# Mastering Problem-Solving in Tech [#03 Nate Herk]

## Метаданные

- **Канал:** n8n
- **YouTube:** https://www.youtube.com/watch?v=Hy47hfRPK2k
- **Дата:** 18.10.2025
- **Длительность:** 39:23
- **Просмотры:** 6,947

## Описание

Summary

In this conversation, Dylan and Nate Herk explore the intricacies of automation, problem-solving, and content creation within the N8n community. They discuss the importance of logical thinking when faced with unknown problems, the power of automation in enhancing productivity, and the value of community engagement in content creation. Nate shares insights on overcoming technical barriers, mastering automation fundamentals, and recognizing patterns in error messages to improve efficiency. The discussion also touches on the future of automation and the evolving landscape of AI tools.

0:00 2026 demand & why logic beats code
0:17 Start simple—avoid “too agentic” first builds
0:35 Master ~15 core nodes to cover 97% of use cases
1:06 The hard last 20%: testing & prompt/model swaps
1:28 Meet Nate Herk
1:48 Discovering n8n (Alteryx → “send it to ChatGPT”)
2:30 Learn-by-teaching: how the channel started
3:07 Why teaching boosts retention
3:41 Sourcing ideas: community, problems, data
4:31 Niche vs reach: conversions vs views
5:46 Community themes → video topics
6:33 Multi-agent research workflow for titles/thumbnails
7:35 What Nate automates vs keeps manual
8:10 Staying on AI news & tailoring ideas
8:32 RSS, trends, and niches inside n8n
9:28 Positioning & audiences (beginners → inspiration)
10:38 From education to monetization paths
11:22 The four learner buckets (+ SMBs DIY)
12:48 “I’m not technical” → mindset shift
14:00 Uninformed optimist → informed pessimist loop
15:30 Map the process; one step at a time
16:06 Text-to-workflow helps—don’t rely on it for prod
16:43 New → familiar; SOPs & whiteboarding
18:12 Community win: rebuild from true understanding
19:43 LEGO analogy for workflows
20:39 Template demo at a meetup
21:30 Methodical debugging vs shotgun rage-quit
22:18 15 core nodes & 5–7 common error patterns
23:44 Solving unknowns: read, search, ask AI with context
25:29 n8n assistant as a backup troubleshooter
26:22 What to keep manual; where automation shines
28:18 90/10 rule: AI drafts, human finish
30:02 AI thumbnail system that impressed Nate
32:01 Next 3 months: natural-language builds & computer use
34:08 Production-ready = testing + guardrails
35:01 n8n’s mission: 10x coder power
35:55 Think like a human when stuck
36:37 Map it; forums, friends, and community
37:12 Painful lessons stick—share them
37:37 Tool-agnostic skills transfer
38:57 Where to find Nate (YouTube, community, LinkedIn)

Takeaways

The mission of N8n is to empower technical teams.
Logical thinking is crucial for problem-solving.
Painful problem-solving experiences lead to lasting lessons.
Teaching others reinforces your own understanding.
Community engagement is vital for content creation.
Automation can significantly enhance productivity.
Technical skills are not a prerequisite for using N8n.
Pattern recognition aids in troubleshooting errors.
Mastering a few core nodes can automate most tasks.
The future of automation is focused on natural language processing.

#n8n #n8nMasterclassPodcast #podcast #masterclass 
The n8n Masterclass Podcast

## Содержание

### [0:00](https://www.youtube.com/watch?v=Hy47hfRPK2k) 2026 demand & why logic beats code

Those are the skills that are going to be in extremely high demand heading into this new year of 2026. It's crazy to me how much easier it's gotten. If you are coming from an absolutely zero technical background and you're opening up Nen, you're already going to be overwhelmed probably just because the interface is new. But once you get past that, you realize fundamentally it really just

### [0:17](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=17s) Start simple—avoid “too agentic” first builds

comes down to logic. If you can do something manually and you know the process really well, you can automate it because it's just one step at a time. But I think maybe where people get into issues is they go straight into a process that's very complicated and very agentic rather than kind of taking those baby steps. When I really think back at

### [0:35](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=35s) Master ~15 core nodes to cover 97% of use cases

all the automations and agents I've built, I think it really boils down to like maybe 15 core nodes. And if you can master those 15 core nodes, you can automate probably 97% of anything really. When I think about building workflows, the last 20% is always the toughest because you basically need to test tons of different inputs and then you need to test all those inputs again with a different model and then you need to do it again with a different system prompt and you know all these changes. That's really going to be the more important skill rather than

### [1:06](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=66s) The hard last 20%: testing & prompt/model swaps

just building workflows is going to be making them actually ready. Now that you have this knowledge of a automation, what are your options? you know, how can you move from here to become a more valuable worker to pick up some side income to maybe even start your own business? Now is the best time to be getting into it. — Nate, what's up, buddy? Welcome to the

### [1:28](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=88s) Meet Nate Herk

podcast, brother. — How's it going, Dylan? Good to be here. — So, what first got you to start making content around NAN? — Yeah, absolutely. You know, when I discovered NADN, I was working my full-time job and that was pretty much automation. So I was you know building things with no code tool called alterx and I remember having this one product

### [1:48](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=108s) Discovering n8n (Alteryx → “send it to ChatGPT”)

that I was working on where I hit this decision point in the automation and it took a level of reasoning that typical logic couldn't and I remember thinking to myself man I wish I could just like send this data to chatbt and then just have catchb it back and then the automation continues and then I looked into it and I was like okay that's a automation it is something that exists and it is possible. So that's how I kind of initially went down that rabbit hole of AI automation. And you know, from there, I just started building stuff in my free time. I would literally sprint home from work just to get on and build stuff. And one thing that I've always kind of known about myself when I want to learn something new is I like to teach it. You know, if you can explain something from

### [2:30](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=150s) Learn-by-teaching: how the channel started

the beginning to end and someone can understand it and learn it, then that validates that you really do know what you're talking about. So the motivation was I'm just going to create some videos, throw them on YouTube just kind of for fun, also to teach myself, document my journey type of thing. And then because there was not a huge supply of NAND content on YouTube, but the demand was growing, the videos just started doing well. And then obviously that was the motivation from there was to just continue to make more and you know was getting nice comments and stuff. So that helped it. That helped a lot too. And it just kept it going. And then from there it just evolved and it's been a lot of fun. Cool, man. Yeah, I

### [3:07](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=187s) Why teaching boosts retention

feel the same way like with a lot of like channels is if you want to like learn something, right? You can watch a video on it, you get some retention, say 10%, you want to then do the thing, you get like 30%, then if you teach the thing, you get 90% retention. And that's so true with uh and what's really cool about that in making content is that it's kind of like this double-sided thing where you're learning something and then you get to share something. So, you get like extra value out of like learning the thing. you just have to go through the comfortabilities of making content which is a which is — can be a challenge in and of itself with making content side. I'm curious about like the ideation phase when you're

### [3:41](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=221s) Sourcing ideas: community, problems, data

coming up with ideas for making content and like you know how that's shifted from you just starting now to you really being tapped into the community of in like how do you source ideas come up with ideas figure out what do what you should make content on or what you should dive in to solve? That is a really good question because there's so much that goes into that and that's something that I had to kind of learn throughout my journey because when I first started it was just like anything that I could actually build because at the time you know if you go look at anything content from September when I started it is so much more technical you know you actually had to like build out this little JSON payload of schema you wanted to send over to different nodes and it's just it's crazy to me how much easier it's gotten but from there I was like anything I can get that works I'm just going to make content on. But the reason I say it's an

### [4:31](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=271s) Niche vs reach: conversions vs views

interesting question is because you kind of have to start with that end goal in mind. You know, if you are trying to build services for, you know, like a very niche specific audience, you can make that content with an optimized title for that audience, you'll get not very many views, but probably a really high conversion rate of people that are watching are genuinely interested in what you're building. So, are you optimizing for, you know, conversions on a specific niche or in my case at the time, I was just kind of optimizing to grow the channel? So, I was trying to think of templates that people might actually want to download. I was thinking of, you know, my best performing videos are the ones that are, you know, agents that have like a cool thumbnail and they're agents that have that that's kind of say they can do a lot of stuff compared to if you think about what is like a very high value ad that someone could actually put into their business right away and actually see like an ROI that kind of scales up as their business scales up. Those wouldn't perform as well on YouTube even though it's like a higher value video. So, it's a very interesting topic. But to get back to your original question, really the community is it's so helpful to see, you know, we offer tech support and stuff like that. So, whenever people are

### [5:46](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=346s) Community themes → video topics

posting about their issues, we always see common themes. So, a lot of times I'm able to make some videos that address those common themes or I take a look at use cases of what other people are building. And the thing about custom automations is there's an infinite amount. And also, if you ask someone, hey, can you build me like a customer support agent? there's an infinite amount of ways that they could build that for you. So, I'm never really hurting for ideas. I've got a to-do list on a Google doc, a very, very informal to-do list, and it's like video ideas, and it grows quicker than I can fulfill them. So, so far, don't really hurt for ideiation, but I'm also able to do things like scrape my comments, scrape YouTube for high performing videos in my niche, and have AI, of course, help me look at my positioning, and then generate me a list of new ideas. So it's

### [6:33](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=393s) Multi-agent research workflow for titles/thumbnails

always kind of staying fresh. — Love it. Yeah. I like the you had an AI agent if you want to call it swarm or multi- aent use cases where one of them was generating, you know, find me four high performing in videos and then send it to somebody and all that fun stuff. That was a great use case. I've actually used it at uh some of the meetups and uh and said, "Hey, this is something uh you know Nate Hurt made and all that. " And so I think it was really cool that use case because it's a specifically what you're talking about figuring out what's working well in terms of high performing videos or looking at your comments but like looking for these ideas of what do people want you know so whether you're a business and you're trying to figure out what services are provide or you're a content creator like what content should I make a lot of it comes back to like getting insights from what your community is asking for what other people are doing. I'd be curious about the automations. Like what automations do you have in place to help make your content creation life easier or better? — That's a good question and you may be disappointed by my answer to be honest. I don't have too many. It's a very

### [7:35](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=455s) What Nate automates vs keeps manual

manual process. I'd say it's more about the ideation where I have, you know, something that scrapes my comments so I can just kind of stay tapped into what people are asking about. I've got one that looks at things in my niche. Of course, analyzing like high performing titles and thumbnails and stuff like that. I don't take that into consideration as much given the nature of my content is very educational and it's less on the side of like curiosity driven. There's a little bit of an element there, but not the same. And yeah, and then the other one is just like staying tapped into AI news. So getting like a little email summary every morning of here are things that

### [8:10](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=490s) Staying on AI news & tailoring ideas

are going on in the space. And then I have an AI that's obviously kind of system prompted on me and my channel and it kind of says here are maybe some ideas that you could use this like you know a GPT5 job but here's how you could make it a little more specific to your audience rather than the hundreds of other YouTube videos about GPT5 just killed other models but you know that sort of thing. So — yeah it's interesting what you're saying

### [8:32](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=512s) RSS, trends, and niches inside n8n

that is just sourcing data. I know one of my inadin buddies, he hooked up like an RSS feed to NAND to try to get the download of what's going down and so you do you can do that or see what else is performing and because there's always new models coming out quickly, you know, Gro 17 or GPT32 or whatever we might be having coming out soon. And so it's cool to aggregate that data and get that all together. Uh then part of what we're talking about here is this content creation and figuring out niches, right? You said this is my niche. I've noticed that there's different types of like in it in content out there. Uh for example, you don't do as much voice agents as much as others do. Like what do you think are the different niches of innate in content or in interests that are out there? — Yeah, that's another that's a good question. So like you said, I don't I haven't made too much with voice. I haven't done too much either with like full systems as far as like front end

### [9:28](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=568s) Positioning & audiences (beginners → inspiration)

backend, bam, productize it. So I think in my mind, it's kind of the way I'm positioned right now is just kind of very general, but more so when I'm making a video, I'm thinking of could a beginner watch this and be inspired to download NN, try it out, and feel like they're feel comfortable, you know, actually building something. So that's kind of the way I approach it. I think there's obviously the people that come from a more technical background and their whole mindset on nodn is a little different. I think it's more of like a noden is insanely powerful. We've got the visual aspect, but me as a native coder, I'm going to use this to spin up a PC really quick and show the power of it. And then I'm going to turn that end workflow kind of into more of a programmatic agent. So then there's like the technical piece and then like you said the voice agents and then I think there's the piece where people are maybe less looking to spin up like an AI agency or become like an AI freelancer with workflow automation and they're more looking to create like a SAS and they want to be able to productize something wrap it up and just scale to the moon. So those would kind of be the areas I'd say that I don't touch on too

### [10:38](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=638s) From education to monetization paths

much yet. you know, everyone ultimately that's learning this kind of stuff is looking to either monetize it in some way or to become more valuable in their work. So, that's kind of the next natural evolution that I think I'm moving towards. I haven't touched too much on like the sell these templates or blah blah, but am starting to look at moving more towards the now that you have this knowledge of a automation, what are your options? you know, how can you move here from here to become a more valuable worker to pick up some side income to maybe even start your own business? So, that's kind of the foundation I've been building is just purely education for no coders. But, there's so many other ways to go with it. — Yeah. What I've seen too is in a couple

### [11:22](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=682s) The four learner buckets (+ SMBs DIY)

of categories. You're either a developer that wants to have a better toolbox because you want to be more valuable at a company or you want to get hired. you're an entrepreneur and you're trying to figure out how does this thing work in my business and I maybe am not that technical but I'm just trying to figure out what this looks like or you're some sort of agency or you want to be an agency. you're like, "Okay, how do I step into a business that I will enjoy and get paid well for with having an AI agency? " Or you're enterprise and you're afraid of the little guys running rampage all over your stuff and you want to figure out how do you stay relevant with with — Nad to really, you know, get more value out of the company by automating the low-level hanging tasks that take up a lot of time and don't really add a lot of value. And so I to me that sounds like the categories in terms of the types of people. Is there anybody else I miss in that those types? Um maybe the only other thing that comes to mind is very similar to what you said about enterprise, but just the small businesses that are knowing they need to get ahead. They don't want to pay an agency or a freelancer to do so. So they're learning to help grow their own business without making a hire. But I think ultimately, yeah, those kind of four main buckets is pretty holistic. — Yeah. And uh if there's another one, let us know. Comment down below if there's another type. I'm c I'm actually curious in terms of uh what we're talking about is you said to talk about being not that technical and then kind of coming at it from a code first perspective. Can you talk to me about this thought and belief around I'm not

### [12:48](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=768s) “I’m not technical” → mindset shift

that technical so I can't do inade or your own experiences with being how technical you need to be to get into inadin and what that looks like for beginners just getting started up and running with inade. — Yeah. So if you are someone that has no background with JSON code any of this kind of stuff and you want to get into NIN sort of like what the mindset should be is that kind of what you're asking — yeah well some people say okay I in order to learn in I now need to learn a bunch of coding languages or I you know this seems kind of daunting or like I don't know where to get started like there's a thing about I would say a misconception or maybe it could be a conception of I need to be x amount technical in order to use naden right is — any thoughts about that — I do have some thoughts about that I do I would say that it is a very common misconception I think that what happens you know it's kind of referred to as the entrepreneurial life cycle that I talk about a lot in my community and I know like people like Alexi talk about it but

### [14:00](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=840s) Uninformed optimist → informed pessimist loop

it doesn't apply just to entre entrepreneurs. It basically applies to whenever you try something new like a new initiative, a new venture. It's this life cycle where you start off here where you are an Hold on, I don't want to mess this up. You're an uninformed optimist because you see the opportunity and you know it is something that you know there's lots of upside to but you don't know enough about it to really understand how it works. So that's when you dive into it and you kind of are on this high and then you turn this corner where you become an informed pessimist because now you know the complexities that go into it. It's very overwhelming as anything is like even to me at this stage if I think about trying a new tool I don't even want to because I don't know where the buttons are. I don't know how the data flows but if I just spent you know like 10 minutes and really dove into it would all become very clear. So then you come down to the valley of despair where you can either become an informed optimist or you just drop off basically. And so this is where the value of despair that's where I think a lot of people need to be able to push through because I will say if you are coming from an absolutely zero technical background and you're opening up niten you're already going to be overwhelmed probably just because the interface is new and whenever something's new it's like whoa where do I click but because it doesn't look like it's unfamiliar but once you get past that you realize fundamentally it really just comes down to logic and I believe in this so strongly Like when I started doing automation at my job, I

### [15:30](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=930s) Map the process; one step at a time

was insanely overwhelmed. But I realized that all it takes is basically understanding step by step what you want to happen, you know, and it's basically if you can do something manually and you know the process really well, you can automate it because it's just one step at a time. But I think maybe where people get into issues is they go straight into a process that's very complicated and very agentic rather than kind of taking those baby steps. But also with the way that all of this, you know, like cloud oper 5 can help you generate JSON in order to, you know, create an end in workflow to start with. I think that that's going to help a ton

### [16:06](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=966s) Text-to-workflow helps—don’t rely on it for prod

because that eliminates a lot of the barrier to entry of like where do I start? But I don't think people should rely on that to actually be like a production ready system. I think it gives you a really good place to start. It gives you an idea of like a certain API you may want to use or how data should flow. But um yeah, I don't think that you need any sort of background. And I think that that's already been proven with over the past six months how many people are building some insane stuff nowadays. — Yeah. What you're talking about this is also there's two things we're talking about. So one is when you step into something new, there's this tend tendency to get overwhelmed and

### [16:43](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1003s) New → familiar; SOPs & whiteboarding

then bounce out, right? In the hero's journey framework, we call that the trial of tears. You jump in, you get smacked, you're like, "Nope, too much. " And you bounce out, — right? But if you can go inside there, especially if you have a mentor or a guide that can kind of lead you past the process, you can stick with it long enough to where it becomes, you know, this new becomes normal, right? And then when you go into this normal thing, you're like, "Okay, now I'm very comfortable with this. " Remember when I first saw in I was like, — looks cool, but like what is all this stuff? — Now it's just super familiar. I can go in, I know what needs to happen. I know what I need to automate. I know what nodes do what. So it all makes a lot of sense. And that second step of going into this clarity of thought is a lot of people whether you're in businesses or whatever they have an idea but it's not really refined idea in their head. They don't actually have a clear model. And so if you can map out those steps on digital whiteboard I use lucid use mural but like whatever things work is how can I see what am I doing? What do I really want to do? And SOPs are so awesome and if you can get those and start to map those things out then it seems to make those things possible. But you don't technically need to be technical. Ideally, you might have a technical friend, like a phone, a friend. You hit a technical button, they show up. But overall, it sounds like it's uh it's solvable because I'm sure you've seen Do you have any like stories or things of people either part of your communities that have come in with maybe limited technical knowledge and be able to seem some success in the back half? Do you have any stories about that come to mind? Yeah, I have so many cool moments in the

### [18:12](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1092s) Community win: rebuild from true understanding

community where people kind of hit that realization where they they're initially drawn to the community because they see a template that they want, they download it, they hit run, and they see some errors and then they're just kind of confused. And so they come for support and they try to figure out what's happening. And once we kind of like guide them in the right direction as well as like explaining maybe you know like this is a great place to start because you can see data move through but if it works don't just celebrate that it works actually look at every single node and click throughout the whole process to understand and so this one person I'm thinking of in particular she did that and I remember she posted on LinkedIn about like this is an awesome automation and then she also came back to me like a week later and said that she took that advice basically and looked at how the data moved through each step. And then what she did is she rebuilt the system without looking at it again. And I thought that was a huge win because that basically validated to her that she really understood how that exact system worked. And I think that kind of highlights this theory of like start with the end in mind. And if you have a very clear picture of what you want to build, the analogy I always use is, you know, if you dumped out a bag of Legos, but you put the instructions like under the rug, it would you could probably get there, but it would be really hard compared if you just looked at the instructions that said A connects to B and then B connects to C and C connects to D. It becomes very clear once you understand the steps and order that you

### [19:43](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1183s) LEGO analogy for workflows

need to follow. — Yeah. It's almost like learning how to read sheet music if you play songs, right? You don't really need to know the song, you just need to read the sheet music. So if you can map out these steps and you're like, "Okay, I get what these nodes do. notes do. I'm going to look at this map and then I can follow these maps and then left to right follow it out. " And I love that because you kind of say, "Okay, this is a working model because you make great workflows like the Aiden Swarm one. " I used that one at a meetup, right? I went and I said, "This is neat. Cool. " Download it, uploaded it, and I did it maybe I don't know, it was maybe 45 minutes before the event. I downloaded it, connected all got all hooked up, — and I was like, "All right, guys. Just want to check. Just want to let you know, full disclosure, this isn't mine. But the power of in it in is I could connect all these things. But because I've already hooked up Telegram before and Google and Ampify and these other ones, I already knew the process. I already knew how these things worked and functioned. And because I could do that, I could get up and running and showing off something really neat and really insightful. And so I think that's one of the powers of the of Nad is being able

### [20:39](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1239s) Template demo at a meetup

to have someone you trailblaze a path through this technical jungle. And then I was able to go back and go, "Oh, thank you for this map. " And I just download it, connect all the dots. And then it's nice because then you can go by step and I had to teach how some of the stuff works. So I had to go through these things line by line to show them off to the community say this is how this system works and this is what this looks like. And so I there is I've noticed this with masters uh because uh talk in shockingly has several init masters a part of their group um internally is there's always this thing of like methodical like you go through you make the map and then you methodically move through one node at a time and you see what's going on and you understand what's moving you know versus the shotgun it doesn't work okay I'm going to throw my computer away or you know rage quit or whatever. So I love it's seems so simple what you're

### [21:30](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1290s) Methodical debugging vs shotgun rage-quit

saying man but it's such a valuable key of like just look at them and like oh become familiar form a relationship with it you know oh this is my Google sheets I love you so much I've missed you thank you for holding all my data Mr. Telegram bot, could you please help answer my questions? You know, if you get familiar with these things, it is better. — Yeah. And just to jump off that real quick, too, it's like — when I really think back at all the automations and agents I've built, I think it really boils down to like maybe 15 core nodes. And if you can master those 15 core nodes, you can automate probably 97% of anything really. And similarly, when I think about the error messages

### [22:18](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1338s) 15 core nodes & 5–7 common error patterns

that I've received, it's probably a total of like a range of like seven error messages. You know, it's either kind of like an invalid API key or, you know, like permissions are wrong or the JSON body was not proper JSON. It's like a five to seven core error messages. And the problem there is I think a lot of people will tend to see that error message and just kind of get overwhelmed. But if you can actually understand the pattern of okay, when I got this error message, this is where I looked or when I did this got this error message, I had to look through the logs to see where the error happened and pinpoint the issue. Fix it, but then don't just like move on. Remember why that broke and think about this broke because I did because this happened. I fixed it by doing this and that's what addressed that error message. And now I get to this point where it's not ideal. People will send me just a screenshot of an error message. Ideally, they'd send more context, right? But they send me just a screenshot of the error message and I say, "I'm assuming this is what's wrong. Check there. " And they'll be like, "How in the world did you know that? " And I'm like, "Because I've seen this error message 500 times in the past week. " You know, it it's just pattern recognition. A lot of it comes down to that. — What do you do if you don't know? If you haven't solved this problem before, what's your go-to method for solving unknown problems? — Yeah, that's a good question because, you know, ultimately it will still happen at some point. And the first thing I try to do is before I go to the

### [23:44](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1424s) Solving unknowns: read, search, ask AI with context

internet, I want to see if I can do it myself because that's how you really break down and understand it better. And that just basically means I read the error message and follow what it tells me to do. A lot of times it will tell you exactly what's breaking. And you kind of just have to go there and find out. But let's say for some reason I spend a couple minutes and I just I'm stumped. I'll just copy that error message and just paste it into Google and then I'll just say end it in at the end of it and almost, you know, nine times out of ten that's just going to pull up like five links to the end forums or like to my community or someone else's community where someone's basically pasted that exact same error message and then the community helps them solve it. In that case, you just kind of follow those steps there. Like I said, that pretty much always does the trick. But if it doesn't, then I'm just going to paste it into a thought partner, an AI thought partner, whether that is GPT or Claude, and just kind of give it context. That's really what's important when you're dealing with these things. You can't just paste the error message right in. You have to say, you know, like, here's my flow. Here are the variables I'm sending over. Here's the error message I got. What could be the problem? Not just like, fix this. Maybe help me think of what could be the problem. And it will give you some options. And then a lot of times I'll also just like put a screenshot of my flow into there and it will help me think even better. So that's kind of what I do. And then I'd say last resort I would go to like a paid community or something like that where someone else maybe like an expert could help me look at it. have no it makes me want to build an automation that like I have okay this is where the typical answers are found and then have an AI agent that will then you know check the forums check this check there check Google and it like looks for what's the chance of this one being found and then serving that back. I wonder if there's something like that out there. — Yeah. Well, I I honestly don't use

### [25:29](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1529s) n8n assistant as a backup troubleshooter

that AI assistant too much, — but I think that's going to start to become my plan B after looking at it myself because of the way that that's like where it's pulling its knowledge from and because it's natively right in there and because with all of this, you know, text to workflow stuff getting better and better, I'm assuming it's going to get better and better at troubleshooting as well. And so if NDN is natively looking at the JSON you've got going on in that JSON payload. Not to get too technical but there's the error message in there and you and AI can understand like the flow. So I'm sure that's going to get really good too. — Love it. Yeah. And using that you can use the AI agent assistant inside of there and then be able to call different things like you know appy or other types of services to be able to get data sources from everything that you need. I could see that being helpful. But there is a an appreciation and sometimes a skipping of the manual process which I think you

### [26:22](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1582s) What to keep manual; where automation shines

seem to be really good at. Even though you know you're one of the who makes a lot of content on automations, it seems like a lot of processes are manual for getting certain things. What areas have you seen that are like valuable to keep as manual and what things are valuable to keep for automations? Do you have any like personal use cases or things that you see as common things in your communities or the companies you work with? — It's another good question. When I think about my personal day-to-day, there are a lot of things that I still do manual. And I'd say, you know, one of my biggest time commitments is the community. And obviously, I want all of that to be manual. I want the interaction from me my live calls support to be manual and all that kind of stuff. Now, if we zoom outside of that, I really like to have a lot of the creativity process at the moment be manual or at least with a big human in the loop step because um all of my posts on LinkedIn, I completely handwrite. I don't even use AI to give me an initial draft because I think sometimes when you have an initial draft, your brain just like subconsciously gets swayed to write the way that the AI gave you that first template. So with LinkedIn, I like to handw write all of it, but on the back end, I do like to have AI help me with generation of ideas because idea generation is like you have full control over what you want to pursue or what you don't want to pursue. I think it may just depend on who you are as a person or what your goals are with, you know, your content strategy. So, I do see some use cases though where I would have a system that I would trust to just pump stuff out automatically. I think it really depends there. But when it comes to emails, my email agent, everything's a draft no matter what. Like, you just never know what could happen. I end up always just kind of sending them off. I skim

### [28:18](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1698s) 90/10 rule: AI drafts, human finish

through them, I send them off, and I've never really had like a big issue with one, but still, those are always a draft. You never know what could happen. You know, maybe there's a bug and the agent fires off a hundred to the same person. It's just like well I don't want I don't I think the way I approach like manual versus automation is manual for anything that I just want like you know full control over no matter what or I need my 100% personal touch on it as of now and then when it comes automation very rarely is stuff like 100% AI automated for me automated yes you know data prep moving things from one sheet to another I would automate that 100% of the way. But if we're really bringing AI into play, I like to have drafts. human in the loop and I like to have a step of review. But you know, if you're automating 90% of your work, that's still awesome. Like that is still 100% valuable. So — yeah. It's like the 9010 rule, right? You want AI to do 90%, you go the last 10%. You take it the last mile. And you look at and say, "Okay, this is before I send this out either publicly or in an email or whatever. Right? Is this cool? We cool with this, right? And then it sounds like the other side is the 10% to get started. — So, how do I get started with it? Might be the ideation phase or coming up with options or become aware of things. But really that that section about, you know, getting started or closing it out are really helpful bits. And then the redundant tasks in the middle like moving, you know, contacts into a database or moving sheets over or any of these things that you need to go through and like, oh man, if I do this all day long, this is taking up time. Then automate those practices. I dig it. What have you seen in terms of

### [30:02](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1802s) AI thumbnail system that impressed Nate

like what is something that's like been like one of the most exciting or mind-blowing like automations or workflows that you've seen you're like, "Oh, that is cool. " Is there anything lately that you've seen as of recently that have been mind-blowing? — Oh wow, that's actually a really timely question. I saw a post this morning on LinkedIn from Nadia. I don't know if you've seen her YouTube channel or anything, but she's pretty involved in the end community, I'd say. — Mhm. and she entered a contest. she won the contest where it was like a YouTube thumbnail automation and it was really really cool the results you know I always kind of thought that's something I would never automate but after seeing the results it at least opens my mind again to like this could be something where I would at least trust it to give me like 10 drafts right away you know so she basically like had a system to get a bunch of high performing thumbnails analyze why they were doing well she also used like face swapping to make sure the right face was on all the thumbnails it was a really Cool. Cool post. And I think I haven't just because I've been I feel like I've been pretty tapped into what Anidin can do. There haven't been too many use cases where I've been like, "Wow, like that's insane. " But that one I was like very impressed by. Yeah, — that's cool because we do know like especially with content creation, thumbnails take a ton of time. Thumbnails, hooks, and the premise, right? You're just trying to okay, what am I going to say? How do I get started with this thing? Um, and I could see how that could save a ton of time on like and sometimes you just need to know see options to know what's available. And that's what I think is really cool about that because it may not generate it out the gate without you ever looking at it, but you're like you could look at one out of the 10 and be like, "Oo, I actually like that one. I didn't think about it that way. " And then start to iterate on it. — That's really really cool. In terms of like — the next where do you see this going in terms of AI automation? We'll say three months out because like I if I said three years it doesn't really quite — I wouldn't be able to give you an answer. — Flying cars and jets and mobiles or whatever we have. But like no literally

### [32:01](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=1921s) Next 3 months: natural-language builds & computer use

like three months out like where do you think we're evolving really quickly? I mean you've only been doing this for it's like a year. Has it been like a year? — Almost a year. Yeah. — Almost. You've almost been doing it for a year. Right. And uh had a marginal impact on your life I would say. Where do you think the next three months is going to go? What do you think we're evolving into? What can you what do you expect — the next three months? I think what I see happening is a lot more of the building stuff with natural language. I'd say I think that and a mix of computer use because when I think of what I'm limited by when I want to automate certain things, it is times where there's like authentication you can't get past with an API or they just don't have an API. So I think computer use is start going to start to get more and more used and also some more hype around it. But like I said, I think a lot of people are really craving that natural language to workflow building, natural language to agent building, whatever you want to call it, just because, you know, we've seen that kind of blow up with Claude Opus 4 and the N MCP server and now with GPT, I'm sure there's going to be videos about it. I'm sure every company right now, you know, just like we saw with the whole lovable bolt, the vibe coding era of natural language to web builders, I think every company that has any sort of workflow automation tool, you know, Google dropped Opal or Opal, however you want to say it. And I think they're all more than capable of being able to spin up templates with natural language. And I think all that does is lower the barrier to entry even more. And that's why I'm very adamant about I don't want there to be a false sense of security across the world of like I know how to build agents because I spit a sentence into there and it built it, you know, and but that's definitely where I see it going as long I just hope that people are still getting to their first iteration, understanding exactly what's going on, customizing a little bit, but most more importantly knowing how to improve it and add error handling and add guardrails. Because when I think about building workflows, whether I build it

### [34:08](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2048s) Production-ready = testing + guardrails

from scratch or whether I have an AI give me that first template, the last 20% is always the toughest because you basically need to test tons of different inputs and then you need to test all those inputs again with a different model and then you need to do it again with a different system prompt and do it again with, you know, all these changes. And that's actually how you get from a template that you see on YouTube, any of mine truly. None of them are production ready because they need to be tested on your data, your use case. So that's really going to be the more important skill rather than just building workflows is going to be making them actually ready. You know what I mean? — Yeah, that's a great point. If you don't know the mission of the company of inn, the purpose is to give technical teams the power of a 10x coder. And so if you look at this, the prompt workflows really do get you up and running and it kind of gets you that hit of dopamine where you're like, "Oh, this is cool. I can do this. " But really

### [35:01](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2101s) n8n’s mission: 10x coder power

anytime you get any incremental growth, like you learn some JSON or you learn how the nodes work, I think the definition of being technical is being redefined. And if you start to get familiarity with what Nadin can do and you can start to level those things up, you really can massively explode your 10x development beyond just connecting credentials with template workflows. If you start to actually level up those skills, it really does give you the power of a 10x developer, which I think is super cool. — Yeah, completely agree. It really just the way I always try to think of it is just you just got to be logical. And it sounds super common sense. It sounds super simple and obvious, but whenever you're really faced with these issues, I feel like if you just like kind of take a step back and you think about let's just forget that we're dealing with like an end or something on a computer screen. Just like logically think about if I was

### [35:55](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2155s) Think like a human when stuck

doing this manually and this is the issue I ran into, what would I go to change, you know, or where would I look next? And you'd actually be surprised how many people I've said that to when they're running into an issue and I've said like step back and just look at it like this. It becomes very clear. You can see in their eyes they're like oh you know like you know if my automation breaks because I'm having an error reaching like the Google Drive server. If you were a human and you were having an error reaching the Google Drive server, what would you do? So it becomes a lot clearer and I think sometimes people get tunnel vision a little bit because they're in this like end box. — Yeah. Sometimes helpful to step away and go okay what's really going on

### [36:37](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2197s) Map it; forums, friends, and community

here? And that's why maps are helpful. If you have a design map of any kind you can go okay this you know this node and this node are breaking. What is it? What are the different things I can do? And we just talked about like you know different ways you can deeply understand it. You can look at for communities that have it. You can look in the Nin forum and you, you know, you phone a friend, go to a paid community. These are all ways to overcome that. And then once you solve that, the next time usually I've also noticed this is the more painful it is to solve the problem, the less you're ever going to forget how to solve that problem, you know? — Yeah. — Uh I just like I was like, I will never forget this lesson again. and I'm going

### [37:12](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2232s) Painful lessons stick—share them

to put that in my pocket save that and I'm going to share it with as many people as possible which I think ties back to like the whole making content on YouTube is you're like I want the world to know how to avoid this pain — do this thing. — Oh yeah, — dude. I love it. Uh Nate, uh love you being on the podcast. This was super epic. Um is there anything else you'd like to let people know about before they tell them how to get a hold of you?

### [37:37](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2257s) Tool-agnostic skills transfer

— I would say now is the best time to be getting into it. And you know, obviously I'm a strong advocate for Nadn, but another thing that I feel like I address a lot in my community is what tool do I learn? You know, like am I taking a gamble? And ultimately I think that you because of like what we were just talking about with logic if you actually understand the process of automation and high ROI opportunities. I don't think it matters which tool you use because ultimately you know you're becoming tool agnostic if you understand those fundamentals and the foundation. So let's say right now you are learning make and you want to make the switch over or you haven't started with one yet and you're trying to figure out which one to do. Don't feel like you're gonna make the wrong choice because I think ultimately those skills will transfer over. The difference there is just kind of the UI and the terminology or where the buttons are, but the skills are all the same. And that those are the skills that are going to be in extremely high demand heading into this new year of 2026 in a few months. And yeah, that's kind of what I'll end off with there. — Yeah. Yeah, the future is bright and — these are like the true core skill sets of, you know, AI automation, thinking deeply, understanding these things, knowing how to work with the agents. It's all critical pieces to be able to solve real world problems real quick, which is incredible. Nate, if

### [38:57](https://www.youtube.com/watch?v=Hy47hfRPK2k&t=2337s) Where to find Nate (YouTube, community, LinkedIn)

people want to find you, how do they do that? — Yeah, you can find me on my YouTube channel. Just type in Nate Herk and I should pop up. And then from there you can find my free community or my plus group where we do you know support live calls monthly hackathons stuff like that. And then if you're on LinkedIn I'll be over there too. So rad. Honor and pleasure my friend. Much love and I will see you on the other side. Take care. — Awesome. Yep. Thanks for having me Dylan. All right. — Bye now.

---
*Источник: https://ekstraktznaniy.ru/video/15226*