After 10+ years building The Blogsmith into a content agency serving B2B giants like HubSpot and Klaviyo, Maddy Osman French made an unexpected move: she went in-house. Now at DigitalOcean, she represents a “content engineer”. A marketer who blends process thinking, technical depth, and AI workflow automation to scale content operations efficiently.
What makes Maddy’s perspective particularly valuable is this combination of agency and in-house experience, paired with genuine technical depth. She’s a bestselling author of “Writing for Humans and Robots,” a certified content engineer from AirOps, and someone who actually builds RAG pipelines with n8n and Pinecone rather than just talking about AI in abstract terms.
We’ll explore how content operations are evolving, why she believes most SEO problems are actually process problems, and what it means to be a “content engineer” in 2026 when AI is reshaping how we create, distribute, and measure content success.
Check out the full episode
Check out the episode on YouTube, Spotify, or Apple Podcasts.
Mit dem Laden des Videos akzeptieren Sie die Datenschutzerklärung von YouTube.
Mehr erfahren
Auto-generated Transcript
Niklas Buschner (00:02.094)
After more than 10 years building The Blogsmith into a content agency serving B2B giants like HubSpot and Klaviyo, Maddy (Osman) French made an unexpected move. She went in-house. Now at DigitalOcean she represents a content engineer. A marketer who blends process thinking, technical depth and AI workflow automation to scale content operations efficiently.
What makes Maddy’s perspective particularly valuable is this combination of agency and in-house experience paired with genuine technical depth. She’s a bestselling author of writing for humans and robots, a certified content engineer from EROPS, and someone who actually builds RAG pipelines with N8 and N pine cone rather than just talking about AI in abstract terms.
We’ll explore how content operations are evolving, why she believes most SEO problems are actually process problems, and what it means to be a content engineer in 2026 when AI is reshaping how we create, distribute, and measure content success. That being said, Maddy, welcome to the podcast.
Maddy Osman (01:11.424)
Thank you so much for having me. And I really love that intro. think it touches on a lot of the things that I’ve been thinking about lately that I know we’re going to get into today.
Niklas Buschner (01:21.25)
Thanks so much. spent a lot of time on it and really tried to nail it here.
Maddy Osman (01:24.012)
And it shows.
Niklas Buschner (01:28.458)
Nice. So let’s start with the elephant in the room. After building the blocksmith for over 10 years, as I said in the intro, you went in-house at Digital Ocean, a brand that probably a lot of people in the marketing space know. Walk me through that decision.
Maddy Osman (01:39.722)
I did.
Maddy Osman (01:45.26)
Totally. Yeah. I mean, I think the biggest thing and something that we’ve been talking about is just like personal lives, right? And becoming parents and it changes things fundamentally in many ways. And so I am about to welcome my second into the world. And I feel like, you know, since having my first two years ago, it’s like so much has changed in this marketing landscape and
operating a company and I’ve changed too. I think that’s like the biggest thing is just it’s a different world now. And so I think I’ve always enjoyed the hustle. I’ve always enjoyed being an entrepreneur, but I’m kind of just looking for something that’s a little more normal, I guess, in a sense, a little bit more focused to, I think, on
the content itself on the marketing versus all of the aspects of running a business, the administration, the sales, just like the constant changing and rebranding as an agency in this AI world. And so I do still run the blogsmith just at a smaller scale to the side of this job. But I really saw it as an opportunity to try something new. Also,
I think there is a benefit in terms of switching your perspective, right? Coming from the agency perspective where you work with lot of different brands to then being able to fully focus on one in your role. And the great thing I think about working at DigitalOcean is I still get to explore a lot of these different sides. I’m an editor, I’m a strategist, I get to touch on GEO, I get to work on AI projects that are really exciting.
but it’s just kind of a different focus now.
Niklas Buschner (03:41.575)
Has the team at DigitalOcean been surprised that a 10-year entrepreneur is applying for an in-house role?
Maddy Osman (03:52.22)
Yes and no. One thing that was really interesting, I applied to two jobs after coming to this realization and got offers from both. The feedback that I really got from both of the hiring managers in this process was, I think, an interest in working with somebody who’s a founder who has run their own thing for a long time. think that
That’s something too, for anybody else who’s a freelancer or an agency owner or something like that who’s maybe thinking about making this type of move, it really is something that helps in terms of, of course, how you position it. But it’s something that I heard really from both of those different job offers that it was something that they were interested in and that they employ a lot of people who have gone through that same change.
Niklas Buschner (04:45.454)
I’ve also seen a couple of posts from startups. think it’s even been Aerobs, for example, where they said, yeah, we have like X amount of ex-founders on the team and we’re actively looking for people that have like this entrepreneurial spinner spirit. And I think from a, like a founder myself, from my perspective, makes a lot of sense to have people on your team that know how to run a business, right?
Maddy Osman (04:53.387)
Mm-hmm.
Maddy Osman (05:07.51)
show.
Maddy Osman (05:11.5)
Right, exactly. mean, I think there’s just, it’s like that word, intrapreneur, right? Where it’s like being an entrepreneur just within an organization. And so I think one of DigitalOcean’s either core values or just kind of how they operate is they want people who are looking for different opportunities, right? And then being able to capitalize on those opportunities, not waiting for permission.
And then once you get the green light, you know, being ready to take that and running with it. And so I think, I think any company would benefit from having that mindset. There are certain companies, maybe more so even in like the tech space that, that really go and like run with it though.
Niklas Buschner (05:58.647)
And did you have to limit yourself from the full autonomy you had when you were basically the CEO, so to say, now that on the one hand your spirit is valued, but on the other hand you also have to align with a team and somehow probably stick to certain guardrails more?
Maddy Osman (06:07.34)
Thank
Maddy Osman (06:11.052)
Mm-hmm.
Maddy Osman (06:21.42)
Sure. Now it’s an interesting question. think the answer is mostly no. I can still, like I still get to design my day. I have deadlines, right? Like that’s the limiting factor, but that’s the limiting factor of every job, including the one that I built for myself with other clients. Other than that, it’s like, if you get your work done and you want to
you know, explore a new learning path or if you want to build a new AI process or if you want to look into this thing that, you know, maybe even hasn’t been greenlit yet, but you want to pitch as a project. All those things are like well within the bounds of, you know, what I’m allowed to do. And then from more of like the agency perspective and running my company, you know, also from
just like a non-compete perspective or whatever, they’re pretty lax. I don’t know if that’s exactly the right word, but there aren’t a whole lot of guardrails in terms of what I can and can’t do, except for, you of course not to work with like a direct competitor. That’s pretty much like the guideline that I have to follow.
Niklas Buschner (07:34.019)
Hmm. Now I can imagine maybe people listening that are currently working as a freelancer or maybe consultant or maybe even like small agency that like if they maybe struggle and then applying for an in-house role that others could see it as a failure of their like entrepreneurial endeavors. So to say, what would you, what would you say to those people?
Maddy Osman (07:44.833)
Mm-hmm.
Maddy Osman (07:57.014)
Mm-hmm.
Maddy Osman (08:00.556)
Yeah, I I think there is a part of me that thought that about myself too, you know, like, is this me giving up on that entrepreneurial dream? And on one hand, no, because I’m still doing that to the side just in a different way. I think on the other hand, it’s like, the world changes a constant, right? And I think it would probably be more of a failure to not be willing to adapt with both
the changes in my life and the changes in the market and the changes in technology and things like that. And so it’s like, would I ever, would I never, you know, do something like this again, where I start a business and I work for myself, like I’m not ruling that out. I’m not really thinking about it right now. I’m not ruling it out. I think it’s like, it’s like there’s different seasons, right? In your life. And right now my season is.
I’m an entrepreneur, yes, but I also have to put the mom persona first. And I have to work in a way that supports that without me stagnating and without… I still wanna use my brain, obviously. That’s really important to me. And that’s something that I’ve talked to my work about.
I’m about to go on maternity leave, but I’m very anxious to come back as soon as I can because there’s only, knowing from the first child that I had, there’s only so much time that I can spend by myself raising a kid before it’s like, okay, time to get to work.
Niklas Buschner (09:45.487)
Sounds really empowering. think your journey can actually be really inspiring for people that might have started like a freelance career and now there’s all these changes in the ecosystem, et cetera. And like they feel like I can’t let go of that. Like applying in-house would be something that other people like think badly of me. Now that you’re a few months in your new role.
Maddy Osman (10:06.508)
soon.
Niklas Buschner (10:13.53)
Have there been any surprises going from agency founder, obviously still running it on the site, to an in-house content leader?
Maddy Osman (10:20.257)
to.
Yeah, I mean, I think it’s just been interesting, right, to go from, kind of like make your own schedule and you you get to choose exactly what you do every day. And again, like I still get a lot of that within this role. It’s honestly a perfect fit for somebody who’s following this path coming from running your own thing to then working in house because it’s very remote friendly. It’s very much like get work done.
You know, not a whole lot of meetings, having deadlines, but having the autonomy to meet them, you know, within however you structure your day. So I think that was surprising to me, like that I wasn’t necessarily giving a lot up in order to make this work. And, you know, just, just, yeah, like, like it didn’t feel like a total loss of the way that I like to work to, to
transition into an in-house role. I think it’s also like, and this is kind of a no-brainer, but it’s like, really miss working on a team with other people. I still ran a team, right? But it’s a different dynamic to be the boss or whatever. And a lot of the people that I work with were contractors, so it’s even a more different dynamic. But now it’s like we’re coworkers, we’re kind of on the same level. It’s more…
collaborative in a sense and people are less afraid, you know, to either say no or to like challenge you and I like that. I want it to be more collaborative and that’s how I learn. yeah, I mean a couple surprises, but mostly just that like it’s kind of working out, you know, exactly the way that I would want it to for my ideal work style, I guess.
Niklas Buschner (12:18.594)
Hmm. I think I know you’re passionate about and have been passionate about for quite some time already our content processes or like the process element of content. Your LinkedIn bio even says content process consultant. Can you enlighten us? What does that mean?
Maddy Osman (12:28.022)
you
Maddy Osman (12:31.574)
Yeah
Mm-hmm.
Maddy Osman (12:38.678)
Sure. Yeah. mean, I think at the end of the day, it’s like, something that I’ve learned by throwing myself into like this AI builder world is like the easiest way to go from not knowing how to use AI to then experimenting and like building things that actually work is the foundation in systems thinking. And so that’s kind of how I’ve built the blogsmith is on the backbone of
processes and checklists and standard operating procedures. I think, you know, one of the reasons I was able to scale the blogsmith from me as a freelancer to running it as an agency is by really building out a detailed style guide that represented the way that I think, that represented my writing style so that I could repeat across any new writer that we worked with, as well as any new editor that’s enforcing it.
so that’s been, that’s been like one of the more, I think, critical pieces of documentation that’s allowed me to scale, but it’s also things like, you know, what is our process for creating content in general from topic ideation to keyword research to outlining to drafting, to editing, to creating visuals. And so we use a tool called Process Street, which is just, it’s very flexible.
It’s very much like you can create conditional logic so that it shows only the steps that are relevant to the task and brings people in at the exact right moment and not before like their part of the workflow is ready. so anyway, all of this is to say that processes having structure translates really well to then building with AI and using LLMs, whether that’s even just like simple prompts in a chat.
where you’re prompting and you’re giving context or you’re building out repeatable workflows. And so, yeah, in terms of like the LinkedIn bio, I think I was kind of playing around with just like, yeah, different versions of myself, right? Where it’s like, okay, I’m this agency owner, but I also like have this, you know, content process focus. And so maybe, you know, that’s a new reinvention of self in terms of different services we could offer.
Maddy Osman (15:02.876)
And I was kind of playing around with that before landing on the idea to go in-house. And so it’s like, use all of these things now. But yeah, I think that that was a big learning for me was to understand that processes, documentation, being clear, being structured, plays really well with AI. And so I encourage people, if there is a process that you want to…
build with AI that you have to start with that kind of thinking.
Niklas Buschner (15:34.337)
Let’s dig into the systems thinking approach deeper in a second. But first I want to do a personal load because you said you were playing around with the LinkedIn bio. I really love that. And I think I never shared that, but a little anecdote of how we like transition the value proposition and like, let’s say the, the, the key statements, like not a slogan, but like the main headlines and like.
Maddy Osman (15:38.06)
Thank
Maddy Osman (15:50.22)
Mm.
Maddy Osman (15:59.177)
Mm-hmm.
share.
Niklas Buschner (16:02.679)
stuff at radiant was I was just going like logging into our WordPress backend from from our website and I was just going to the home page straight away and like just changing the headline and changing the sub line and a colleague of mine was making fun of me and like yeah you change it again it was like just two three times every week during a certain period because I felt like if I put it into a Google Doc it doesn’t like it doesn’t
Maddy Osman (16:10.028)
Mmm.
Maddy Osman (16:25.228)
you
Maddy Osman (16:28.716)
Mm.
Niklas Buschner (16:31.487)
Lead to the same thoughts and feelings I have when I put it on the home page It’s like okay. Now. This is in the this is life. This is like in in the in the public So and it challenged me to really think is this the direction we’re going to is it not so I just I love this idea of putting something out there Just seeing how it feels like sleep a night and then we revisit again. So yeah can can highly relate to that
Maddy Osman (16:34.316)
Mmm. Mmm.
Maddy Osman (16:39.626)
Right? Totally.
Maddy Osman (16:46.123)
Okay.
Maddy Osman (16:50.795)
Right.
Maddy Osman (16:56.768)
There is a certain psychology to just putting it out there and trying things. And I think whether it’s LinkedIn or your homepage, it’s forgiving, right? You can change it, you can get some feedback, you can look at the metrics or whatever, do some comparisons. And so why not just try and see where it takes you.
Niklas Buschner (17:19.587)
Yeah, I think if I would be a VC backed startup and would have raised like a hundred million dollars or something from like top 10 investors, I will probably be more cautious. But for me, like it always felt right. Let’s dig into the system thinking approach a little bit deeper because I can imagine that people find this very abstract, especially the people that have not like pursued this path a little bit more. So.
Maddy Osman (17:37.334)
see you.
Maddy Osman (17:44.566)
sure.
Maddy Osman (17:48.118)
Mm-hmm.
Niklas Buschner (17:48.961)
If you would have to teach like an intern or like someone that is like super junior, like maybe first role after college or something, how would you try to teach someone to start like thinking in systems?
Maddy Osman (17:51.862)
Yeah.
Maddy Osman (17:57.772)
Mm-hmm.
Maddy Osman (18:04.044)
Totally. Yeah. And I feel like to an extent I’ve had to do this with every role in my team, right? In order to be able to delegate it and to document it. So it’s definitely relatable. I think, you know, the beginning of it is first of all, just starting really small. Like I think it’s very tempting to try to take that whole content process workflow that I just described, drafting, outlining, researching, whatever.
and trying to think about that from start to finish. The problem with that is it’s very overwhelming. There are so many different parts to it. And even after I created that workflow, we’ve since updated it a year or two later, just going in and being like, here’s all the things we learned. And so it’s like lot of these things are different now. What would be better is to take a component and maybe even a smaller component than
outlining, right? Maybe it’s like outlining for a specific type of client because we did kind of, or we do have like different directions, you know, if it’s like a certain topic, right? Like if it’s maybe something in more of like a regulated industry like finance, it’s like there’s going to be a slightly different approach to outlining than if it’s more of like a lifestyle topic or a tech topic or whatever.
So, yeah, it’s breaking things into components and then trying to break them into smaller components and then going through and documenting. And there’s different ways you can go about that. There are certain tools that you can use. You can film yourself doing a loom, right, of the task itself. And then you can use the transcript of that. You could use AI to help you to take that transcript and to break that into steps or even just to guide you as you write it yourself.
You could use a tool like Scribe that records your screen and you can add comments and things like that. You could record a voice recording of you just kind of like doing the task and your commentary. So you kind of have to find the way that works best for your brain. I have found that just trying to write something from memory, even if you have, you know, a tab open with whatever, you know, it was you were going to do, it’s very hard because it’s like you have to confront
Maddy Osman (20:26.442)
your inner sensor, right? Like as you’re writing, you’re like trying to edit at the same time. And it kind of like gets in the way of just getting something done. So that’s, that’s some of the how. And then I think the other big thing, especially if you’re thinking about it as a freelancer or an entrepreneur, and it’s maybe more of an operational task than it is like a creative task or something like that. Although
there’s applications for both in terms of what I’m saying. But if it’s more of an operational task, it’s like, you have to kind of like put on that hat. So it’s like, whenever I was ready to expand the blogsmith in terms of a new role, so it started with hiring writers. Well, it started really with hiring researchers and then it was writers and then it was editors. And then it was, you know, things like SEO keyword researchers.
things like, you know, having account managers and project managers and then bookkeepers and someone to help me with social media. I tried to be that role for like a month and use, you know, what to basically give them to make sure that when I put that person in that role, they had something to do and that they would be able to do that job, you know.
to the standards that I hold myself to essentially. Because I didn’t have time for extra administrative headaches. I didn’t have time to be their manager necessarily outside of just like, here’s how that work is changing or here’s major feedback. But I needed to give them a structure because I couldn’t take on much more than just basic approvals with each new person that I added to my team.
Yeah, I mean, essentially it’s like you are that role. You have to document it really well. And then you have to see like how it lands and if it makes sense. And if you do have to make updates, which oftentimes you will have to do, but you want to create a solid foundation before you bring new people onto your organization.
Niklas Buschner (22:38.359)
And how would you approach prioritization for documenting things? Because I struggled with this myself and I still see a lot of people struggling with that because if you tell them, like if I tell a maybe little bit smaller agency owner I talked to, yeah, you should start documenting things. He tells me like, but then I have like…
Maddy Osman (22:43.537)
Maddy Osman (22:50.284)
Sure.
Niklas Buschner (23:04.335)
2025 things I have to document, like, and there’s all this stuff. So how would you approach prioritizing the most impactful things to document?
Maddy Osman (23:07.276)
Maddy Osman (23:12.245)
Amen.
Maddy Osman (23:16.844)
Yeah, I don’t know if this is going to answer it exactly, but I think the first thing again is breaking it down into smaller components. Because if you think that you have 25 things to do, you’re just not going to do any of them probably. Right. So it’s like, can you create one new like SOP per week, you know, or one, you know, every couple of days or. And yes, you should be thinking in terms of like, what’s going to be the most impactful for my organization, because there are going to be certain documents.
or processes that you use every day versus something like a month end creating financial statements, right? That’s gonna happen once a month or so. And so if you have to be a little bit more hands-on with making sure that fits what you were looking for, then maybe that process could wait, but it is something that you should still think about. One thing I think that helped me too is like, for a time the Blacks, but did have a full-time employee.
who we called her our director of content, she was helping with the client side. She was helping manage some of these different contributors. And it’s just having regular meetings with her, right, where we’re talking about the things we need. And the things that I need to do to empower her to do her job is also a really great, I think, reinforcing mechanism. So it doesn’t necessarily have to be an employee. That’s just the example that I had. But somebody who is checking in with you and keeping you honest.
I think is another good way to make sure it actually happens.
Niklas Buschner (24:50.063)
Now, speaking of systems thinking on a completely different level, let’s look into something you have built, which is a rack. So retrieval, augmented generation, upload agent in N8N4 podcast transcripts, which sounds, think for some people, super complicated, like WTF.
Maddy Osman (24:58.815)
Bye now.
Maddy Osman (25:11.244)
Sorry, sorry.
Right.
Niklas Buschner (25:17.039)
Can you quickly walk us through how that works?
Maddy Osman (25:20.78)
Totally, yeah. the project that you’re talking about, actually built for a client. So it was a very specific need, but it’s kind of a reusable system now. Like if I were to, you know, work with another similar type of problem, it’s like, kind of have a structure in my head for what that would look like. And so the problem that needed to be solved is this client works with a bunch of different,
podcast, like has like a podcast marketing agency. And so, you know, they’re always going to be putting out new episodes. All those episodes have a corresponding transcript that represent the text of what was said, the context really, right, of each episode. And it’s actually part of like a separate, bigger agent component that is referencing that context in order to then suggest things like
relevant, well actually for this one, it’s used to be able to respond to social media posts where somebody’s talking about a certain topic. And so you’re able to take the text of that, take that post essentially and say, yeah, we covered this on this episode of our podcast and have a little bit of a text blurb that talks about why this podcast was relevant to the topic being discussed.
it came from a need to be able to do that action. And it needed to be something where, again, it sounds really technical and complicated, but it needed to be something where this person, this client could sustain it himself, right, without going into N8N and having to add new…
podcast transcripts and also so that like if you needed to pass it on to a client, like they wouldn’t have to get into the backend either. And so the way I built it is like there’s basically a Google Drive folder and it integrates with Google cloud console or something like the Google cloud projects feature. And so it’s like Google cloud projects is watching this folder for any new activity and
Maddy Osman (27:39.636)
You can add Google docs for each new transcript. And it’s basically like the title of the doc is the title of the episode and the body of the doc is, you know, just like the transcript itself. And so it has to be, you know, kind of these simple components, but then I built a component on top of that, that also goes on the podcast website and it tries to find a relevant podcast episode URL because we need something to link to when responding to this social post.
And it just uploads that transcript to Pinecone for those who are unfamiliar with RAG. It’s a database. It’s a very contextual database that makes it kind of easier for these AI tools to search and find meaning than just like a straight up keyword search. And so it breaks down the text into chunks. And so there’s like a billion chunks, you know, for all these different.
transcripts and all the transcripts together. And then it also adds metadata like the podcast URL is associated with each chunk of text. And so then, yeah, when somebody has a social post and he wants to respond to it, for example, it’s pulling the most relevant context to create a response and to add that link. And yeah, as long as new transcripts are going into that folder, then they get to the right place.
which is Pinecone, which is storing all this data. And then Pinecone also then is suggesting the answer. So it’s like, yeah, it sounds very complicated even when I explain it, but that’s essentially how it works.
Niklas Buschner (29:19.767)
Okay. And do you feel like that? Yeah, no, I think you explained it perfectly well. But obviously for people that so it might still feel abstract, but I think it’s something that you can’t give someone like all the knowledge you care to understand it in like a three minute explanation. They obviously also have to have to do their own research. So sorry, guys for everybody listening. But what I’m interested in is
Maddy Osman (29:26.083)
thank you.
Maddy Osman (29:40.246)
Sure, sure.
Niklas Buschner (29:48.953)
Do you think that everyone working in content roles needs to get comfortable with stuff like that?
Maddy Osman (29:58.742)
think it’s worth at least exploring. And I guess like the last thing that I can say about RAG that’s maybe relatable, especially for content and SEO, is the whole point of RAG is that it’s grounding the AI response in reality, in true context, and in the context that you give it. And this is a really big turning point for me in terms of using AI in my job and like getting comfortable with it.
I definitely had my head in the sand for a while where I was thinking about, you know, AI is confidently incorrect. Like we can’t trust the outputs and that’s a problem, I think, in terms of using it to help with aspects of content marketing. as a brand, you don’t want to rely on something that is unverifiable and even rag. It’s not perfect, right? Like you still have to have your due diligence, but,
Yeah, a concept like that really helps me to get comfortable with spending more time thinking about AI, building with AI. And so to an extent, yeah, I mean, I think you don’t have to be like a coder, right, in order to get a rag workflow in place. There are a lot of tools like N8n, for example. But even, you know, there’s even like less complicated ways of doing
kind of what I said or components of what I’ve said. I think that, yeah, the future of content marketing is going to be this person who is both creative and technical. And I think it’s been trending that way for a while. It’s just becoming more and more important. The more that we learn about AI, the more that we start to implement it. And so I would say it’s a mistake to ignore it. I don’t know that you have to be.
learn development necessarily, but I think that you do have to be comfortable playing around with some of these tools and sort of like learning about like automation, for example, if this, then that and the different tools that you have available to build workflows.
Niklas Buschner (32:11.286)
Do you think that leaders should focus more on these skills when hiring either in-house or when choosing an agency?
Maddy Osman (32:18.878)
Yeah, I think that it should be something that definitely is part of the conversation. And I think even more than that, it’s about hiring for curiosity. It’s about hiring for people who are willing to get their hands dirty and just build stuff. It’s not so much expecting that whoever you hire is going to figure it out immediately, but it’s giving them…
the time and space to figure stuff out and to try things and to talk about it with the team. One thing I really appreciate about DigitalOcean is they kind of have, how do I say this? It’s very much encouraged to use AI and to play with it. And I think especially because DigitalOcean’s offering has really evolved to encapsulate being a place where AI builders can build applications.
But within individual teams, it’s expected that you’re going to play with AI tools and see what you can build and use them to leverage more efficiency. so one thing that was kind of fun recently is they had a build day where was just like go and experiment and see what you can come up with. I think every company should do that.
Niklas Buschner (33:38.702)
Now there’s this concept in AI workflows of human in the loop. For everybody that is not familiar with the concept, can you quickly explain it to us?
Maddy Osman (33:49.536)
Mm hmm, sure. So human in the loop is basically, I guess, a form of a guardrail, which is like, at some point in this workflow, a human needs to be involved to sort of like review or like accept or reject the output of the workflow or an interim step of the workflow. And what else is worth saying about that? Yeah, basically before
before you take that output and do something with it, especially if it’s like a client facing thing.
Niklas Buschner (34:22.698)
And where’s the line for you? Like when do you also, when you do your systems thinking, when should humans intervene and when should you let automations or agents just run?
Maddy Osman (34:29.312)
Uh-huh.
Maddy Osman (34:36.396)
Yeah, so I mean, I think they should always intervene if it is client facing. That’s not to say that you can’t run a full workflow before then reviewing the output, right? Like I’m not necessarily, how do I say this? Like I’m not necessarily encouraging that you have to have a bunch of interim steps for like human in the loop within a workflow. It’s like have that workflow run.
have a way to get it in your face, you know, so that you have to interact with it in some way versus like it runs and then it just sits there because, you know, something else distracted you in your day. I mean, I think there is a big line for me right now with just kind of like running something and then using it for whatever the intended purpose is. There has to be some sort of review in there. And oftentimes for me, it’s pushing that output to Slack.
because it’s like, know I’m going to check that and it’s just, it’s convenient. works with my workflow for you. It might be something else, right? Like it’s like, what tools are you using on a normal basis? It could even be your Gmail or something, but like getting it off the tool where the automation is occurring, for example. But yeah, I would err on the side of having more human in the loop components versus less. And then over time,
You start to test things, right? And you see what happens if, you know, I have a workflow with maybe three points of humans interacting with it. What if I take one away? Do I still get the same results? so removing those guardrails is really the result of testing. And it’s also the result of technology getting better, AI technology.
Niklas Buschner (36:23.137)
And what would you say to people that if they hear like you should always have a human in the loop, maybe you should even have like multiple checkpoints. What would you say to people that then say, yeah, but what about like the whole beautiful AI productivity gain then? Like it’s all going away if I have to still like let my team check this stuff like all the time at like these.
Maddy Osman (36:34.156)
Thanks
Niklas Buschner (36:51.545)
multiple checkpoints, what would be your response to that?
Maddy Osman (36:55.508)
Yeah, I mean, would say it’s just like flawed logic, honestly, because it’s like, I mean, it’s just like, at the end of the day, like, AI does speed up a lot of workflows. I don’t think that having a human review them changes that fact, right? Like, I think you have to build in that understanding that a human should review the output with the state of where AI is right now. The fact that it is confidently incorrect, the fact that it is
non-deterministic, so you’re always going to get different results. It’s just kind of the reality of how we use AI right now. But yeah, mean, for example, another workflow that I built for this same client is it’s taking relevant news from the past 24 hours and creating social posts using a specific style guide. There’s a couple of other pieces to it, but it’s like, do I have time to go out and like
look at all this news and decide if it’s relevant and then also reformat it in a way so that it works that this client could post it on his clients social media profiles. I really don’t. And so it’s like just the fact that this workflow is running means that the work is getting done. Whether he uses it or not, whether he reviews it or not is his choice ultimately. But he has…
the results of that workflow being delivered to his inbox every day. So he has to confront it in some way. And yeah, he could post it or not, but like none of that work would have gotten done if not for this workflow or, you know, it would have been a major time suck. And so it’s like, it’s like people who maybe have that question have to maybe reconsider what they mean about ROI or like, you know, like what should be or shouldn’t be. It’s like at the end of the day,
it’s already a much more efficient workflow than it was.
Niklas Buschner (38:52.475)
What are you most fascinated about with AI?
Maddy Osman (38:56.3)
Probably the rag concept and continuing to figure out, like I think that’s like my 2025 and really my 2026 goal is like, how can I take what I know about this and apply it to what I’m doing in content marketing to make sure that whether it’s for my clients, the blogsmith or the work that I’m doing at DigitalOcean, that I’m creating content in an efficient way that’s really rooted in the reality of the brand.
I think there’s, I think there’s a lot to it. And I think it’s like, one of the things I maybe struggle with is I think that rag has been used in a lot of different ways in a lot of different departments, but I haven’t seen a lot of marketing use cases. So I feel like I’m trying to, to figure that out without a lot of guidance. And I’m the type of person who I learned best when it’s like sharing stories and experiments and things like that. So it’s.
maybe getting a little bit outside of my comfort zone to just experiment myself and try to figure out how to apply it there. But I think it’s a worthwhile use of time to experiment in that way.
Niklas Buschner (40:07.867)
super interesting because I feel like that people that are more advanced in the use of AI, they often rather turn to these like concepts where you have a lot of base knowledge and like you have a lot of also input thinking like what’s the best input for this workflow, et cetera, whereas people that are just starting out their AI journey, so to say.
Maddy Osman (40:16.661)
I know.
Niklas Buschner (40:34.787)
still see the AI itself as like the super brain where like all the all the stuff is coming out whereas other people more see like a synthesizer in the middle like you have to have really strong input and then be able to like use database concepts etc. Do you also have you also observed this?
Maddy Osman (40:38.677)
Maddy Osman (40:50.379)
Right.
Maddy Osman (40:57.876)
That’s a great question. think it’s kind of different for different projects. But I think, think, yeah, it is kind of more of like that middle, middle man middleware, whatever, where it’s like, you have to be the way that I’ve always thought about AI, at least thus far is like, you have to be an expert operator in order to get the best results. Because otherwise, you ask mediocre questions, you get mediocre outputs, you know, garbage in garbage out or
whatever people say. And then another concept that maybe changed AI for me is the idea of vibe coding. So I have a background as a web designer, so I have some coding knowledge, but I don’t have software coding knowledge. And I’ve just been having a lot of fun using tools like Replet or Bolt or Lovable and just building little components, whether it’s just like
kind of for fun, to see what I could do. Or if it’s something that’s helpful for me, either with the Blogsmith or with DigitalOcean components that can help me do my job better. And in that case, it’s like, I’m asking the AI, how do I build this specific thing? How do I describe it? Because I don’t want to just brain dump the idea in one of these tools. I want to have a structured way of thinking about it.
And so in those ways, it’s maybe, it’s kind of leading me, but I’m also still leading it because I’m saying what I want, right? And then I’m evaluating what the AI is telling me and taking some of what I like and, you know, leaving some of what I don’t. So yeah, I really do think it kind of operates in this sort of like middle layer, right? Where it’s like, you have to give it a lot of guidance to get what you want out of it.
Niklas Buschner (42:50.673)
I love the expert operator idea. Maybe little side story. I also told a couple of my colleagues. There is this car rental company from Germany. called SXT. They have actually also expanded a little bit into the US. They were probably not as big as Hertz or the big US brands, but still they should be there at the big airports. And the old SXT
Maddy Osman (43:02.444)
Mm. Mm.
Niklas Buschner (43:20.591)
So he’s now, I think, 80 or something. Like, I think in the 80s or in the 90s, and he told the story on a podcast, they had to build a software for themselves, like with dispatching the cars and doing the yield management and all that stuff. And he decided to just learn coding himself and build the software himself because he said,
Maddy Osman (43:43.776)
Mmm, wow.
Niklas Buschner (43:47.537)
His fundamental belief is you have to know the business inside out to be able to make the right choices and the right calls in the software. And I feel like it’s a full circle moment now with AI and Vibe coding because a lot of people like the key missing piece of the puzzle has always been the coding knowledge. And obviously there’s also a lot in terms of infrastructure thinking, et cetera, like how do I make something scalable?
Maddy Osman (43:51.372)
Hmm
Maddy Osman (43:56.49)
Yeah.
Uh-huh.
Maddy Osman (44:10.603)
Right.
Maddy Osman (44:15.478)
Totally.
Niklas Buschner (44:17.222)
But in the end, it’s exactly this expert operator idea. If you know the things inside out now with AI, you have the tools to actually build what you previously could have only imagined. yeah, it’s, it’s, it’s a car and the story. anyway, scaling content is something that I wanted to talk to you about also because, scaling content is actually way easier now with
Maddy Osman (44:21.718)
Mm-hmm.
Maddy Osman (44:30.868)
No, 100%.
Niklas Buschner (44:44.762)
already been talking about AI workflows, et cetera. people, this also why I brought like the devil’s advocate question about like the AI productivity gains are all gone with humans in the loop. I just want to have my agents that are creating like hundreds of pieces of content every month and like get it out there. So what would you say, what are the biggest mistakes you see content teams making when they try to get content, especially now that AI makes it easier than ever?
Maddy Osman (44:53.174)
Stay.
Maddy Osman (44:59.998)
say.
Maddy Osman (45:08.086)
Mm-hmm.
Maddy Osman (45:11.798)
Mm-hmm.
Maddy Osman (45:15.36)
Yeah, I mean, it’s falling into that trap, right? That you can do it so easily. So a couple things. I think the first thing is, again, going back to that expert operator idea and really thinking about the inputs that you give it and starting from a place of originality, right? Where it’s not just like, hey, come up with this idea and then come up with a draft. It’s like, you have to have the idea.
and you have to give it some context. like, for example, something that I like to do before I write pretty much any article is I like to just like do a voice recording of myself like rambling through the topic. You know, it’s like, okay, so this is the topic, you know, that I’m gonna write. And, you know, here are some of the things that I know that I need to cover, whether that’s a brief from a client or just maybe some things that I know about the topic that I wanna make sure to include.
and it’s going to need this section and I’m probably going to need to, you know, research a stat about that to substantiate my point. And I remember, you know, this was something that was being talked about on LinkedIn recently that people have a lot of opinions about. So that’s, you know, something to weave in just from like a human interest standpoint. And so it’s like, again, like confronting that inner sensor. If I sat down and wrote this, I would get in my own way. But if I could just talk through it.
What AI is really good at is then organizing those thoughts into more of a brief or more of an outline. And from that point that I might use AI to flesh out section by section, there’s almost no world where I would have AI take that and make a draft though. It’s still kind of like, just like without AI, it still has to be a step-by-step process in order to get a really good end result.
I think that that’s a fundamental flaw of AI. At least the AI that we have right now is this idea of going from start to finish, whether it’s a really good prompt in context or it’s your army of agents or whatever. I just don’t think that we’re there yet. And I’m not saying that we’ll never be there. I’m just saying I wouldn’t rely on that. And so another example or way to look at it is at DigitalOcean, we do a lot of
Niklas Buschner (47:11.25)
.
Maddy Osman (47:38.912)
competitor comparison, like listicle pieces. And so they are somewhat formulaic. We’ve really spent a lot of time creating good guidelines for what we’re giving to writers in terms of like, you know, this is the structure we’re looking for. This is the why, and this is like a checklist of like what we expect to see. And so again, like this process documentation is the backbone of any sort of AI systems we could build. And so I kind of ran it.
an experiment recently where it was like, okay, I’ve sort of like written one of these by hand, right? Like without significant AI help. But then I had another one that was like very similar that I had to write. And so I was thinking like, okay, well, how can I use AI to speed up these like really formulaic parts of the process that we’ve put a lot of thought into, you know, this is how this component goes and this is how this component goes and.
and we have really great documentation. So foreseeably giving that to an AI system is going to be get really good results. And so I started to create like custom GPTs and we use Gemini. So like the GEMS version of that within like our ecosystem. And so it’s things like, okay, this component now has this custom GPT, this component has this custom GPT. And so again,
I’m not trying to create this article in one go, but as I go to each section, I give it the context of the article so far, or even like a previous article that I wrote that’s gonna be very similar to that. And it has all this context and it knows that I’m not looking for exactly the same thing, but that I’m looking for the context of the topic I’m working on, but the backbone of the information from the guidance. also created like, here’s like,
a world of like links about all of our competitors that you can use for context and you can use to kind of match up. And so when I turned that article in, there were very few edits on it because it just like, you know, it like understood, you know, what I was looking for and, it just kind of worked. And again, I didn’t try to do it all at once. It was still a very involved process, but it was a
Maddy Osman (50:02.636)
it was a lower lift, know, it was like easier on my brain to get it done. And even just like checking through it’s like, yep, that’s exactly like what I would have written myself.
Niklas Buschner (50:13.447)
Now I have to get your expert opinion on a client request that tapped into like this AI making content scalable, et cetera area. The request was, I’m paraphrasing here, why can’t we just take like the top 50 pieces from our competitor based on like SEM brush or whatever.
Maddy Osman (50:24.661)
Anyway.
Maddy Osman (50:38.892)
So.
Niklas Buschner (50:43.357)
based on traffic, as long as they feel relevant. So if it looks like something that is completely irrelevant to us, let’s skip it. But why not just take those and then give it to an AI and then rewrite it? Shouldn’t this also be a solid strategy? What would you respond?
Maddy Osman (50:47.616)
Mm-hmm.
Maddy Osman (50:57.6)
Mm-hmm.
Maddy Osman (51:05.548)
would say first of all, sounds like plagiarism. And so that would be my biggest ethical line. I think this is something that we covered too on the Blogsmiths blog, an article that I was really passionate about, which is like, yeah, is AI plagiarism or whatever? And I think at the end of the day, it’s like as humans,
Niklas Buschner (51:11.728)
No legal advice.
Maddy Osman (51:30.176)
there’s that book, Steal Like an Artist, right? Like we’re always learning from our environment. So it’s impossible to say that like we’re not inspired by the work of others. Like you should be, but there’s kind of like a way in terms of how you implement that knowledge. And it’s like, if you just kind of like blindly like copy paste this URL or the context of, you know, this article into an LLM, it’s like, where is the human judgment though, in terms of like going through it yourself and like,
coming up with your own connections, right? And like that sort of like preponderance of the content itself. If you’re just giving it to an AI, it’s like, at what point do you use your own judgment? And so I think that’s a problem is that you’re not taking the time to sort of like figure out for yourself why it’s relevant or not, or like what parts of it are useful for that informing your take. I mean, it’s…
It’s not really giving you an opportunity to create a point of view. And AI can help you with that, right? Like you can give it brand guidelines. You can give it, you know, different facets of like your voice or whatever. But it’s like, do we, mean, I think it comes back to the idea of like, do we really want AI doing like all the fun parts of our job? You know, like it’s here to, again, like the competitor comparison listicles. There’s only so many ways that.
Like I can take joy out of like rewriting the same thing five different times, right? From like just like slightly different angles. But it’s like coming up with a topic that has been covered before, but from the unique angle of the brand that I’m writing it about, like that’s like the part that I actually enjoy about, you know, this writing process and creating content. So, so yeah, I would say on one hand it’s, it’s an ethical dilemma. On the other hand, it’s like, then what
What point is there in like employing me as a human to do this job? What was the other thing I was gonna say? One of the guidelines that we give at DigitalOcean and at the Blogsmith is like, you know, what is it about this article that you’ve written that somebody couldn’t just go to an LLM and get themselves, right? And that really is like the…
Maddy Osman (53:51.858)
unique angle, the personal perspective, you know, trying to make it relatable or interesting or adding a subject matter expert insight that’s original. And so I would say like, those are the problems that I would have with doing something like that. And then the last thing that I’ll say too is like, it’s been easier than ever to connect SEO data with like your favorite LLM chat using something like Model Context Protocol MCP.
but again, that’s another point of failure potentially, you know, if the AI is hallucinating data, it’s not perfect yet where, know, it’s sort of like one-to-one going into that dashboard and getting the data yourself. Sometimes the data itself is flawed and needs to be checked. And so if you’re just, again, living in quad or chat GPT at this point in time, that can, that can create some major blind spots, even just with like the data itself. And if it’s.
totally relevant and accurate.
Niklas Buschner (54:52.595)
Got it. Now I want you to challenge at least part of the answer I have given to them and tell me maybe what you think about that. So here is it. I said to them, hey, you know, if we take the competitors content and we basically copy paste it, it’s basically just a derivative of what they have already created. So it would feel strange to me to expect
Maddy Osman (55:00.352)
Yeah, let’s hear it.
Maddy Osman (55:17.866)
Anyway.
Niklas Buschner (55:21.671)
that our content could perform better than those. So are we really going to settle for being number two right from the get-go? This does not feel like an ambitious target to me.
Maddy Osman (55:34.86)
No, I love that and I should have brought that up too. It’s just the fact that the more that we feed stuff into AI, it becomes recursive almost. Or even if you’re not using AI, but you’re still settling for rewriting what’s already out there. If you’re never adding something original, if there isn’t anything new, then
then yeah, why would you expect that to rank and why would you expect it to perform? So I think that’s probably like the answer that a brand needs to hear. And then my answer is more about like, but think about the ethics too.
Niklas Buschner (56:18.11)
Yeah, yeah, I felt like the people asking me were very like conversion and like performance driven. So obviously this is a context that I could have given you earlier, but I didn’t want to like make the question too overly complicated. So I feel like your point is also very important. And I mean, it’s the same with like the beginning of the internet where you could do a lot of stuff, but
Maddy Osman (56:26.966)
day.
Maddy Osman (56:32.918)
sure.
Niklas Buschner (56:45.896)
you shouldn’t do everything that is possible. like reminding us of ethics, us of some basic legal principles. of plagiarism is definitely an important part of a comprehensive answer. Now I want to tap into obviously your unique knowledge from like being
Maddy Osman (56:45.9)
shit.
Maddy Osman (56:49.31)
Right.
Niklas Buschner (57:12.316)
an agency owner and then also working in-house again because I feel like that a lot of brands like they aren’t really sure which roles to keep in-house which stuff maybe to hire an agency for so as you have worked both sides now what’s your opinion like what should be kept in-house and where can an agency be most impactful?
Maddy Osman (57:15.446)
you
Maddy Osman (57:25.868)
Anyway.
Maddy Osman (57:29.43)
to.
Maddy Osman (57:38.24)
Totally. Yeah, I think the really simple answer to this question is like in-house is like the master of the brands and probably also the master of the strategy. Although I think as you get to smaller operations, they might lean on an agency to help supplement that or to set it, especially if it’s just for a specific capability that they don’t have in-house. Like if they don’t have, you know, an SEO team or maybe they just have one person.
then strategy might be something that they lean on an agency for. I think where an agency is really impactful or a freelancer really, or multiple freelancers, is the execution side of things, being able to sort of like scale up or down according to current needs, to be able to address certain capabilities they don’t have in-house. Because maybe you have somebody who handles like your blog writing, but then you have a research report.
Or maybe you need to write a white paper or an e-book or something, and that’s a slightly different skill set. so agencies can just supplement essentially whatever it is that you don’t have access to in-house and in a really, I think, cost-effective way, more so than onboarding, hiring, going through all that stuff.
Niklas Buschner (58:59.582)
Got it. Now, wrapping up the conversation slowly, I want you to look into the glass ball with me. So where do you see, you obviously already shared a couple of thoughts on that, but where do you see the content marketer role heading in the next, let’s say two to three years?
Maddy Osman (59:04.652)
See you
Maddy Osman (59:12.086)
Thank
Maddy Osman (59:15.724)
Mm-hmm.
Maddy Osman (59:23.372)
Yeah, I mean, I think it is this technologist side is going to be important. I think, know, as like a practice as an industry as a field, whatever we want to call it, it’s like content marketers really need to sort of heed the call of like AI and, get their heads out of the sand, you know, kind of like, like I did, I really
I really got stuck in this perspective of like, you know, like I don’t want to use it. It’s, it has all these like negatives. think anybody who’s still holding on to that, you’re really at risk of your, of it just being hard to make money in this profession in the next one to two plus years. And so my recommendation to people like that would be to just go out and experiment and figure out like what flavor of AI.
meshes with kind of like your line in the sand, your ethics, even just like your enjoyment, right? So yeah, think that’s the biggest thing is that you are going to have to be sort of increasingly more technical. You are going to have to be willing to change and frequently, whether that’s within your own freelance agency.
whatever world or even going in-house. mean, things change with the guidelines we’re creating at Digital Ocean constantly. Some things are foundational and they more or less stay the same, but the way that we approach problems I think differs even in just a few months that I’ve been there. So yeah, it’s being comfortable with the ambiguity, it’s being curious, it’s being willing to learn, it’s being willing to build.
I don’t know, it is hard though to look into the future and see exactly what that looks like. But I think it’s fair to say that the skill set needed to succeed will continue to change. And so you have to put yourself at the forefront of what those changes are. You have to try to anticipate them. You have to follow the trends to an extent. It’s not necessarily chasing every shiny thing, but having some discretion in terms of what
Maddy Osman (01:01:44.428)
what’s impactful, what brands care about, right? And at the end of the day, like what helps you do your job really effectively in terms of both efficiency, but also in terms of quality.
Niklas Buschner (01:01:59.253)
But it sounds like then that it comes down even more to mindset than it comes down to actual skills. Is that fair to say?
Maddy Osman (01:02:09.228)
think so. I think even from my own experience, it’s really easy to get in that sort of doomsday, like, will it even matter seven years from now and the robots take over and money doesn’t matter anymore or whatever. I think that’s getting really ahead of ourselves. so it’s like, I don’t know, to some extent you have to kind of just live each day, right? And do what you can with it and try to use it to…
learn and to experiment and to stay positive for sure. We don’t know what the future holds, so it’s kind of an exercise in futility to forecast too far ahead. I think that’s hopefully something to protect your mindset. Don’t go too far down any rabbit holes. Just learn what you can, follow the trends, see where it takes you.
Niklas Buschner (01:03:08.03)
Sounds good. Do you have anything that helps you personally like keep this like optimistic view of the future? Because I mean, not only with AI, but somehow also with all the stuff that’s going on in the world. I have the feeling that there are people that tend to have a more pessimistic view of the future than maybe five years ago or something.
Maddy Osman (01:03:15.82)
Mm-hmm.
Maddy Osman (01:03:25.803)
Yeah.
Maddy Osman (01:03:34.198)
shave.
Niklas Buschner (01:03:37.542)
Is there anything that helps you in keeping also this optimism?
Maddy Osman (01:03:37.586)
fair.
Mm-hmm.
I think in a way it’s like a forced optimism because as a parent now, you know, I’m bringing new life into this world and I want to be a good example for this new life. And I also want to like build the world that they will inherit, right? And so it’s like being a good citizen, standing up for what I believe in. And yeah, just trying to like pave that path for them. It’s like,
It’s like I have to be positive because otherwise like, why am I doing this? You know? And that’s not everybody’s experience, right? And I’m not saying that you have to be a parent in order to think that way, but I think it definitely, it changes you and it forces you to think and act in a way that meshes with that reality.
Niklas Buschner (01:04:37.055)
That’s extremely powerful. Thanks for sharing that. Now we have talked about like all sorts of stuff from like content to M8N to REC, humans in the loop, car rental companies building their own software and stuff. Exactly. And I always keep one question for the end, which is,
Maddy Osman (01:04:45.101)
yes.
Maddy Osman (01:04:51.931)
yeah.
Maddy Osman (01:04:55.456)
all the important things.
Maddy Osman (01:05:03.414)
Mm-hmm.
Niklas Buschner (01:05:06.355)
what should we have talked about that we didn’t talk about?
Maddy Osman (01:05:10.47)
That’s a great question. Maybe like our favorite.
Niklas Buschner (01:05:13.277)
I stole it from Lenny’s podcast. I am giving credits, this is not plagiarism.
Maddy Osman (01:05:18.132)
not plagiarism, yes, always cite your sources, which by the way is an excellent podcast and community and how I’ve been able to afford my vibe coding addiction by using all the tools that he includes in his product path.
Niklas Buschner (01:05:25.95)
Yep.
Niklas Buschner (01:05:35.145)
Yeah, this is actually super, an incredible value for money.
Maddy Osman (01:05:41.022)
Absolutely. I would say like books we’re reading or it could even be articles or something like that, something that’s resonated with you and then I can answer the question too recently. Things of interest there. Because that’s how we learn, right? It’s just different resources. It could be a course. So I’ll pose it to you and then I’ll think about my answer while you answer.
Niklas Buschner (01:06:11.189)
it’s a good one. I wasn’t prepared for that. So I would actually say that. Yeah. Yeah. Yeah. No, no. I definitely think that it has actually been something that I listened to. I would say it’s probably a couple of months ago, but like this whole story with like the, the rental, the car rental company, because like, I’m not particularly interested in cars. have nothing to do with the car rental business, but.
Maddy Osman (01:06:15.596)
I’m putting you on the spot.
Maddy Osman (01:06:26.604)
Mm, okay. Okay.
Mm-hmm.
Maddy Osman (01:06:39.692)
sure.
Niklas Buschner (01:06:40.597)
this way of thinking, the mental model of it’s better if I learned coding because I’m an expert in the business than getting an expert in coding and teaching him the business because the context of the business is much more complicated. At least these were his words. Unfortunately, the episode is in German, but I hope I gave it a TLDR about the
Maddy Osman (01:06:47.752)
Mm-hmm. Okay.
Maddy Osman (01:06:54.924)
share.
Maddy Osman (01:07:03.016)
Hahaha.
Niklas Buschner (01:07:10.229)
But like these stories that have basically nothing to do with AI at first sight But that tie to like these fundamental principles they really inspire me and I try to think more about those so That’s definitely one thing that Really stuck with me. Now. How about you?
Maddy Osman (01:07:17.292)
shame.
Maddy Osman (01:07:29.248)
Yeah, no, I love that. And I think that’s a good thing to keep in mind in general is like sometimes the best inspiration comes from like outside of the things that are strictly about AI or marketing or whatever. And I’m sure that the founder of 6T would have loved vibe coding if it was like available to him at the time because it would have then got him much quicker to…
to his prototype that he then could have shipped to a software dev to finalize.
Niklas Buschner (01:08:02.313)
I think they’re still doing decently well. I think they’re doing a couple of billions in revenue.
Maddy Osman (01:08:07.116)
Yeah, they’re doing fine. A book that I read, I don’t know if it’s like the most impactful, but it’s the one that comes to mind for that question, is this book called Co-Intelligence. And I think it was written a year or two ago. Like it’s not like it was just published, but it’s from a professor, I think from Wharton in the US. And it’s just about like kind of how we think about how we work with AI.
and how we can make it mesh in our existing paradigms and to some extent what students should think about because that’s his perspective that he’s bringing into it. But there was just something about the book that really resonated with me in terms of coming to terms with using AI and making it work for my mindset. I’ve read some other books since then about writing an AI or different
industry or domain specific books, but it was this one that was more general that was like the most useful to me, think, in like developing my mindset about it.
Niklas Buschner (01:09:16.989)
Awesome. We will put the link to the book in the description. looks very interesting. thanks for sharing that. Maddie, has been an insanely insightful conversation. I learned a lot. I, also feel like that there was a lot in it for people to take away. So I once said,
Maddy Osman (01:09:31.202)
good.
Maddy Osman (01:09:43.114)
I hope so.
Niklas Buschner (01:09:44.855)
that my goal for this podcast is always to make it entertaining, then informative, and then also practical. So I think these are three dimensions that at least maybe I’m not scoring a 10 out of 10 in all of them, but at least if I’m like, I have this clear north star, then I can make progress. And I think today we are really close to a 10 out of 10.
Maddy Osman (01:09:50.636)
Okay.
Maddy Osman (01:09:54.282)
Absolutely.
Maddy Osman (01:10:01.132)
That’s right.
Maddy Osman (01:10:10.432)
Nice.
Niklas Buschner (01:10:11.88)
It has been a lot of fun. Thanks so much for coming on. If people want to hear more about the stuff you’re doing and read more about your thoughts, etc. What’s the best place to follow around?
Maddy Osman (01:10:13.782)
Likewise.
Maddy Osman (01:10:21.238)
Mm.
Maddy Osman (01:10:29.196)
Totally, my AirPods keep falling out. I need to change the little like bud size, but the next one up is like too big. I don’t know. But anyway, LinkedIn is probably the best. So it’s Maddie Osmond French. I tried to keep it Maddie Osmond, cause that’s my maiden name. And you and I talked about this, but you know, it’s like they changed it when I changed my last name and.
Niklas Buschner (01:10:53.045)
Yep.
Maddy Osman (01:10:57.068)
and they periodically drop one of those names. So anyway, just type in the full name and you’ll find it. And then, yeah, if you want to maybe dive deeper into my brain, I would definitely recommend checking out my book, Writing for Humans and Robots, The New Rules of Content Style. It’s a little bit more of that formulaic, process-driven approach, and it’s also a LinkedIn learning course now, if you would prefer to watch instead of read.
Niklas Buschner (01:11:23.496)
Awesome. Okay. We’ll also put the link to that in the description. It has also like superb, Amazon reviews, as far as I can see it. so definitely big recommendation. thanks so much for coming on. Thanks so much for making the time for sharing, a lot of insights. I wish you all the best for, your, role in digital ocean. I wish you all the best for the.
Maddy Osman (01:11:42.675)
Absolutely.
Maddy Osman (01:11:51.884)
Thank you.
Niklas Buschner (01:11:53.354)
the second child. This is probably the most important part here. Yeah, and other than that, just like keep your optimism and I think people can learn a lot from that. I did. So thanks so much for today.
Maddy Osman (01:12:06.143)
Absolutely.
Maddy Osman (01:12:10.102)
Yes, thank you and thank you for all the great questions. They were a joy to answer.
Niklas Buschner (01:12:15.284)
My pleasure. So, see you, bye bye!
Maddy Osman (01:12:19.072)
Bye bye.
