getting better. They keep leveling up. And so in that regard, maybe AI is, is a competition that's kind of good. Do you know what I mean? Um, one of the things that coming from a writing background and from an English background, one of the things that stands out about AI content creation specifically is when you can pick up on writing patterns that previously didn't exist in someone's writing. And so we've started to notice them, right? When, uh, it's not this, it's that.

Um, that is a pattern. Like, that would be a structure that AI would use.

M dash is maybe the, like, you can pry the M dash out of my cold, dead hands, but exactly that kind of thing that genuinely or quietly fails. Um, That pushes a real writer to really hone their voice, to to say okay, if I know that AI has trained off of decades of writing and this is what it's come out with as the average, and I feel attacked because that's how I was writing, it's taken then then that's a tip for you as the writer, as the creator to say how do I level up my voice? How do I make myself stand out against that? Because really, that's kind of a gut check moment. If you sound like AI or AI sounds like you, it's a chicken egg and you, you have to get better. You have to improve.

Mm-hmm.

So I think just from the content creation perspective, from the writing perspective, that's one way that we can, can think of AI. The other thing is visuals. When, you know, the first time you see a really, an AI flyer for something at your kid's school, PTA, whatever, you think, oh my gosh, look, the PTA, this graphic looks so Oh, great. And then you see another school's PTA and the graphic looks exactly the same. And I went to my pediatrician's office and the flyer hanging in the pediatrician's office yesterday looked exactly the same as the one the PTA is putting out, which is the same one as the football boosters putting out. So you really do start to see that regression to the mean. And I think that's where visual artists get an opportunity to excel too. Like, how do we level up our creation? Because again, these AI tools are showing us the average of what we've been doing for so long.

Mm-hmm.

And that is why it also feels so kind of uncanny valley is suddenly we're seeing the middle is very exposed.

Yeah, it's, it's such a great point because I think that one of the things that we have to do as a creator is stand out. We hear about this all the time. How do you stand out as a creator? And there's a lot of different ways to do it, but the best way is to be excellent. And how can you be excellent? There's a lot of different ways you can be excellent. You can be excellent written communication. You can be excellent in systems. You can be excellent in photography and video, in strong hooks. But excellent, I think, by necessity means that you are above the average. And so I think sometimes what it can feel like is, hey, AI can write this email for me. Now I check the box on emails because this email goes out and it was done for me. But if it goes out and it feels— this happened to Lindsay the other day. She got an email from a creator, and she said it felt like a ghost town. And it was when we were having a conversation around the context of AI and what does AI look like, how do you use it? And her pushback against it as somebody who is like very, she doesn't use AI within the context of like creating a thing. It's like her writing a thing. She'll use it for other things like a tool, which we're going to talk about in a little bit. I use it as a tool as well. But it's almost for me, not even like ethical. There would be that consideration for sure. It's almost marketing. It's like, you have to be, you have to stand out because if you don't, then it's going to be average. If you are average, then your reach within a platform, within organic search, within an email, isn't going to reach an audience because it's going to be the average. It's not going to be something that compels people and draws them in. But at the same time, these are incredible tools with incredible power. The tweet that I saw from Peter Diamandis or something like that yesterday, he said, he was talking about this concept of like, if we had I'll pull it up because I feel like it's worth getting it exact. Imagine I handed you 10,000 brilliant employees tonight. What would you do with them tomorrow morning? That's exactly what AI agents are about to give you, and most people have no answer. And I think for a lot of us as content creators, the first answer that we come up with is, I'm going to have it write all my emails and write my post. But that should probably be one of the—

That should be the last thing.

Last thing that you should do. So what should we be doing if we have these 10,000 brilliant employees at our disposal? What does it look like to use them in the most strategic way possible, like within the context of a tool versus as a content creator?

Do you want me to take that or do you want Andrew to take it?

We might both have answers.

Uh, yeah. Why don't, why don't you start, Andrew? And then Miranda, if you have additional information, you could Yeah, Dan.

Um, my answer to that is to, to use it for the things that humans aren't good at. Um, and especially if you're a human that's not good at something. Now, you may be really awful at writing emails, uh, but hopefully not the case if you're a content creator. Um, that's where I'm finding it, like, it ha— it has abilities that we do not have. Um, data ingestion, analyzing things, researching things, finding that email and connecting the dots between certain things that, like, It's fascinating because it's like the smartest thing with the shortest possible memory. And, and you kind of have to constantly steer it, right? But I'm using it to like, like I wanted to know, I wanted to, I wanted to go back and look at like, what are some of our accomplishments over the last 6 months at NerdPress? And it put together this amazing artifact of like 80 different things. I was like, holy crap, we did all the— I'd forgotten we did this and this and this. And, you know, it does that by searching Slack.

Yeah.

And emails or whatever, whatever context we give it. Right. And it would have taken hours and hours and hours to do that work. There was no creativity involved really. Right.

It's, it's research, it's information gathering, and then it's compiling that.

And, you know, not, not, I don't want to diss on researchers, uh, but, you know, it, there are some jobs where I don't think it's going to take jobs away, but I think it's going to change how those jobs are done.

Sure.

And, you know, for programmers, that's already happened. Um, but I, I think that's where I'm like, how cool, I can do this now. I have a superpower that I didn't have before.

Yeah.

And that like has made me level up. It's made a lot of us at Nerd Press level up and the whole company has leveled up. Whereas when I try to add it, when I have it try to write an email for me, if I, even if I give it so much context, I spend an hour transcribing my thoughts, you know, to an AI note taker and say, I want you to cover this point, this point, this point, this point. It spits out an email that's pretty good. But it's soulless. And I just sent Miranda a draft and I was like, okay, this is a good framework. Now we have to not use it at all and just write it from scratch following that framework.

Right. Which is hard. That's really hard. Yeah. Yeah. Yeah. Yeah. The information is good that it has in it, but in terms of like, uh, what it feels like, that, that's not good.

Yeah. Even when I have it like create a writing skill and prompt, I mean, it's, It's weird. Like, I, so I spent some time having it create an Andrew writing style, right? And I write a lot. I send emails, I write in Slack, I do blog posts, I do one-on-one, you know, like I communicate and my style is slightly different on each one, right? Um, and it actually did an amazing treatise on what my writing style is. Like I read it and I was like, yeah, I do that. I don't do that. I do, you know, it was an incredible analysis, but then when it goes to write, it kind of follows those rules, not necessarily. But it's not me and it's soulless. Somehow it just, it's that intangible thing and it's close, but it's never going to be there. I don't think, no matter how good it gets.

The interesting thing, you know, yeah, I think of it outside of the context of content creators. I think for people who are poor writers, it can be a really wonderful tool because suddenly you can articulate what you want to communicate and it can bring it from subpar. It can bring it from really poor writing to average, which is like a huge level up. And I think of some people in my world who I can tell, like, um, you know, this isn't necessarily working relation. Well, it's even within like work stuff, I can tell like, oh, this was, I can tell this was AI written. Like I think of a text that I got from somebody and I was like, it was, it was articulate and was like expressing a thing of gratitude. And I was like, I think that's, I'm like 95% confident this was AI written and good for you for using a tool to put your thoughts together and then being able to take something and pass it along. It wasn't, and this was not like a content creator, but I think within the context of content creators, that feels different because of that conversation that we had around needing to be above average, needing to be different, needing to be novel. Uh, story, I think, is another important component to it. Like, you have to have, the more that you can have story and involve that. So we've kind of checked that box, but now, okay, there's this tool, there's this incredibly powerful tool. We know that maybe it's, at this point, it's not going to be the best for articulating information in a way that creates human connection. It's going to be great at doing that if it's just pure information exchange. Um, but there are a ton of these opportunities for us as business owners to use it really strategically. So Miranda, can you talk a little bit about some other instances of how you see it being used as a tool? And can you also speak to this idea that Andrew talked about, which is context and how important it is for us to give it context?

Uh, context is huge. And when you, I find that when I don't give Claude, we use Claude at NerdPress. We use it. And then I also have a personal Claude. So, um, and they have 2 different contexts, right? Because my personal Claude doesn't really do my work stuff and my work stuff doesn't do my personal stuff. And so the context I'm giving it in both places is different. Um, but when I don't give the tool the right context for what I'm asking it to do, the result is terrible. Um, for instance, I do our News from the Nerds, which is our biweekly site report. We write those blurbs that go in the email, but I can't read the entire internet every day and surface all of those, like all of the articles that come out every day. I, I am one person. I can't read all of that. So I have a Claude skill that's written that runs every day that scans various blogs, websites, topics, hits keywords, and looks for what's new. in the industry and then surfaces the top, you know, 3 or 4 posts a day for me to read. So, I'm not hopping from site to site trying to find these articles myself. But when we started out with that, I didn't have enough context in what News from the Nerds was, what the site report needed to look like, what the structure was. So, it was giving me things that were out of date. You know, it would give me something about like a vulnerability that was already 6 months old and had, was like, 3 releases ago had already been patched. Well, that's not relevant to me. So even thinking in terms of like, what are the time limits of the context that you want it to have? What are the, what are the specific websites that you only want it to look here? Or you only want it to look for things that have, you know, these keywords, these topics. That context is so crucial. And once I really dialed in that skill, My, it started working a lot, a lot more smoothly and I started getting a lot more relevant, um, articles that I could surface for that newsletter. So I think it took us from maybe it taking us 5 days to prep it to now we can get it done in about a day, which feels fantastic anytime you can get it done that, that quickly.

Um, I'm going to say though, that's testament to you too. You've gotten very good at writing the news, like, and curating it and using your brain.

Um, right.

You know, uh, and figuring out what is relevant to all of our clients and all of that. So it's credit to you, not Claude.

Well, no, it's credit to Claude for doing that research. You know, when you're talking about research, and I think that's, that's the thing to talk about too, is like, uh, when we talk about research, what do we mean? Because if we're talking about scientific research, obviously we don't want Claude to like develop and design, uh, you know, science experiments and then go try, but But if it's crunching the numbers, if it's pulling from a broad swath of the internet and it's, it's collating, uh, that information for us, that's a little bit different. I think that's, that's a, it's just a, it's a use of, it's a better use of my time.

Mm-hmm.

Um, and so I try to look for, uh, how do I use this tool in a way that either can do the jobs that I don't like doing? Um, like making my daily to-do list. I don't like doing that. So my daily briefing does that for me and it like pulls from everything and tells me what I need to do today. Um, and what do I not have time to do? If I'm also doing all of these other things, I don't have time to read 10 websites a day to find the latest news articles that I can use to pass along to our audience. So that's 2 ways I think to use it. And then the other way is, is to Mm-hmm. to use it to find your blind spots. Um, and that's where I think it comes in really handy for creators, especially as we start talking about kind of agentic AI and blogs, where it's, we're not using it for content creation, but we're using it to help us figure out what we don't know. What do, what do we, what did we miss 10 years ago that now we need to fix?

Yeah. So it's, it's, um, One of the beliefs that I've come to develop over the last, I would say 2 months, is I generally feel like Claude, Fable 5 Max, uh, ChatGPT Soul Ultra. So these are both the thinking levels, uh, like the model level and then the thinking level are going to be better than me. In terms of knowledge, uh, on 95% of things, I'm just obviously picking that out. Math, English, it has the knowledge of the internet, but what it doesn't have is, uh, my conversations with people, my interactions, my history, unless I give it that context. Now, in a hypothetical situation, let's say that Claude 5.5, you know, and ChatGPT Soul, um, actually had an entire database of my life. I, I know this is impossible. All my interactions, all my conversations, all of my history. It would be probably better at analyzing situations in my life than I would be because it would have complete knowledge of all of those things. It can't. It doesn't, but there are ways that I can give it some of that contextual information, um, to the degree that number one, that I'm comfortable. And number two, that I have access via an MCP or an API to that information. And so one of the opinions that I've started to develop over the last 2 months is if I believe that generally as a knowledge processor, That this tool is one of the smartest tools possible. A really narrow example is like, as a financial analyst, it's gonna be really good at analyzing investments, and it's not gonna be great if it doesn't know what investments I have. So I need to give it access to that. I wanna hear from each of you the areas that you've opened up to Claude in order to broaden the context that it has around you. And how you're strategically using those connections in that context to help Claude give you better answers. And Andrew, we can start with you. Before we continue, let's take a moment to hear from our sponsors. Building a creator business isn't getting any easier. Between AI, constant platform changes, and the pressure to keep growing, it can feel like you're running just to keep up. That's why we're excited to partner with Raptive. Raptive works with thousands of the web's leading independent creators, including us, to help them them earn more, grow smarter, and build sustainable businesses for the long term. They combine industry-leading advertising with personalized guidance, data-driven insights, and a team that's constantly helping creators adapt to changes in search, AI, and the broader digital landscape. They're also leading the charge to protect creator content and advocate for a future where independent creators can continue to thrive. And now it's easier than ever to join. If your site gets at least 25,000 monthly page views, you can apply to become part of the Raptive creator community and get support that's designed to grow with your business. To learn more or apply, visit raptive.com/fbp. Thanks again to Raptive for sponsoring this episode.

Um, I'm kind of grinning and so is Miranda because I think we're, we've had the same timeframe with you on that. And I think a lot of people have where it's like, you know, I wrote down on an index card before we started talking, context is everything. Um, without it, it's just average. But once it has that thing, it starts connecting pieces that I didn't even realize it could. And like every single day, it's like, Oh wait, check with Colin about this thing because he said this over there in Slack. And, um, so in terms of how I'm using it, like, I'm, I'm letting my guard down a little bit more than I ever thought I would. Um, I'm, I'm letting it into my Gmail. Um, and I'm not having it write emails for me, but I'm letting, I'm, and like, I'm, I'm pretty privacy focused, right? I don't even love that I, I'm using Gmail. I've been considering moving to a different platform because I don't want big bad Google to have my emails.

Sure.

Right. And now I'm like, oh, sure, Claude, go log in. Right?

Yeah. Here's my OA. Yeah, sure.

And, um, I, so I don't know, about 2 months ago I saw an ad. I was influenced. I saw an ad on Instagram for granola.ai. And the catchphrase was, you've probably heard of it, uh, the AI app for people in back-to-back Zoom meetings.

And you're like, that's me.

I feel seen. You get me. That is exactly what you needed to say. I am in back-to-back. I will be in Zoom meetings for 6 hours a day. And so I have, I installed Granola and it's basically spyware, right? I've literally willingly installed spyware on my computer to listen to conversations.

Mm-hmm.

Now it doesn't always do it. Like it goes, you know, when I'm on Zoom or something, it says, hey, would you like me to take? And so then it starts listening when I click okay. So it's not always spying, but I've literally like have it doing that.

And that's, and one example is meeting notes. And then the, to draw the connection here within Claude, you connect Claude to Granola. So now Claude has context around every meeting that you've opted in to record within Granola. So that's one concrete example. That's the special sauce. Absolutely.

Like it's not, oh, here's a bunch of notes of here's your, your to-do item. Here's, it's not, here's your action items at the end of the call that I have to copy and paste and create. tasks with. I don't even look at it most of the time. Um, but what it does is in my morning briefing tomorrow, it's like, hey, don't forget you told Miranda you'd do this, right? And it pulls it from the transcript just automatically because of that MCP connection. And, um, you know, if you take that a step further, there are now apps, um, I think Little Bird is one, littlebird.ai, where it's spyware to the extreme. It watches your screen.

All day. All day.

It watches everything you're doing. Um, now a month or two ago when I was like reinstalling Windows 11 on a machine and Copilot wanted that permission and I like slammed the lid shut on my laptop. Uh, I was like, you want to do what now? Right. And some, it's weird because somehow I'm like, well, I might let Little Bird do it, but I wouldn't let Microsoft do it.

What's the name? Little Bird. It's so unassuming.

It's so unassuming, like granola.

It's like, oh, it's lovely.

Yeah. Bottom line is like, I'm giving some cloud provider really intimate, direct access to anything I say. And then that is staying in a database. And then I'm giving Claude access to all of it. And the question is, is it worth it? And I'm like, well, we're leveling up. So yeah, I guess.

Yeah.

Yeah. It's simultaneously amazing and terrifying.

Yes, it's both.

So, so that's one Granola, uh, MCP connection into Granola. And I think it's helpful for people to understand who haven't done this. You go in, there's a connection option within, you know, Claude or ChatGPT. Many of these applications will then have the ability to say like, hey, I want to make a connection. You sign in and then it gives it, you have to grant it access and then it has access and then you can reference it within your conversations with Claude. So Granola, you mentioned Slack, you mentioned Gmail. Are there other ones that connections that you've made, Andrew? And then I want to hear from you, Andrew or Miranda.

Well, internally we've connected all of our tools and that's, that's actually been the biggest game changer. So like Slack is really an internal comms tool, right? That's how we message. We don't use email much at NerdPress. It's almost all Slack. And it's just like, it's such a big unlock. It's what we kind of call it. We're like, with all this stuff, it's like, since it's never been done before, you have to have these moments of somebody going, Hey, did you know you could do that? And you go, oh my God. And then you like next level up, right? And then the next day there's another unlock because somebody else mentioned something. You're like, I couldn't, couldn't even fathom.

Well, and that's, that I think is the beautiful thing of what we've done is that, you know, when you're talking about a fully remote team, 20 working across the country, across time zones, you know, somebody comes in first thing in the morning and they experience this thing and they're handling tickets in our, in our, um, client software, like they're responding, somebody else comes in later and they get a similar ticket, they get a similar question from another client. Those 2 things aren't talking in, in most cases, but now they are because we've built the tools that help them talk. And so, you know, if, if we're getting 4 or 5 clients emailing in about the same plugin, then our tools are saying, hey, by the way, You've gotten 5 questions about this over the last 5 hours from 5 different people. Maybe this is a bigger thing that you need to look at. And we're not take— we're not doing, we're not connecting the dots anymore. Does that make sense? I'm not sure I'm explaining that correctly. Um, but it just allows us to catch things faster. We're not going to miss things.

Mm-hmm.

Not that I think we were ever bad at missing things in the first place, but this just, it lets our team collaborate more seamlessly.

Yeah, I think maybe one of the key takeaways in all of this that I've started to notice is that I think a lot of times people talked about using AI and have said like, hey, it's going to be great because it's going to be able to replace this work that you used to do. You used to do this and now it can do that for you. You know, it's like, um, for me, I would go in and review our business accounts and make sure that we have like 3 months of expenses in those. That was a routine I had. Okay, now AI does that. Like it's able to go and see like, do you have 3 months? And it sends a report. I think what we haven't talked about as much that is equally or more valuable is proactively, what were the things we never were doing that actually are huge opportunities? To your point, Andrew, like a level up that exists. And my question is, as a creator, What do you think those opportunities are? So if you are in the weeds, if you are, uh, you know, you have one, you're one person, maybe you have some part-time team members and you want to maybe do some of the things that you were previously doing, but also you want to do more without having to work an extra 30 hours a week. Given the tools that we have, which are Claude, ChatGPT, and then some of these connections, these MCPs we call them, which are essentially like a pipe that connects one data source or one login to Claude and ChatGPT. What are some of the things that creators can be doing that allow them to either take some of the work off their plate or do more work proactively without having to do additional hours in a week? Systems, processes, you know, data integrations, things like that. It could, that could go to either you or, uh, you, Andrew, or Miranda. Miranda.

Well, do we want to talk about the tool that we built?

You can do, yeah, I think that would be a great example.

Because I think that would be, I think if we talk about the tool that we built, Andrew, and then I'll talk about how I've used it.

Sure.

Because we're talking about the pipes and the connections right now. So this might be a good time to, to segue into that.

Yeah. So I think, you know, the whole thing that we're realizing is like, you've got to connect all these things for it to work. So it has the context and it has the capability. to do things. Um, but we've also noticed that like AI is a people pleaser, right? It's programmed to like try to keep helping. It's like, oh, oh, what else can I do? Oh, here's an idea. Maybe I can fix this, right? Oh, would you like me to do this now? Oh, I can take care of that. And I think some of that is they're trying to get us hooked, frankly.

Yeah.

But also because people don't know how to use it, so it's kind of self-teaching us. Um, but it's, it's very, it is very eager to help. And so it's very easy to go off on a tangent Because you can be like, oh yeah. And like, there's no end to it. Um, it's, you know, you end up at midnight, you're like, ah, I need to go to bed.

You're like, I'm not done. It's, it's one of the like quantifiable downsides that I've noticed is like the, which you could have an entire episode on that, the unending amount of work that, that you can start to do and how just how work changes. It's, I've been thinking a ton about like—

And that is why I don't pay by usage for my personal Claude account.

I, I, I burned a lot of tokens this weekend. Um, So, so I think, you know, the idea that AI is going to do all the mundane stuff and we're all going to have a 20-hour workweek, flat out lie. No technology improvement has ever done that. What's happened is everybody's had to produce more. Their output has to go up. Competition has never been stronger, right? I've said, I was at the National WordPress Conference, WordCamp, last week. And I said to several people, I'm like, man, if I was the only one who had AI, I would love it.

Yeah.

And, and everyone like immediately laughed and understood what I meant, which is like, but we all have it. So we all have to scramble to use it better and faster and output to stay competitive. So that, that's the balancing act. And you're going to drive yourself crazy if you don't get outside and touch grass every once in a while. So you have to take good care of yourself and just know to unplug and be like, you know what? It's, uh, this will be infinite and we are finite beings.

Yeah.

But To get back to like productivity and actually making your life better and being able to do the things you want to do that you couldn't do before, you know, for content creators, a lot of that is just a lot of work editing your site. There's so much maintenance involved. You know, if you've been publishing for 5 or 10 years and you have 500 blog posts, so many of those need help and love, right? And what we're seeing right now is people are starting to connect their websites connect an AI agent directly to their websites. Um, and I know you had Shruti on recently, and she's, you know, talking about using AI as systems and processes, and totally agree with a lot of the stuff she's saying, right? It's very aligned with what we're doing is like AI's tool to help you with your productivity and workflows. Um, but because AI wants to help and it's so eager, and so like, it will often do things before you ask it to or without you asking. And it's really hard to put guardrails on it because it's so— It's like this weird, unpredictable thing, right?

Mm-hmm.

Um, and so it may do things that are, in the context of trying to be helpful, not what you want. Um, and so I don't think it's a good idea for people to make a direct MCP connection to their WordPress site. Uh, not on a production site that is your business. It's just too risky. Um, because it's gonna be like, hey, I updated 3 other posts for you. Oh wait, I deleted this post 'cause I didn't think it was necessary anymore. And it just, it's well-intentioned, it, or I guess, um, but it can kind of go rogue. And so, And also right now it's pretty difficult to connect it. There's a lot of steps involved with authentication and stuff, cuz WordPress does not do it outta the box, even though they've added API features, sort of like they've added some plumbing stuff, but it doesn't really do it yet.

And to be clear, what you're talking about is if you use ChatGPT or Claude, mm-hmm, and you say, I want to connect to my WordPress website via ChatGPT or Claude, and then work on my website. Within Claude or ChatGPT. That's what we're talking about here is like, usually we're used to going in and doing the actual work. We change the post, we update the titles, we swap out the images, we update the recipe. We're in there doing it. One of the changes in this agentic world is that now we are actually starting to task agents with doing that work. But in this context, you're saying, number one, It takes a lot of work as things currently are with WordPress to connect your Claude or ChatGPT account to your WordPress account. So there's a lot of work involved. And then number 2, one of the issues is that there aren't guardrails, uh, in that connection that keep it from doing things that you probably don't want it to do.

Yep.

That's exactly right. And, you know, there are ways to say like, don't update a post without asking, but still it's like, What's the update going to be? And so what we, what we've created is we're calling it the NerdPress Guided Agent Connector. Uh, we went with a very boring name, very descriptive name.

Um, yep. Relay. Relay and guided connector.

Yeah.

Naming things is hard.

Okay.

Naming things is hard.

We didn't want to come up with a cute name. We wanted to be descriptive because in each of those, each of those words is there for a good reason. Um, but it is literally to help people connect, uh, their AI tool, Claude or ChatGPT, to their website so they can be more productive. But it's easier and we're trying to make it safer. So we've added audit logging so you can see what's been done and undo functionality. So if it does something you don't want, you can hit the undo button, um, you know, where it's possible to do that. Um, but I think the special sauce that I'm most excited about is the guardrails and the guidance. So we've added extra guardrails where you could have to, we can say, no, no, no, you can't do that. Like just at the, at the MCP level, we're able to say that's not allowed. Um, Or, or it can ask for, we can require the agent to ask for approval as well.

Give an example of what that would be.

Uh, so if it's going to be, let's say you want to delete a post or it wants to delete a post, what we can do is say to, we actually could communicate directly to the agent and say, ah, that needs extra permission. Give this link to the user, make them click it, and they can click okay. And so the agent will dutifully follow our instructions, send the link in the back in the chat to whoever's using it. And say, hey, NerdPress wants to be extra sure you really want to delete this thing. Click this link and say okay, and then come back and tell me you've done it. And until the user has proactively done that, um, it won't, literally will not be able to delete the content. And then once, once, you know, whoever's using it says that's okay, then we're good. We don't want to introduce too much friction, right? That sounds like a pain in the butt every time. And so we're only doing that for really important things. Um, uh, maybe a gentler way to to, to explain it is like, we're going to say, hey, you know what, make a, make a backup before you do this. And we're going to tell the agent, hey, before you delete that, run a backup. And it's going to go, oh, I'm supposed to run a backup. Now I'm deleting this thing. And that's just going to happen automatically. And so we're essentially baking in like 10 years of NerdPress expertise directly into that MCP connection. Um, because over the past 5 years, we've been building a library of internal documentation. Um, because scaling a team is really hard if you don't have documentation. So we have all these documents of like, When somebody asks a question about Yoast, what are the best practices? Like, what should my featured image sizes be? All of this stuff that like we've been talking about for a decade on this podcast, that is like, here's the NerdPress way to do something. Here's guidance. So it's, it's very specific to our clients rather than just the general website, uh, word, uh, internet at large. And so if you upload a featured image, it's going to know, oh, I should size this to 1,200 pixels.

Mm-hmm. Yeah, it's, I almost see if you, if we're talking about this analogy of a pipe, the pipe is the connector. I almost imagine maybe a better way to say it is like, it is a tunnel and agents are people and you can have a tunnel that is like open access and it's just a database to Claude and the agent can go back and forth as they please. What you're describing is almost like key fob access. And so These agents that are going around are going to have to have this special key fob access if they're, if they want to get through and do a certain thing and, or they'll try and get in and they just can't do it because of the guardrails that you've put into place. And I think it's one of the things that will be important for us to know as we continue to use tools like this is to understand the directives that we give something. The example that I have, um, there was, uh, so I use OpenClaw, which is an extreme example of like a tool that can do many different things. And one of the things, uh, I asked it to, um, I don't remember exactly what it was, but I remember being like, oh, that was farther than I wanted you to go. Like it was like canceled an appointment or like, Oh, this is what it was. Our stove wasn't working. And Lindsay was like, can you follow up and figure out like how we need to like get this taken care of? And so I gave it a prompt that probably wasn't the best prompt, something like, we were trying to figure out our stove. Can you follow up? Can you find who previously— find where we bought it from and figure out what the customer support process is or something like that? And then it just submitted a ticket on my behalf. But it was a ticket to the old, uh, company that we had used before. And so they called and they're like, hey, do you want us to come out? And I was like, no. And so it doesn't happen a lot, but that was an instance where I was like, oh, there was a pipe. It had access to do it. And it did a thing that I didn't want it to do. It was probably a bad prompt, but then also there wasn't a guardrail in place for it to check back with me.

And it goes back to context is everything, right? It's really hard to prompt well. Um, to, to give it, give it all the information it needs to make the choice that you've got in mind. And, and so we're basically adding that extra context in with that connection. So it's, it's almost like there's, I love the key fob analogy, but there's also like, there's instructions posted on the wall of the tunnel too.

Yeah.

Right.

That's great.

So the context thing is really interesting. And if I can just a minute, uh, really quickly, one of the things that I've been using the agent connector to do is like clean up a lot of the technical debt because we're talking about a website now, 20 years old. Andrew mentioned 500 blog posts and I was like, 2,000?

Wow.

And like, you know, uh, this is Blogspot days, so that's before categories and tags were even a thing. I know Shruti mentioned that she used her agent to do taxonomy, which I thought I immediately texted Andrew and I was like, Shruti used to do the exact same thing that I just used one to do. So, cause I was like cleaning up categories, but in the, in the lack of context. So one of the things that Um, cleaning up my site, I have a lot of thin content because in the early days it was, you know, your blog posts might just be like a quick little diary entry. Um, but maybe that diary entry also came along with a baby picture of my son. And when I started this project, Claude wanted to delete over half of my website because it was like, this is all thin content.

Mm-hmm.

And didn't understand, like, the name of the blog is Caffeine and Cabernet. That doesn't necessarily lend itself to automatically thinking that this baby photo from 16 years ago, 17 years ago is relevant. So it was like, we, it just, we need to delete 900 posts. And I was like, uh, those are actually, this is like a baby book kind of. So maybe we don't delete all of those things. And so then it got a little more context. It got a little, it's like, oh, well then all of these posts we need to delete because they're. you know, from their giveaways and old link-ups and that sort of thing that we were doing in the dark ages of blogging. Um, so, but, but I felt safe with that because I had the NerdPress guidance. Like I had like, hey, we actually don't recommend deleting posts, move it to draft or, you know, delete this, but maybe this is when you don't delete. Um, and it was giving me that right there in In the chat with Claude, it was telling me like, here's what NerdPress recommends and here's how we can do this in a way that protects the legacy of your site and what you want it to be while also leaving it open for you to make it whatever it's going to be in the future.

Yeah. I almost feel like with AI, there's 3 layers. There's the layer of like the broader internet, the open web, everything that AI has trained on, which is pretty comprehensive, but then there's an additional layer below that, that's harder to access, which is like industry knowledge. That would be an example of like the NerdPress documentation. That's not something that Claude can crawl. And it's probably not something that's super well documented on the web. The nuances of best practices for a WordPress site that is, you know, primarily food sites, like Okay, what do you do within the context of Yoast if this question comes up? Like there's going to be Yoast documentation on the internet, but probably not that specific. So there's like the open web, there's the industry knowledge layer that a lot of times is closed. And then the last piece is like the individual and your understanding and your beliefs around your business and your contextual knowledge that for sure isn't available. It's not like crawl, uh, Claude can crawl. Your knowledge base, which is like your brain and you knowing that, hey, I want to keep this post. I don't want to keep this post. And so I think most of us are operating within the context of Claude or ChatGPT, the database that they have, which is extremely comprehensive, smarter than almost anybody, but we're missing that like middle layer that exists, which is the industry layer, which I think is what you guys are building, which is really cool. And then I think we need to remember. That we have expertise and knowledge and we need to still apply ourselves against the responses that we are getting and think strategically around those as opposed to just YOLOing and let it, letting it do whatever we think it should.

Yeah.

Um, as we close out, I want to talk about the different layers that somebody can step into if they are starting to get into AI. Maybe better ways to say it, levels. Like, so if somebody's level 1, they're probably just starting out. They are using Claude and ChatGPT to answer questions almost like you would Google. What would you say are the other levels that would exist beyond that? And where do you think that we are trying to go as creators? So maybe let's say there's 4 levels, maybe there's 5 levels. Could we kind of help people think about stair-stepping? And we could do this collaboratively, stair-stepping up from one level to the next. We can see how far we go. Let's say that first level is you've downloaded ChatGPT, you've downloaded Claude, and you're just asking it questions, general questions around What's, you know, uh, when is Christmas this year? Like, uh, what day of the week does it land on? These informational questions. What do you think is the next level that people can go up to after that initial kind of level 1?

I don't know. I don't know if this is a level, but my first thought is understanding memory. So as you use these tools, they automatically have memory that they Right. Um, and so under, like, I think the next level is starting to understand how the tool works, maybe. And so memory is one of the key things that makes this actually really cool is it'll remember that, uh, oh, you're Jewish, you don't celebrate Christmas, right? Uh, if something's coming up or, and it might just, and it, it can like bring that out of nowhere. You're like, oh, I want to make a cookie recipe in December. And it'll be like, oh, how about dreidel cookies? Right. It'll, because it has that context.

Mm-hmm.

And it does that automatically. And I think what people don't realize also is you can actually see that memory. It's just, uh, what's called a Markdown file. It's most of the stuff is stored in text files. Um, AI agents love Markdown, which is just a way of formatting text. So you can have like headers and links, really basic formatting. It's kind of human readable. It's, it's not the prettiest to read, but you can read it. Um, it's not code, right? It's like you can, you can open a Markdown file in a text editor in like Notepad and You can totally read it. Um, and so it'll, it'll actually, like, its programming language basically is English or whatever language you speak. Um, and so understanding that that is the programming language of it is maybe a way to start understanding it. And then, and then maybe on top of that is like standing instructions.

Standing instructions, meaning?

So you, um, well, one that I have is that I don't want Claude to be a sycophant. I don't want it to just always agree with me. Um, Uh, and so in my instructions, in both my Nerd Press Quad and my personal quad, is like, you can push back. If, if, if I'm brainstorming an idea, if I'm working through something and it's a bad idea, you can tell me it's a bad idea. So that, so that some of the instructions are as simple as that.

Mm-hmm.

Uh, I know that Andrew loves to, to tell, like, his instructions are interview me to get the answers that you need. Like, always ask me questions. to get more context if I haven't given enough context in the prompt. Um, so it's almost like shaping the tool.

So starting to say, here's how I want you to act. Here's what I want you to do. This would be the next level, shaping it, not just assuming there's a default, but like, here's how I want you to partner with me in this context. I think that's great. I would almost say like a level 3 would be starting to develop, you, uh, talked about this, Miranda, like a scheduled task or a skill.

Mm-hmm.

Can you talk about what those are? And, you know, we could go deep on them, but at a high level, how could somebody create their first skill or create their first scheduled task?

Well, uh, I am the least technical person to be answering this question, but I, what I'll say is, and I think that makes it like relevant for me to answer it, right? Because I'm not a technical person. I literally just go into Claude and I, I use Cowork when I'm doing skills. So I open up a project in Cowork. So let's talk about my daily to-do. Um, and I just start writing. I need you to help me write a prompt that I can turn into a skill that will develop it, that'll write a daily brief for me that'll run Monday through Friday at 7:00 AM. What questions do you, like, what questions do you need to, do you need answers for to be able to, um, formulate this. I want you to pull from Slack, from my email, from my project management software, from my calendar to see what meetings I have. And, and start giving it all of that and ask Claude to help you write the skill or ask ChatGPT to help you write the skill. They're really good at writing skills. This is also a trick sometimes is you can ask one tool to write the skill for the other tool. And sometimes you get better results that way. If you ask ChatGPT to write you a skill for Claude or ask Claude to write a skill for ChatGPT. But you can ask the tool write the skill for you if you know sort of what you want to do, but you don't know how to write it and you're worried about like, am I going to be technical enough? And then I'd say beyond that, constantly refine your skill. And when you're refining your skill, you know, if I notice that it's over-indexing on old notifications from the project management software, I'll be like, we need to, we, this needs to be less important. in what you give me in my daily brief. And if things have been lingering in the project management software for 2 weeks, don't tell me about them anymore. They're on hold for some reason. So constantly refining your skill and, and kind of letting it do it itself is a, I think a great way to go about it.

Yeah, I just did this today. So I was nerding out, NerdPress appropriate, um, and I had it watch my screen. And so it was actually what I was working on is like, I want to have a process for Ensuring, usually this is again a little bit nerdy, but like usually we have an investment allocation split for our investments. Okay, we want to do 70%, you know, total US stock market, 30% foreign, you know, total foreign stock market Vanguard funds. Usually we would allocate that like on a yearly basis, we would rebalance. I want to do it every time that I'm doing an investment. I want to do an analysis to see what the what, how it's drifted, and then make an investment based on the drift. So it's more real-time. So I just had it follow me in Vanguard to see how I go about doing that with the skill initially looking at the investment allocation. And so I would almost say like, you can do skills, you can do scheduled tasks. It feels like that next and maybe last level that we could talk about would be Starting to do those MCP connections. I feel like for any creator, the NerdPress guided agent is a great example. There's a ton, you know, Kit has one for email, all of these integrations into these platforms that we're using. Clarity, we launched one for Clarity so people can interact with their analytics data. Raptive has API data. So making these connections. Uh, and then I'd be interested if either, either of you, this is within the last month for me, have done computer use. Have you turned on computer use and like watched it navigate a website or your computer at all?

I have seen it navigate a website.

I've used Claude for Chrome.

Yeah.

Which, you know, so the, the, basically you can let it open a tab and navigate in that tab. I have not let it do anything else on my computer. Sure. Um, because of Terror, fear, security. And also, like, the funny thing is, is like, I'm actually really bad at delegating. My team, my team will laugh because it's true. And so it's like the whole, like, imagine tomorrow you have 10,000 employees. What are you going to tell them to do? I'm like, oh crap, I don't know.

I like, I go, go home and take a day off.

And so, like, just because I'm so used to being like, oh, I'll just do it rather, you know, Sure. I've, I've, I've gotten better at delegating as I've built a team. I've had to.

Um, well, you obviously have, you have 20 people on your team. Yeah.

But there's still a lot of stuff that I'm like, oh, you know, and, and so it's like, but I'm going to trust Miranda, a smart human, more than to, to do something autonomously than I am going to trust Claude to do something autonomously, especially when it comes to controlling my computer, which has access to a lot of things.

Yeah.

Yeah.

Yeah. Yeah, I've started, I mean, the majority of the way that I'm using it is within a browser. Um, but it's kind of one of those things, and I think maybe this speaks to the levels piece. I think what I've found, especially as these models get smarter and smarter, is like you start with like, okay, I'm going to make this connection into, uh, you know, into my email. You kind of experiment it with a little bit. You kind of let the rope out a little bit and you continually do that.

And starting read-only.

Yeah, I think that's a really key piece. Read-only, meaning it can only read, it can't write on your behalf. And that's exactly how it started for me with all of these little pieces is like, okay, let's experiment with this. What does it feel like to do this? Okay, what if I turned computer use on or Chrome access and it was able to go in and audit DNS settings for my email. Oh, actually, like I have some issues with my DNS settings. What does it look like if I just watch it and see what it does? Okay. Actually, it updated those. That to me is part of it is I feel like it's a little bit of my job, which is testing the upper boundaries of some of these tools. Um, but another part is the question that I've come away with is like, Like, at what point do I stop trusting it? And there definitely is that point. Um, but I think that's a little bit of what these levels are, is like an encouragement to people to kind of level up a little bit, try something new, but it's really important to remember that having those guardrails is important, which I think why what you've built with the guided agent at NerdPress is so great. We could talk for another 2 hours here. Yeah. Miranda, you can share that. Share that. Yeah.

Yeah. Those, you've kind of indicated what is so difficult about the AI conversation for creators, because so many of us have been sole providers, have been single operators for so long that the idea that this AI can suddenly do so many things for us, the, um, the temptation is definitely there to say, you know, how can I scale this up? Not just in having it write my content, but in doing all of the things that I don't wanna do.

Yeah.

But still maintaining that humanity and being really cognizant of what you're letting it do, I think is, is crucial.

Yeah, it is. I think one of the, as a last thought, I think one of the bigger considerations for us is how do we maintain our sanity when suddenly we have 10,000 employees that can do incredible work and are super smart? And I'm not saying like, then the only thing you do is like work with AI agents. It's like, all of us within the context of companies and knowledge workers, how do we maintain our sanity when we technically have this massive pool of knowledge that we can resource and that we can put to work? And I think the biggest shift for me is in the categories where I previously was an independent contributor, meaning I was doing the work, where now I am tasking the thing to do the work. And I'm not talking about like creating content. I think we've talked through that idea of like, we want to be novel and unique and human as a competitive advantage. I'm talking about like, one of the things I've been thinking, doing a lot is like auditing. I'm just like, it's audit everything, audit my Facebook account, audit my investments, audit my like Delta Amex card to see where I'm missing opportunities. Suddenly I realize it could audit everything. Like, And when do I stop? You know, that's the hard part for me is like, it feels like it's endless and it could go on and on. But I think the key takeaway in it is what we need to be doing is thinking strategically around how we can be using these as tools to help us with our day-to-day, those layers of knowledge and making sure that we involve ourselves in that. And then maintaining our humanity as a creator. And I think those, when I hear you guys reflect on that, it's like so obvious that you are deep in the creator world because you understand the nuances of the needs of this community. Um, for people who would want to use this tool, uh, or work with you within the context of NerdPress, can you talk a little bit about what that looks like? We know, you know, countless people who are, uh, NerdPress, uh, members and have used your services and I just feel so grateful that a tool like, or a service like NerdPress exists. Talk a little bit about what that looks like if people are interested. Maybe is there like a, people should start looking at it at this level as a creator? What does that look like?

So as we speak right now, as we record this, we are beta testing it with some clients and internally. I'm guessing by the time this airs, I'm hoping by the time this airs, we'll actually have launched it. So at this point, we are going to be making it available for all of our clients on our support plans. Um, so, uh, for no extra charge, we're just going to basically give this to all of our clients.

Awesome.

Um, and, um, that's where we're stopping at the moment. Um, we may open it up to others, um, depending on how that goes. Um, you know, I love the idea of one of our clients, uh, uses Clarity and they get recommended, they're working in Claude, they get recommendations from From Claude and Clarity recommendations in their Claude, and then they use our connector to safely make those changes and they're able to manage that whole project, right? It helps from recommendation to implementation and they and Claude can be smart in this. So we, we really want to empower our clients and help them be more successful. So basically on, on any of our WordPress support plans, it'll be available soon.

That's awesome. Yeah. Miranda, Andrew, thanks so much for coming on. Really appreciate it.

Thanks for having us.

Hey there, Anna again from the Food Blogger Pro team. Thank you so much for listening to that episode of the Food Blogger Pro podcast. If you enjoyed the episode, we would greatly appreciate it if you could share it with your community and leave a rating or review wherever you listen to podcasts. Join us next week for our next episode as Bjork interviews Yoli Weah from Food Wellness Equity about the Food and Wellness Equity Collective and their upcoming creator retreat. We'll see you back here soon. And in the meantime, we hope you have a wonderful week.