Agents vs Workflows: Pick the Right Tool or Pay the Price [#02 Nick Saraev]

40:40

Agents vs Workflows: Pick the Right Tool or Pay the Price [#02 Nick Saraev]

n8n 11.10.2025 8 565 просмотров 233 лайков

Machine-readable: Markdown · JSON API · Site index

Смотреть на YouTube

Поделиться Telegram VK Бот

Транскрипт Скачать .md

Анализ с AI

Описание видео

Summary In this engaging conversation, Dylan and Nick explore Nick's remarkable journey from a door-to-door sales background to becoming a successful content creator and entrepreneur in the AI and automation space. They discuss the essential skills needed for success, the importance of authenticity in marketing, and the value of deep thinking and mental models. Nick shares insights on community building, time management, and the patterns of success observed in his community members learning AI tools. In this conversation, Nick and Dylan explore the importance of embracing uncertainty and a problem-solving mindset in the context of automation and AI. They discuss the significance of learning through challenges, the simplicity of effective automation, and the need to choose the right tools for specific applications. Nick shares practical tips for using N8n, emphasizing the value of quick workflow hacks and the importance of understanding risk and reward in AI applications. The conversation concludes with a reminder of the responsibility that comes with powerful tools and the exciting opportunities they present. Chapters 00:00 Tech moat is gone; business skills win 00:59 Where AI agents fit (high-volume, low AOV) 01:33 High-stakes risk: agents vs. workflows 02:02 How we met (leaked number → 100k) 03:05 Nick’s journey prompt 03:32 Door-to-door GMB sales 03:54 COVID crash → pivot online 04:46 Into AI & automation 05:25 GPT-2 freelancing → prompt engineering 05:47 GPT-3 + no-code → scalable biz 07:07 Skills stack for AI agencies 08:09 Top 3: Sales, demand gen, systems 10:07 First-principles thinking; Maker School ops 12:25 Demand gen: YouTube → capped community 13:41 Authenticity, scarcity, word-of-mouth 15:43 Value-first marketing; ruthless focus 17:56 Time & lifestyle audits (15-min alarm) 19:54 Remove friction: keys & the “Schelling place” 20:35 Who succeeds with n8n (future-self mindset) 22:29 Scoping rule: “Could I do this in 1–2 weeks?” 24:03 Docs, HTTP Request, and credential re-use 26:28 Guarantees + simple 3–10 node wins 28:31 Agents vs. workflows: decision rule 32:15 n8n 80/20: pin data 33:58 Testing patterns: limit / filter 35:27 AI latency: test smaller & faster 36:48 Learn fast without shiny-object FOMO 38:15 Best way to learn n8n (docs & quickstarts) 39:16 Power & responsibility (closing) Takeaways Nick's journey began with door-to-door sales, leading to digital success. COVID-19 prompted Nick to pivot to online opportunities. Sales skills and interpersonal skills are crucial for success. Technical skills are becoming less of a barrier in business. Authenticity is the most effective form of marketing. Time audits can improve productivity and accountability. Eliminating friction in daily life can lead to greater efficiency. Building a community requires authenticity and value delivery. Nick's community model focuses on quality over quantity. First principles thinking can lead to innovative solutions. The people that rise to the top are those who think about future possibilities. You don't need to know everything to solve a problem. Focus on what you can learn in a couple of weeks. Offering guarantees can drive better outcomes for clients. Simple automation can transform businesses more than complex solutions. AI agents are best for high volume, low risk applications. Understanding risk and reward is crucial in technology choices. Pinning data can save significant time in workflow testing. Incremental testing is key to efficient workflow development. Learning should focus on core tools rather than shiny new objects.

Оглавление (29 сегментов)

Tech moat is gone; business skills win

I believe the technical barriers to doing basically everything are plummeting. We see this not only with no code tools like NADEN and their AI builder and whatnot but just in general tech is no longer the mo hasn't been for I think the better part of like a year or so already and it certainly won't continue to be. So the most important thing in set of skills that you need to cultivate are business skills right things like sales skills things like interpersonal and soft skills. The things that actually transform my target market anyway, small business pieces for the most part, are like three to 10 node flows arranged in very simple linear fashions that do things like automatically generate emails when somebody does something, automatically notify a team member add a line to a Google sheet when somebody does something. Because they're so simple, you can absolutely do them even if you have very little technical experience to start with. AI agents are incredible piece of technology, but they work very well in specific applications. Those specific applications tend to be high volume, low average order value applications where you have a large volume of requests coming in where you need some sort of superficial decision-making with high reliability and accuracy but not necessarily 100%

Where AI agents fit (high-volume, low AOV)

and you're capable of making a few mistakes along the way. In doing so, because it's high volume, what you're doing is replacing headcount or payroll for an equivalent function and you're not really risking a lot of downside as a result. AI agents make a lot of sense to be used as a like chat architecture for that because if one out of a 100 times it chooses the wrong route or something, it's not the end of the world, right? It's 100 bucks. But if you are, let's say, some private equity firm and you are taking significantly higher average order value prospects and on the front end, your agent has a one in 100% chance of making some sort of mistake or something, then you're not risking $100

High-stakes risk: agents vs. workflows

anymore. Now you're risking maybe a million dollar. I remember reaching out to you in the very beginning when you were I think maybe a hundred subscribers and now you're at like over a 100,000 subscribers and uh seeing what you were doing was uh was pretty epic. I it's crazy to watch your growth trajectory and like what you've been able to make happen on YouTube. — Yeah, thank you. Uh you reached out to me at 100 subscribers. Really? — Yeah, you had only a couple of videos. You were very small. you and I you

How we met (leaked number → 100k)

accidentally left I think your uh phone number or something on one of the automations and I shot you a text message. I was like yo — that's what happened. Dude, that was wild. You totally did. I remember that. And I'm not going to lie, like some people have sent me, you know, IME messages uh after I leak my phone number in one of the builds that I do. And not all of them are as eloquent as yours for sure. But yours was done so artfully and you were just reaching out I think like asking some sidelong question mentioning that I leaked my phone number some something like that. And uh Yeah, man. Holy crap. Can't believe I forgot that. That's how we met. — Yeah. Exactly it. And it it's crazy to me how you can you see somebody online and then you see their YouTube videos and then you connect with them. And I mean since then I mean we've connected quite a bit. Um and I think it's been whatever like a year and a half now, — something like that. Yeah. I met you in person a couple times. Yeah. — Yeah. It's it's fun meeting people in person. You know, again, you have legs and feet and all these weird things that I don't normally see. — Do the jury still out. Apparently, I'm a robot according to the YouTube comment. — Yeah, man. Uh I'd love to like I'd love

Nick’s journey prompt

to hear just like one thing about it is um I'm super fascinated. You know, you've been able to have some accelerated growth in the content creation space and also you're one of the first people into the whole automation space. I mean, I think you're working with like the GP2 back in the day. I'd love to get just a little bit of your highle journey from starting out uh you know, with your copywriting agency and then the evolution into becoming a uh a content creator.

Door-to-door GMB sales

— Yeah, I mean thank you very much for giving me the platform to do that. Um I started doortodoor. So my background was not technical at all. I was selling local Google My Business marketing packages for 250 bucks a month to put my clients on the map. So, I was knocking on between 50 to 80 flower shops, you know, auto repair shops

COVID crash → pivot online

uh, I don't know, lawyer, uh, dental clinics and so on and so forth every single day. And I did that for, uh, quite a long time. And, uh, right around the time that COVID 19 hit, uh, when that occurred, where I was living in Vancouver, they mandated the shutdown of a bunch of businesses. Um, well, basically every local business. So, as a marketing agency that was based around local businesses, you can imagine that did not impact my uh the my numbers very well. We went from, I don't know, a little over 10K or something like that down to zero overnight. And as it was like a fledgling business that was paying all of my bills, basically completely and utterly destroyed me. So, I started asking, okay, like can I find a business model that's a little safer? And I didn't mean safer in the way that you might expect, but safer in so far that like, you know, I don't want my revenue to be tied to a specific location where that could happen to me again. And so I started looking at, you

Into AI & automation

know, online opportunities. And I started being like, okay, like what if instead of doing this whole door to door thing? What if I h, you know, find a way to accomplish the same result, but then I do so digitally. And so that led me down a big rabbit hole of cold outreach. Led me into like AI and automation tools. And it also led me into just like you know computer science and whatnot in general. So very thankful for that setback. At the time it felt terrible and obviously horrendous to see this little baby nest egg that I'd grown just turned into nothing overnight. But uh yeah, looking back it was extraordinarily valuable. And then from there I launched a small little I launched a couple of businesses. Um, but I started playing around with GPD2 as you were talking

GPT-2 freelancing → prompt engineering

about. Now, obviously we're up to GPT4. 5 or whatever. GPT03, you know, it's so high up that we're not even using the same naming convention. Um, but uh I started experimenting with GPT2 to uh pay my bills at the time with a little bit of freelance. I was doing some freelance writing and uh that was really interesting too because then I got to see basically like the first permutation of like prompt engineering. I mean that

GPT-3 + no-code → scalable biz

was basically a prompt engineering ground zero, right? And uh yeah, just as I was sort of figuring all this stuff out and you know, learning a little bit more about these tools and freelancing and whatnot, um all this came together uh and sort of like got me into my first content writing business where I leveraged the next model up which is GPT3 which had just come out and AI automation tools and no code platforms to you know make something that could grow sustainably without me having to uh you know knock on 50 to 80 doors a day. So that in a nutshell is like my journey, you know, very much sales. Um that's where I started with and then the AI and the automation side of things was sort of tacked on out of necessity and it was more so opportunistic. I was just looking for ways to guarantee my stability and safety. Um and that's where I landed. — Yeah, it's interesting that the core set of skills that you developed as a business owner in terms of like sales being the primary one out the gate and you know doortodoor cold sales is a is a very hard thing to do. Uh but you also evolve certain skill sets and then you jumped into you know like website design and copywriting and then codew writing um and then you know automation in general. It's interesting how they've stacked together right like being able to do sales being able to make something look good being able to use automations being able to design things and you put it together in a

Skills stack for AI agencies

bundled set. Are these like the cornerstone pieces that like an AI automation need company needs or like what do you think are like the the key pieces that are skill sets that you know um someone that's looking to build a company out of this need to have in order to succeed and flourish. — I believe the technical barriers to doing basically everything are plummeting. We see this not only with no code tools like NADEN and their AI builder and whatnot but just in general tech is no longer the mo hasn't been for I think the better part of like a year or so already and it certainly won't continue to be. So the most important thing and set of skills that you need to cultivate are business skills right things like sales skills things like interpersonal and soft skills. So you talk about stacking. Yeah, I think I stacked like the right way. You know, I started with like the core foundational thing that is the most important which is like interpersonal and soft skills that I slowly wrapped that in layers and layers of technical skills. Um but that core nugget remains the same and even if those technical skills end up changing or the need for them do because of new no code platforms like n

Top 3: Sales, demand gen, systems

uh you know like I still have that core nugget. So yeah, you know, if I could zoom out a little bit, I think some of the core ones are interpersonal sales skills. I think demand genen, so the ability to take an idea and then generate like leads from that. I think that's massive. And I think the importance of that is only going to go up as we quickly reach a place with no code platforms and AI where like you'll be able to have a business idea and realistically have your product within 24 hours. Like at that point, the only thing that constitutes whether or not that business will be successful is like can you acquire leads? Can you get demand, right? And then I think um I think also just like some fundamentals about systems. But when I say systems I don't mean like computer science systems. I just mean like how systems work in general like some mental models for thinking. You know some stuff like first principles for instance or modularity. You know these are things that I see a lot of people asking me questions about specifically when it comes to NADN. But when I try and explain it I don't as hey like this is how you refactor an NAND flow um or I don't know make things more modular. It's like, hey, you can actually take this approach to like everything in your life. So, those are probably my core three, I would say. — Yeah. And, you know, one of the other uh skills that I've seen you really hone and develop because you talked about mental models and I would say be able to develop your own a strong foundation of a mental landscape. And so in terms of, you know, going into a copywriting business and then I've seen some of your blogs, you have the the depth of your writing of going deep and like being able to analyze, okay, here's what's going on in my world. Here's my priorities. Here's the Eisenhower method for, you know, important, unimportant, urgent, not urgent situations. I mean, you what's your process? because I think a lot of it comes down from this foundational element of you being able to think deeply on a topic and then ruthlessly execute. So talk to me a little bit about the thinking deeply part and the mental models you use to get clarity before you take action.

First-principles thinking; Maker School ops

— Yeah, I think the biggest one is that first principles idea like everything that you see around you did not exist. It was thought up by people. Steve Jobs has that famous quote, people that are no smarter than you. I think some of these people were smarter than me, Steve. But I digress. Um, if you just approach everything not from like the standard conventional way of doing things, but then just ask like what do I actually need? You know, if all of this existed in a vacuum, I think you can go much farther than most people think. I am bringing up my community as a good example, maker school. We ended up doing uh an insane amount of revenue per staff member at the community. And the reason why is because that whole community is ran by me and it's just me. Um an equivalent community like the equivalent communities of my revenue size typically have teams of you know 5 10 15 20 people in order to maintain that level of revenue. Then obviously their margins suffer but I'm able to do it um with just myself and everybody from the outside in is like my goodness how the hell do you do this? This is insane. You know nobody else is doing it that way. And then uh you know as proud as I am about that. I then in my head I'm like well it really isn't all that much work because I've just constructed it in such a way where it doesn't really require it. I didn't look at how everybody else was doing things. Hiring you know oh as your community grows every 300 people you need to hire another person. I just ask well like what does a person actually do in the community? You know what sort of needs do I have on literally a per user basis? How much time on average does the average person that joins demand of me? and how high could I realistically scale that? Similar in nature to um you know Elon Musk and SpaceX, right? He had a bunch of people around him advise him and tell him that uh and I'm not just trying to draw parallel between me and Elon Musk. I mean you want to talk big galaxy brain like that's number one. But uh similar to how he talked about the rockets uh you know he has this big rocket example where like he asked a bunch of people how much money would it take to design this rocket and I think they were shouting figures at him like $50 million, $75 million. And he asked, well, what is the total weight of the aluminum in the rocket body and what is the cost of aluminum per gram? I don't think that needs to be that expensive. You understand what I mean by that? Like he just took a step back and he said, "hm, let me just start from first principles. " Um, I think that informs most things. — Yeah. I mean, it's interesting too because the way that you're looking

Demand gen: YouTube → capped community

at things, I saw you do a couple unique things that I think a couple other people might have started to parrot a little bit. Uh but when you talk about demand genen is that you know you built up the YouTube channel right and then when you went to go launch your school community right you got in some people inside of there you know get the energy moving and then what happened was I always noticed that you had a um one um as more people joined you raised the price which I didn't see that anywhere else and then I always noticed that you capped it at so many people to be in and you had a demand of people still trying to get into that makers uh make money with um school group uh and you were able to kind of not just open the gates and lower the quality, you were able to focus on, you know, having more demand than you had supply. Was that the logic behind the reasoning or can you talk me through that process? — Yeah, the way that I always see things now is that the most important thing you have is your authenticity and it's your reputation. And so I thought that there would be nothing better for me than to put this cap in place at least for my initial community make money with make way back in the day. Um because it would allow me to provide that onetoone personalized support to the degree that nobody else was capable

Authenticity, scarcity, word-of-mouth

of at the time. Um I had a lot of people being like well Nick there, you know, we there's I mean I had like five uh 400 people in the community at one point and then I had like 1,100 people that wanted to get into the community. And so I had a bunch of people send me like, "Dude, you got to take off this cap. You got to increase the limit. You got to do this. that. " And I the way that I always see things is like when you have a big group of people that want to buy your product, like a big waiting list, you don't actually need to act on that like right now to secure that revenue. If you function from a place of authenticity and renown and positive reputation because these people like you for you, they don't even really like you for the specific product you're selling or whatever. these are this is basically potential energy that you could later convert to some sort of product sale or something later on down the line. Like I you know there's not there's nothing I'm not holding a gun in my head being like oh my god I need to convert these people or they're never going to want to work with me, right? So I kind of thought about that and then um that strict cap um enabled me to like build a ton of value and goodwill within the community. Um sort of like the Costco model where it's like you don't advertise, you just talk to the people inside of the group, make them love you so much and they naturally advertise. Then everybody and their mom obviously wanted to get in. So when it came time uh for me to build another product which solved like a related demand but a you know a different need um in this case it was zero to one getting your very first customer with some sort of AI and automation solution you know something like um NAN for instance uh I just had massive interest massive surge I was basically able to like take all the Tinder that I had built up over the course of the last year and then just like grab a match to it if that makes sense. And yeah, I think that was one of the reasons why everything just grew so explosively, you know, really just not focusing on the advertisement or the marketing side of things, but just like making sure that I uh I treat the people within the community as high quality as possible and I really just give 100% of myself every day. Another thing that I thought through first principles, I would say, — yeah. Well, I mean, I think sometimes marketing gets a bad rap when you say marketing because I think there's like uh low-level marketing and then there's high value ad marketing

Value-first marketing; ruthless focus

if you want to call it that, where all you're doing is just constantly delivering value to the community versus like, hey, click this link, buy my thing, do my stuff. It's more like, hey, I'm going to keep giving you so much stuff and I'm going to keep helping you so much. I mean, that's part of the reason why, you know, I think I called you the Alex Shrimosi of automation back in the day is because I saw you constantly gave value and you gave so much value consistently. There was kind of like that meme Fry from Futurama that just shut up and take my money, you know, just waving my hands across. But that but I think it comes down to this is this I think of this trifecta whether you're working in a business or company or working with people. It's like a trifecta of you want to work on something meaningful, right? uh with people you like and then you want to get paid well for it. And so, you know, with you making a lot of great content, adding a lot of value things out there like, "Oh, wow. This is a new era of AI. There's new opportunities. There's new things. This is meaningful. Um you know, I like Nick. He makes these side jokes off to the side while he's creating content. I feel like I'm connected to him. " And then, you know, there's this, you know, there's this opportunity that I think I can get paid well for because there's a lot of opportunity in this marketplace because this is kind of currently an eclectic skill uh that is in high demand. — Yeah, absolutely. I think you hit the nail on the head there when you said that uh that was still marketing and I should have rephrased like that is marketing. Like being extraordinarily authentic in an age where authenticity is at a premium is marketing. That is about as marketing as you can get. And I would go as far as say is the most effective form of marketing. Download this thing and uh reply this and I will send you this thing. I mean, you know, these are tactics. These are gimmicks. They're not like at the core of what makes somebody actually grow um a personal brand. So, well said. — Yeah. No, just something I've just noticed and it's and it serves you well because I've seen your growth um in the time we've known each other and it's been epic to see. welld deserved too because I see that you're very um I would say ruthless with your time or ruthless with things that are distractions in your life in terms of your ability to cut things out and be willing to say, "Okay, what is the most important thing and what am I uh wasting time on? " You did a time audit back in the day where you were like, "Oh, I don't know where my keys are. I

Time & lifestyle audits (15-min alarm)

need to get this. I don't know where this is. " And so, like, do you always do these time audits or what is that like? Talk me through that a little bit. Yeah, just for context in case somebody doesn't know, the uh time audit is essentially where well there's two. There's a lifestyle audit and then there's a time audit. The time audit is where you just have this little alarm that goes off every 15 minutes and then it asks you to describe what you just did in the last 15 minutes. And so I consider it a form of like radical accountability. You have this alarm that goes off in 15 minutes. Not only are you forced to be like very objective with yourself the next day when you look back and you're like, "What the hell did I actually do? " Um, I also find that like what gets tracked gets improved and so simple knowledge of the fact that you have this thing that's going to go off in 15 minutes makes you want to justify what you are doing right now preemptively. It's kind of a hack in that way. So that's the time audit and then the lifestyle audit you talking about like where the hell are my keys? There are so many tiny little points of friction in an average person's life that they just chalk up to being how life works. that example of keys. I distinctly remember there were like a couple days when I was hanging out with friends or something or you know meeting somebody for a meeting or going to the dentist and uh I just didn't know where my keys were. So I was running around my where the hell did I put the keys again? Were they over here? And you know took me like 3 minutes to find my keys and then I was off on my day. That happens like realistically to most people like once every two weeks, right? If you count up the three minutes and all the stress that um you know generates for you like over the course of a year that could be like a good hour and a half of you just being hyper stressed all the time. That's a very small example of friction but friction like that is present all over our lives. So, what this lifestyle audit is it's just where you consciously and intentionally using the time audit and using um you know self self-reflections and journaling to go through all of the points in your life that provide you a tiny bit of friction and then just seek to eliminate those. Good example of the keys are you get what's called you build what's called a shelling place. And the

Remove friction: keys & the “Schelling place”

idea of a shelling place is just the same place that you put all of the stuff that you need to leave the house every day. It's where you put your wallet, keys, where you put your I don't know your glasses. and it's just a little I don't know like a stand by your door or something, right? If you just know to put your keys there, you've effectively solved all future instances of that key problem for the rest of your life. Instant ROI. Talk to me about this cuz one thing I've cuz I mean you know you run a couple of um communities and you've seen a lot of people come in and some people come in um they don't really do anything. They kind of fizzle out because they just don't put in the time or energy. Other people rise and they excel and they do very well, right? in terms of like learning NAD in terms of just

Who succeeds with n8n (future-self mindset)

overall increasing your AI and automation skills, what have you seen as like patterns of behaviors or these types of mini hacks amongst your community that people have done to really be able to go from I don't know anything about NAN to actually being able to use it, utilize it, and have it serve them whether they're getting a job or be developing an AI agency or anything like that. Man, what a brilliant question because the most important thing to me is all revenue and savings for the clients that I work with. I'm obviously a big proponent of this AI agency model where you use AI and you use no code platforms that are high leverage to deliver things that clients care about. Money in their pocket, savings in their pocket, whatever. But when people come into the community, specifically people with an NAD background, NAD betrays a certain level of technical sophistication. You have things like inline JavaScript in the expression editor. And so because you don't have to do that. You know, they also make it extraordinarily simple for people that don't, but because the skill the potential skill ceiling is higher, you have a lot of people that just like think that they need everything to be perfect and they need to become NAND pros in order to go there and actually deliver the sorts of revenue or savings based outcomes. So just to delineate between the two groups of people, the people that like bubble up to the top and the people that like go down to the bottom. It's usually the people that like get really lost in the technical weeds that come in and then bubble down to the bottom because they are obsessed with certainty and they're obsessed with feeling like they have to know everything like it's a freaking multiple choice test. The people that rise to the top are the people that when they talk to customers or clients, what they're thinking is not, can I do this right now? What they're thinking is, is there a future version of Dylan that can do this at some point? And if the answer to that question is yes, then they go ahead and they continue putting themselves out there and interacting with customers and whatnot. Like I'll give you a quick example. Um, you know, if there's some client that you're discussing uh, you know, potential a automation thing with and they're running you through their needs and

Scoping rule: “Could I do this in 1–2 weeks?”

they're saying, "Oh, right now we have this problem with onboarding and we're using this one app and it connects with this other app and this other app. Can you help? The question that you need to answer is not can you help right now. You might not know what the hell any of these platforms are, what the hell they do. You might have never worked with a proposal platform in your whole life. The question that you should really be answering is could I do this in like a week or two? And guess what? With no code platforms like NAN and others, with AI platforms that you could chat with to solve problems or use in part of your flows, in reality, I don't believe that there is a business problem that you can't solve within a couple of weeks of just focus head down effort. So don't try and solve for everything today. Don't try and become a technical pro today. Obviously, it depends on what you're doing with your life and the purpose of you learning these tools. But in the case of driving revenue and savings outcomes for businesses, you need to know a lot less than you probably think today in order to move forward with a client opportunity that can manifest in money tomorrow. — Yeah, I love that. And it sounds like it's this willingness to understand that you're you can figure this out. And a lot of times unless you put yourself in that situation, you won't figure it out. And I realize that yeah, there's nothing more incentivizing uh to figure something out than someone hiring you and paying you money to do a job and then you need to deliver on that job and and need to sort that out. Do can you talk to me was there any examples? I mean, if you had anything with inn where there was a situation that maybe you got put into a a situation you're like, "Oh man, I have no idea what I'm doing right now, but it's okay and I'm going to sort it. " Do you have any practical examples

Docs, HTTP Request, and credential re-use

on your side that you've experienced that with? — Oh man, so many. Oh man, I still get them all the time to be honest. Um, but for me again, it's always a question of like I might not know the answer to this right now, but I know how to find the information. like wrangle through API docs. I know how to go to like Naden's documentation which I'd consider pretty comprehensive and just like follow along top to bottom and figure it out. Uh I've had so many instances with like I don't know some built-in not pre-existing in NAN and me being like oh no this thing doesn't exist. Oh no I guess I can't do the automation. Oh and then I'm like wait a second wait we still have HTTP requests. I guess it'll just make an API call. And you know, it's like a simple copy the curl, paste it in, and you got 90% of a built-in anyway, right? Predefined credential types were really big for me when I realized that you could just like move over a credential that you set up for a platform, and then you could do a thing that like was not built in supported by anyone or whatever, just using an API request while still reusing that credential. Um, yeah, I've had a lot of clients ask for things that are just like completely infeasible as well, uh, that, you know, just nobody in their right mind would ever actually say yes to. and then I have for whatever reason said yes to them and I've just been in that situation where like oh boy you know this is going to be an allnighter but with sufficient API docs with sufficient you know like confidence in yourself you can get basically anything done we live in like we live in an incredible era let's just be real we live at like the dawn of like computers can talk to us now uh you can figure out anything given a couple weeks for real and some elbow grease at least that's the way that I see it though. — Yeah, 100%. And that's you never find your edge unless you go too far. And that's sometimes and when you take on a couple of clients, you realize what are the what projects are grandiose, right? When are you over scoping something? And you know, what does it look like to say, okay, this is this through experience, understandable, is something I know I can sort. And unless you're willing to put yourself out there and take those risks, then you never have that chance for growth. But if you want massive growth, it sounds like what you need to do is be willing to mess up enough times and, you know, probably be willing to refund some people money and do it in an ethical and honorable way. And just realize that a lot of these things are learning lessons for the growth that you get from the lessons that you take from going through these hard challenges. Absolutely. I'm a big fan of offering guarantees when you're at the start line of any business model. Some people disagree with me and of course we're all entitled to our own opinions. But the

Guarantees + simple 3–10 node wins

way that I see a guarantee is if there are two universes and in universe A you do not offer somebody a full money back satisfaction or whatever guarantee for the service and there's another universe where you do that other universe you put your ass in the line. So you're probably going to work harder and be more likely to actually deliver what the client needs. So the way that I always like to to see it is you dive in with both feet. You guarantee your service. You be willing to put your asset on the line. Uh to make a long story short, um you know, if you're willing to do this and put your butt on the line, then both cases are great for the client because now they have some solution that is like halfway done or high quality or maybe low quality or whatever, but they've learned a lot more about product discovery and then they haven't had to pay anything and then they got like significantly further ahead with their needs. you obviously learned a bunch of things on your end and you're much better able to position yourself to deliver their needs or you knock it out of the park and then you get incentivized for it and you get paid for it and uh yeah you can just take those skills on uh to either that client or the next client or whatever and again just because these are very straightforward asks for the most part you know the things that transform businesses are not the five quadrillion node flows with like multi-step AI agents and all the super complicated data wrangling and whatnot the things that actually transform my target market anyway small to mid-size businesses for the most part are like, you know, three to 10 node flows arranged in very simple linear fashions that do things like automatically generate emails when somebody does something, automatically notify a team member add a line to a Google sheet when somebody does something like because they're so simple, you can absolutely do them even if you have very little technical experience to start with. Yeah, that's interesting because there is a thing that happens with you know there's amazing complicated AI agent flows online that are very I mean they're fascinating to see and watch um through inadin but then a lot of these practical business use of cases are simple workflow patterns something that can allow you to reliably and predictably be able to solve a problem that is relieving uh much like we talked about finding your keys the three minutes of something is where are those

Agents vs. workflows: decision rule

little hidden things inside of there. So when you know what advice would you give to like a starting up AI agency when they're trying to find out you know should I build an AI agent? Should I do a workflow? I don't know exactly what to do with this person. What does this look like? How would you recommend them under uh determining you know what's the best path for them to build out workflows and what that looks like? — So basically design patterns, right? Like what sort of architectural choice should I make for this build? I think that AI agents are incredible piece of technology, but they work very well in specific applications. Those specific applications tend to be high volume, low average order value applications where you have a large volume of requests coming in where you need some sort of superficial decision-m with high reliability and accuracy, but not necessarily 100%. And you're capable of making a few mistakes along the way. In doing so, because it's high volume, what you're doing is replacing you're replacing head uh headcount or payroll for an equivalent uh function and you're not really risking a lot of downside as a result. So to make it a lot more concrete, if you have this AI agent that's doing like let's say some lead collection at the front end and you are a business that charges I don't know a h 100red bucks a pop or something like that for your services, AI agents make a lot of sense to be to you to be used as a like chat architecture for that because if you if one out of a hundred times it chooses the wrong route or something, it's not the end of the world, right? It's a 100 bucks. you get a thousand clients at 100 bucks a month, you're fine. But if you are, let's say, some private equity firm and you are taking significantly higher average order value uh prospects and on the front end your agent has a one in 100% chance of making some sort of mistake or something, then you're not risking $100 anymore. Now you're risking maybe a million dollars. And so the downside risk of that is much higher. And then the upside that you get from replacing a human staff member is concordantly lower. So that's the main driving decision for me when I think about whether I use a agent um architecture or I do something that's a lot more linear procedural or traditional right left to right. — Yeah. It's a riskreward ratio and it's kind of like is there massive volume with low risk in terms of each per unit or is it something that's massive risk and low volume that's something that you need to make sure that you nail out and is reliable and repeatable. And that's one of the challenges I've seen with these AI agents is that, you know, they're super cutting edge. They're great. They're amazing, but they're also a little experimental that we're going into, right? Versus that workflows are very predictable, reliable, consistent, and repeatable, uh, but they're not as dynamic. So, I think it's okay. So, if you're looking at the spectrum going, okay, where does this fall into? You're looking to figure out, okay, which bucket does this fall into? And this will determine what my design patterns will be. — Agreed. I think that like anything when you use technology that's on the bleeding edge like the AI agent um architecture. I think that you stand to gain a lot. It's just about nuanced application. If you apply it to a high volume, low average order value industry, you can print you can change the course of a business forever with something like that right now. But using it as it's like when everything is a hammer, all you see is nails. I believe that's the idiom. I may have butchered it, but hopefully you understand what I'm saying here. Like, if you just latch on to the new shiny thing and try and use that to solve every solution, then obviously um that's not going to be as effective. If you understand the nuance of what I'm talking about with the average order uh value and volume, then you're able to apply these things much more um accurately, I would say. H what do you

n8n 80/20: pin data

think um in terms of like with NAN there's a lot of different um I mean nodes in there there's triggers there's different um AI agents inside of there what do you think is the 8020 inside of NAN that people are using and what are the common things like copy curl for HTTP request one of my favorite things of all time um what are some of the other ones that you've seen that are whether they're little known and but you commonly use them or ones that do you think is the 8020 of inade? — Yeah, like quick workflow hacks and stuff. Oh man, one of my favorite topics. I'm actually making a video on this pretty soon. So, we'll hit the print we we'll hit the prints and press with this one first. Um, but okay. So, obviously pinning data, right? Like obviously pinning data. Hopefully, we're all on the same page about pinning data. One of the major issues that I see beginners fall prey to when they start building in N8 or really any other workflow tool is uh every time that they want to test the automation, they'll rerun the whole thing from start to finish. The thing is if you rerun the whole automation from start to finish, a lot of the time these triggers are things like a form fill, right? Some sort of like external action you need to do. Well, think about it. How many times are you going to fill out that form over the freaking lifetime of the build with the testing? You're going to fill this thing out like 50 times. And guess what? It's going to take like 2 minutes every time. So, what have you done? You just added yourself an additional hour and a half to your build time. What if you just pinned all of your data and then you get to rerun off of like pinned to data and not that? You legitimately save hours. That's number one. Oh, and so many API calls and tokens and resources and you bashing your head against the wall being like I've had to all tab to freaking some form solution 5 million times. It's the worst. Um, so that's number one. So pinning data. Number two, um, again in the same vein of testing. I'm a big testing guy, believe it or not. You know, a lot of people are like Nick's just like a move fast and break

Testing patterns: limit / filter

things kind of guy. I'd agree with that to some extent, but like man, when I build my flows, I'm very much incrementally testing them the whole way through. Uh like limit nodes and filter nodes. Um I don't think a lot of people know, but you can just like deactivate them and then the data flows right through. So if you set the source data before the limit or filter node for a node that is after the limiter filter node and you just activate or deactivate the limiter filter node, you now have a little testing design pattern where you can test on single units of data. You don't actually have to run let's say 50 items at a time. You can limit maybe output two so that you can worry about you know um connecting item matching if you have more than one item in an array. And then you can just run it on those two over and over again with the pin data as well. So now you're like you're stacking these wins, man. Testing leverage. Am I right? — Mhm. Yeah. So, so you can, for example, if you had a Google sheet and there's a whole bunch of data inside the Google sheet and you're searching for those rows instead of running those every time and having, you know, a thousand of those run through, you can put a limit on there and just kick through one at a time. And so you can use that as testing and then when you're ready to actually turn on the automation, you can just toggle that on or uh deactivate the limit node and then the entire data will flow through and your system works as expected because it's been tested. — Absolutely. And then you saved yourself uh big pain in the butt having to deal with like test waiting for the thousand items to run and so on and so forth. So that's really big. I mean it's really big specifically because a lot of the flows nowadays

AI latency: test smaller & faster

involve AI to some extent, right? like you are dropping in some call to like some uh AI API large language model API and the thing that is different about them versus like you know procedural uh systems is the amount of time it takes to finish an API request really varies like before you you're sending something to some procedural endpoint like it's going to return the thing back within like. 5 seconds and it's like yeah sometimes it'll be a little bit slower it'll be 6 seconds faster be 4 seconds well you know I think we've all probably sensitive data to some large language at this point like sometimes it takes like 5 seconds, 30 seconds. So like I try and minimize that incremental cost every single time by pinning data and then by limiting the amount of data that comes through so I don't have to wait like 20 minutes. It's also just so frustrating. Super frustrating. Like I don't want to just sit there waiting for a variable amount of time to get my thing done. Uh and that's why I'm really thankful about that. — Is it funny how like impatient we're now getting with if an if the node takes more than 5 seconds, you're like come on I got you just robbed me of 5 seconds. I want back. Give me that thing back right now. And but part of the thing is like all these little micro hacks would allow people to be able to develop so much quicker. Um I'm wondering here what advice and guidance would you give someone someone's you know stumbles across your channel or whatever and they're like oh I need to learn n I need to get up and I need to do it now. Right. What advice and guidance would you give to them uh

Learn fast without shiny-object FOMO

to be able to expedite their journey? — I would hedge against that a little bit. I understand that we all have this like super crazy desire to learn everything today and right now and we got to keep up with all the amazing tools and stuff like that. But the reality is like have you seen the current state of like AI and automation Instagram or Tik Tok or whatever. There's like every 5 seconds it's like new tool comes out that eliminates everything. New tool changes the world. Other tool does this. Like you could just you could really back yourself into a freaking corner. And I see people in my community all the time thinking like I need to learn one new tool a day and if I don't I will be behind. Well, earlier you mentioned that there were two groups of people that join these sorts of programs and that you know work on building a business, right? The people that bubble to the top, the people that like end up um going down to the bottom. More often than not, the people that bubble up to the top, they just pick a stack of some, you know, some tool stack and they just master that. They're really good at it. So, from a leverage perspective, what is a good tool to actually invest your time in? It's a no code tool like NAN because if you think about it, you're going to use that for everything, right? Where what is not a good tool to invest all of your time in some tool that does some tiny little fraction of uh what a business needs. So focus I guess what I'm trying to say is like you should direct your learning less on these shiny objects and more on like core proven infrastructure. I think NAND definitely qualifies as a tool that you I would be happy if you spent your time learning. So, just

Best way to learn n8n (docs & quickstarts)

to, you know, just to make sure we're all on the same page about that shiny object stuff. Um, but yeah, in terms of like fastest way to pick up Nad, they have fantastic documentation. They have fantastic um, like I literally just went top to bottom. I read through the entirety of the Nad docs. Took me a very long time to be clear, but like by the end of it, I knew the platform, you know, like am I going to sit here on my iron throne and be like, well, I have a bunch of NAND tutorials that you should watch and they're the best. You know, I think that they're fantastic, but I also acknowledge it like there's just such a tremendous amount of time and energy put into making super accessible resources for everybody from the NN team that you would not go wrong literally just going top to bottom uh one of their quick start guides and uh you'd probably learn the 8020 right there. So, a lot of different ways, but don't ever feel overwhelmed, — of course. And uh you know with all that being said um is there in wrapping up the podcast now is there anything else you'd like to let people know about before you tell them how they can find you? — Um I would say we live in extraordinarily exciting times and

Power & responsibility (closing)

learning this set of skills is probably just in terms of like raw impact and ability to change the world. Learning this set of skills, AI combined with a no code tool is akin to like wielding the throne a few hundred years ago. It is so gosh darn powerful. The things you can do it with it are so gosh dang amazing that you should also keep in mind that you carry a fair bit of responsibility as well with that power. Um, so understand that like we're in a we're in such a lovely position to be able to do these things. These tools are extraordinarily powerful and they can change the lives of a lot of people. You know, treat them seriously. Um, at the, you know, as much as I love treating a lot of the stuff like a game, I would also say like you have responsibility to use them for good and whatnot. — 100%. Yeah. The uh Spider-Man quote, great power, great responsibility. Honor. — That's me, Uncle Ben. That's me. — Nick, it's been awesome, man. Thank you so much for coming to the show. Uh, much love, my friend, and I'll be talking to you soon. — Looking forward. Cheers. — Take care. Bye.

Другие видео автора — n8n

Ctrl+V

Экстракт Знаний в Telegram

Экстракты и дистилляты из лучших YouTube-каналов — сразу после публикации.

Подписаться

Лучшие методички за неделю — каждый понедельник