Seriously, Please Watch This Before You Declare n8n Dead

1. 0:00 – 2:28

   Intro

   1. JOJan Oberhauser0:00

      We have grown 10X in the past year. The week OpenAI launched Agent Kits was one of our fastest growth weeks ever. Even the frontier labs are using us. Many of our users using Claude Code internally, but still use n8n for business-critical use cases.

   2. AGAakash Gupta0:10

      Meet Jan Oberhauser, the CEO and founder of n8n, which was just valued at $5.2 billion and crossed 100 million ARR.

   3. JOJan Oberhauser0:19

      If you want to build something that's reliable, auditable, and you can pass to your teammates, you really need n8n.

   4. AGAakash Gupta0:24

      This tweet went viral from John Ennis. "Remember n8n? Went irrelevant pretty quickly, huh?" Do you need n8n anymore?

   5. JOJan Oberhauser0:32

      I think there's a lot to unpack there. I think we have been called dead probably a thousand times. I don't remember half of the things that, that killed us over the years.

   6. AGAakash Gupta0:39

      You guys actually existed before ChatGPT, and you talked about in a podcast how you said you were scared a little bit. n8n was literally the hottest AI tool in the world in 2025. Now people are declaring you all dead. Before we get into today's show, please take a second to check that you're subscribed on YouTube and following on Apple and Spotify podcasts. If you want access to all of my favorite AI tools, I've gotten them to give you an entire year of their paid plans. Check out bundle.aakashg.com for an entire year of Bolt.new, Airtable, Speechify, Descript, Magic Patterns, Linear, Dovetail, Arise, and Mobbin. And now, into today's show. n8n was the hottest tool in my newsletter last year. I wrote about it over 25 times. You can use it to learn the basics of AI, like RAG and fine-tuning. You can use it to automate your workflows. You can use it to create AI agents using its visual workflow builder. And then Claude Cowork came out, and then there was the rise of Claude Code. Today, we have on Jan Oberhauser, the CEO and founder of n8n. We're gonna get to the bottom of what can n8n do that Claude Code and Cowork can't? What is the role for n8n this year, in 2026 and 2027? How should you think about using n8n versus these other tools? We'll also talk about how they build product at n8n, what their new latest metrics are, and much more. Jan, thanks for being on the podcast.

   7. JOJan Oberhauser2:26

      Thanks for having me. Excited to be here.

2. 2:28 – 7:09

   Do you still need n8n in the age of Claude Code

   1. AGAakash Gupta2:28

      So n8n was literally the hottest AI tool in the world in 2025. Now people are declaring you all dead. I looked at the Google Search trends, and there might be something behind it if you look at these trends here. Then this tweet went viral from John Ennis. "Remember n8n? Went irrelevant pretty quickly, huh?" Do you need n8n anymore?

   2. JOJan Oberhauser2:51

      I think there's a lot to unpack there. I think first, uh, probably important to call out that we-- I think we have been called dead probably a thousand times. I don't remember half of the things that, that killed us over the years. We're still here, so, um, I think that, uh, probably doesn't hold that too. Also quite exciting, uh, quite interesting, I think if you check out half of the other tools out there and you type into Google Trends, I think you see very similar trajectories. I think it's just there's obviously a hype, hype cycle where it gets a bit overhyped, and then it kind of goes into this kind of more normal, um, framing again. I think that's also where we are with, like, the most products there. But yeah, I think may first things to talk a little bit more about Claude Code, uh, and other ones. I think it's an amazing tool and really deserves, um, to be out there. But I think the most important thing is, like, n8n and Claude Code are very, very different products. It's like cl- you, you need both in the end. Like, Claude Code is more like agentic tool, like it runs Anthropic models in your terminal. And n8n, you can see more as this kind of orchestration layer to really connect your kind of tools, your, your kind of LLMs, your data sources, and it kind of offers you this kind of visual canvas for, for systems to really run reliably and securely. This is especially important for business-critical use cases, uh, where technical and non-technical people really collaborate. I think you, um-- I certainly think it's also kind more important than ever for, for this-- Like n8n is more important than ever for this, especially business-critical use cases where reliability, security, and auditability really matters, where, yeah, we want to be 100% sure that you know what is running. Like, you c- you can inspect what's running, you can see what did run, um, where you cannot just, uh, rely on it working 95% of the times, but it c- it has to be 100% sure that it really works. And I think that's again where our visual canvas kind of really shines because it sees, shows exactly what, how it works, what it's done, um, and just kind of, um, or also when you work with other people where you kind of talk about, like, what is actually running. Like, if you have something like a Claude Code, it generates literally 10,000 lines of code that nobody can inspect, and nobody knows if, if, if it's actually doing the right thing. Also, do we see actually a lot of people that use it together, like they very often prototype with Claude Code because again, you can get something started very, very fast. But then actually kind of, um, migrate it afterwards to something like n8n, uh, for the reasons, uh, that I've already mentioned again, especially for, for the auditability, when you want to handle like large or complex data processing, uh, pipelines, they want to be, like, able to kind of, um, more precisely, uh, define what actually should happen. And where also self-host ability really matters because, like, people care more and more about data privacy and security. And also, one important thing to be aware of, actually, multiple of the model companies are using n8n as well. And I think that also makes sense because in the end, you want to use the right tool for the use case. And actually using us for things like security or compliance use cases because that's again where we really shine. And also, if you look like in, in our stats online, you still see like crazy growth. Like we crossed 200,000 GitHub stars recently. We have one and a half million active users. Um, we have over 1,200 enterprise customers, and I'm not talking about enterprises using n8n. I pa- I'm talking about 1,200 enterprise customers using our enterprise solution out there. We have 300 ambassadors out there. There's literally one n8n community event worldwide every day. Like in September alone, we have over 50 events happening worldwide, um, in a single month. Um, we increased our revenues 10X over the last, yeah, last year. I think the median number of, of enterprise instance users more than doubled o- over, over the last, uh, months. So it definitely, we, we still see a lot of growth. Um, and I think one last thing to probably call out, I think there is still two very cri- critical things that kind of made me, made me very confident about the future of n8n. One is the kind of model flexibility. It's like what people really care more and more about, um, is kind of being able to switch models for the simple reason because new models appear literally daily, and all of them have different capabilities, all of them have different prices. People really value kind of this, this flexibility of n8n to kind of really use the right model, um, for your use case. And we see generally kind of models and kind of intelligence being more and more commoditized, and this again means that the value kind of shifts, um, more in our direction. Um, the other thing is also like when you want to kind of connect, um, multiple models together. Again, they say, "Hey, maybe we want to use for X use case, I want an OpenAI model," or, "For that use case, I want to use Anthropic one," or, "Here, I want to use an open source

3. 7:09 – 10:17

   Ads

   1. JOJan Oberhauser7:09

      one."

   2. AGAakash Gupta7:09

      I used to think I had a retention problem. Turns out I had a messaging problem. I was sending the same onboarding emails to every new user, whether they activated on day one or never logged in again. I had no idea who was slipping or why. Customer.io changed that. Every message I send is now based on what users actually do in the product. Someone hits a key activation moment, they get nudged to the next one. Someone goes quiet, they get a different path entirely. Their AI agent makes it fast. I describe the campaign I want, and it builds the full journey for me: triggers, timing, copy, even branching logic. And when I want to know how something is performing, I just ask the agent directly, and it tells me what to do next. They also have an MCP server, which means AI tools like Claude can see directly what's happening in your Customer.io workspace, your segments, your customer data, your attribution, all of it. So instead of explaining your business context every time you need help, Claude already knows it. Notion used Customer.io to personalize their onboarding and hit nearly fifty percent open rate, improved conversion by six to seven percent with localized campaigns, and pushed open rates up another twenty percent through A/B testing. The idea is simple. Customer.io helps you deliver more impact from every message you send. If you're a PM or founder and your onboarding is still one size fits all, try Customer.io at customer.io. Today's episode is brought to you by the experimentation platform Kameleoon. Nine out of ten companies that see themselves as industry leaders and expect to grow this year say experimentation is critical to their business, but most companies still fail at it. Why? Because most experiments require too much developer involvement. Kameleoon handles experimentation differently. It enables product and growth teams to create and test prototypes in minutes with prompt-based experimentation. You describe what you want, Kameleoon builds a variation of your webpage, lets you target a cohort of users, choose KPIs, and runs the experiment for you. Prompt-based experimentation makes what used to take days of developer time turn into minutes. Try prompt-based experimentation on your own web apps. Visit kameleoon.com/prompt to join the wait list. That's K-A-M-E-L-E-O-O-N.com/prompt

   3. JOJan Oberhauser9:16

      And the last one is really this kind of the community and accessibility. Like, I think n8n, we talked about the, the, the hype cycle before. Um, and obviously n8n started with the kind of very early adopters. We talk about this, like, zero-point-one percent of, of these people. And now we are kind of moving more into this kind of ninety-nine percent who are, like, moving away from these early adopters, but really the people that actually want to get things done. That we kind of lower the entry barrier more and more. We just really make sure not just technically we can only build with n8n, but literally anybody who can use a computer. That's literally our, our mission is to find, like, if everybody uses a computer and technical superpowers, and that's what, what we're thriving for, um, these days. Um, and again, it, it just generally feels like it's just getting started. There's so, so much opportunity out there, so I'm really excited.

   4. AGAakash Gupta10:00

      Wow. So I'm gonna be paying attention when you show this to us in a little bit around privacy, security, auditability, reliability to really understand what those words mean, because I feel like sometimes when we hear those words in abstract, it's hard to really understand what that means in the product.

4. 10:17 – 11:38

   Did Claude Cowork and Claude Code hurt n8n

   1. AGAakash Gupta10:17

      So I'm gonna be on the lookout for that. But I wanna dig in a little bit more on this detail around specifically the rise of Claude Cowork and Claude Code. I had on the founder of Lindy, who I used Lindy personally for some of the similar things I would use n8n for, like automating agentic workflows. He admitted Claude Cowork had a really significant impact on their business. Did Claude Cowork, Claude Code, the rise of Codex have a noticeable impact on n8n's business?

   2. JOJan Oberhauser10:45

      In the end, like, we first go- going back to, to, to the things where we got killed, like, I think that at some point OpenAI launched their, their own agent builder and think like literally we had, I think the best week ever is when we got caught, uh, killed by them. So it means like, I think the nice thing about all of those opportunities is, is that people talk about it more and see more opportunity.

   3. AGAakash Gupta11:04

      Oh.

   4. JOJan Oberhauser11:05

      Um, so yeah, that's, that was quite exciting. And we definitely, I think every company these days like then, I think people, again, hype things up and, and they see a lot of usage of those products. I think what we're definitely seeing is where, where the use cases change. I think a lot of the use cases, uh, in, in the past is where people started to, again, use n8n for this, um, per- a personal use case. They say, "Hey, again, write me an email summary." And what's changing more and more is, again, that people say, "Hey, these users cases are, are great and I need them," but where they really use n8n more is for those kind of, um, business-critical

5. 11:38 – 14:44

   n8n's revenue, users and enterprise customers

   1. JOJan Oberhauser11:38

      use cases, and that's definitely probably the biggest shift we are seeing there.

   2. AGAakash Gupta11:41

      Mm. So I was trying to find some numbers on n8n. Just in May, you guys announced your strategic investment from SAP, where you mentioned you had hit a five point two billion dollar valuation. Startup writers earlier this year put you at a hundred million dollars ARR. You just had mentioned that you guys had 10X'd your ARR over the last year. What can you tell us about your users and revenue now? Feel free to break any news. [chuckles]

   3. JOJan Oberhauser12:09

      And you see, we, we're not sharing, um, much more numbers there. I think I'll say-- can say we, we way across a hundred million by now, uh, still growing strongly And we, um, again, the, the, the user numbers already shared one and a half million users also growing strongly there as well. But we're not, uh, sharing any additional numbers right now publicly.

   4. AGAakash Gupta12:27

      I think you had said twen- twelve hundred enterprise accounts?

   5. JOJan Oberhauser12:30

      Exactly. Yeah. That as well. I think, like, the exciting thing is definitely kind of just seeing what kind of enterprises are adopting n8n. It's like, again, talk about SAP. Again, they're literally-- they're not just made the investment that you talked about with, with, at what, point, uh, five point two billion valuation, but they also, they actually, um, added n8n to their, to their product. So n8n will be available to literally any SAP customer out of the box. They can use it without installing anything else without, um, installing, like, uh, setting up anything else, without adding billing information, anything else. They can just use this out of the box. It means, like, really this kind of makes n8n this kind of AI layer, uh, inside of SAP, where you can really kind of build your agent automations inside of the, uh, of the, of the product. Also, like customers like Mercedes, which probably talk about, um, how they're using n8n and how they see, um, their company getting transformed through it. I think this is, uh, the really exciting things like... I think a lot of the, the numbers people are showing, a lot of that is, is a lot of hype. But what actually, but actually it's an interesting thing when you actually see real usage of a product, and especially really delivering real ROI. Because I think, like, people-- like, in the past, like, revenue was great because, like, revenue was very close to actually also kind of real value. But I think there's a lot of decoupling happened there. Um, and I think what our customers see very strongly is that they, like, we really deliver real ROI for them. It's like they're not just burning money, they actually see that getting something back there. I think why that is happening is I think, uh, another thing that's maybe important to call out about n8n is this kind of combination of AI with deterministic logic and human-in-the-loop. Like I mentioned, we talked about it before, AI is amazing. It's, it offers so many possibilities, but at the same time, it's, it's not a solution for everything. What you really want is kind of, kind of link AI with the, with deterministic logic, because again, it's much cheaper to use a simple if, it's much faster, and it's a hundred percent reliable. And then you still need human-in-the-loop. It's also something, again, I'll demo you later because, um, how it works in, inside of n8n, because for certain use cases, you always want to make sure there is, like, a human still in the loop. Like, and I think that is probably, um, the things people should look, look out on LinkedIn, like not just the, the, the numbers that actually share, but actually the, the real users behind it and the ROI it gets delivered for those customers.

6. 14:44 – 22:46

   Demo, what n8n does that a Claude Code agent can't

   1. AGAakash Gupta14:44

      So I asked Claude to go through my newsletter archive, and it said I had written about n8n twenty-five times in the last twelve months, which surprised even me. It also found that when I did my AI tool ranking a couple months back, we put it in the A tier above Zapier and Make. Now my audience, I've been spending a lot of time talking about PM operating systems in Claude Code. So what they typically do is, you know, they would open Claude Code, connect some MCP servers, and have a working agent in a couple minutes. What does n8n do for that person that they can't do themselves?

   2. JOJan Oberhauser15:22

      So actually, I think it's probably the easiest if I simply show you what, how it works. This is, uh, the, the starting screen for all users, um, right now. I think maybe the first important thing to call out there is, like, what you see here is our AI assistant, which the idea about it is to kind of-- I talked about before that the, um, what, what we're looking for is kind of really give everyone who uses a computer technical superpowers to really kind of lower the entry barrier. And that's exactly, uh, what it's doing here. Like, you can do the same thing you can do in a Claude Code. You can just describe literally what you want to get built, and it's gonna build it for you. Um, but actually I, I have something prepared, but before I, I show you, um, how it got built and maybe show you the, the outcome and to give you a better understanding how n8n looks like and how it feels like. Um, here you can see an AI assistant which actually works with your Google, um, Calendar and, and your Google, um, um, Gmail account. Um, how generally n8n works is you get a, as a trigger node, something that starts a workflow. In this case, you have here an, an, an, an agent, um, and you can, uh, it gets a response. And as well you have, again, different AI models. It's important to call out here that here you have, for example, we use, uh, Claude Sonnet by default. Um, it actually does get used via our own gateway, so you don't have to sign up for, for separate accounts if you don't want to. If you want to, you can obviously still use your own one as well. Um, we even have a fallback model like we know this, this, uh, language, uh, this, this, this provider is not always very reliable, so it's always good to have a fallback in case they don't respond. Um, and then we have a lot of different tools, like in this ca-this case here, list emails. Um, you can get a, a whole, uh, a single email. You can write emails, get information from your calendar, um, and so on. Here you can also see that, uh, here we have s-further actions, uh, like for example, sending an email. That is something I don't want the AI model to do by itself without asking me. So we have here this human, uh, uh, like human-in-the-loop step. But you can define literally, like, only execute that tool if you got approval before. You can define how it should be approved, um, and, and how the, the output should be what it displays you. Then let me give you a fast, um, a demo, like how it would work. In this case, you can use it externally or you can also test inside of n8n. So you can just say, um, "What is my next call?" And now I can also see this auditability piece. You're not just seeing how the whole workflow is built, you can also see how it exe-executes. So you can see here the agent gets exe-exe-, uh, did run, then it kind of used the model. It called certain tools here. It also got a memory. Um, and also here you can go through literally, um, each step, what information got sent in and what information came out of it. And you can see here, um, my next call is actually in, um, tomorrow at Thursday at eleven AM till eleven thirty. Then you can say now, for example, "Please send," uh, or, "Please create, uh, another..." Mm. Actually, "Please create another, another meeting." With my VP of sales tomorrow at 3:00 p.m. And now again, you see again each node execute. Again, if something goes wrong, you would see it red. Then you can very easily debug and say, "Hey, um, I, I now fixed that one thing." Again, you can see, um, what, what it's actually doing, how it's executing, and now you can see exactly that. If you go over here, you see now it's waiting here. You can see, uh, it's, it's, it's not doing anything further. And now it says, um, "I'm about to create a calendar event, um, with, with a certain information. Am I actually allowed to do it?" And now I can either cancel it or say add event. If I say add event, um, it goes in here, actually creates the event, um, and, um, and it's done. I think that is, uh, generally how n8n works. Obviously, like, it, it looks now quite technical, like, and the not technical people just probably wonder now, hey, can actually create that. Again, that's where the AI system comes in. Um, and I can tell you, show you how it was built. Um, here you can, for example, see the prompt. Here it can generate me a workflow agent, um, for personal projects. You kind of see what it was supposed to be doing. Get, get emails, um, archive an email, um, search for events, and so on. Um, it kind of get prompted like similarly like you would be doing in Claude Code or other solutions. And then it starts thinking. It asks some clarifying questions like which model should be used, um, and, and something else. Um, and then it actually thought for eight minutes. Again, very similar to Claude Code. And after eight minutes, it, it kind of gave, came back to you with the answer, what it built. Um, and you can see here literally, again, that's the workflow we were just working, um, at right now. Um, what you can also be doing, again, it gives you some more information about the tools that you used. Also, again, where to use the approval step. Um, and then you can also kind of, um, make, um, additional adjustments. And maybe that's, uh, something else to maybe also demo here, where we maybe just say, "Hey, um, that's great, but I, for example, very often want to schedule one-on-ones with, with, with my people in the team." So I just tell it, "Hey, please now extend that workflow. Make sure I can schedule this one-on-one meetings. It should be thirty minutes long. It should add always a Google, Google link. Um, the agenda should always be in the description, and I should only pass, like, uh, what's required. And also, like, it should get the information from Google Contacts." Um, and now you can see exactly that. Now it's gonna, uh, keep on thinking for a little bit, um, and then it's gonna come back to us. Again, in the meantime, maybe, um, just again, um, go, go back up here how we, um, just to see actually, like, while it's thinking can still, for example, like, uh, interact with the workflow here. You can still ask questions, um, as well and kind of keep on, on, on working, and it kind of keeps on working in the background, and you can just, um, ask further questions there. And then we can anytime go back here, see what it's doing, trying to, uh, see while it's figuring things out in the end. Um, and I know that it's probably gonna take a few minutes. I'm not sure we have enough time. But at the end, what you will see is that it adds additional tools to that workflow. It will add, again, another tool that it, you can get, um, contacts from a, from a Google, um, Google, from Google Contacts, and another, uh, additional, uh, tool to add actually specific calendar events for my, uh, one-on-ones. And I think that is again, um, I think that shows the power of n8n, especially, like, I- if you imagine now this one I can very easily hand over to some- somebody else because, like, like again, even if, if you're not technical, you can still at least go in here and kind of very easily understand, hey, actually, here is an agent. You can see what is the system prompt that actually has been defined. You can see again what model it's using. You can see how they are defined. You can see here it actually creates a draft of an email. Um, you can see what things get auto-defined, like all the things that are simply, uh, totally impossible in code, um, because again, it's just too much output or, again, if it's done 100% by AI, you literally, it's, it's a black box. It can do the things for you, but you have no reliability that the thing that worked yesterday is still gonna be working tomorrow as well. That's again the, the reason why we are, again, great, uh, for anything that's, where compliance is, uh, important or where you, where, uh, anything is really business-critical for you because you can really rely on, on, on those workflows because of the combination, again, because of AI, but still deterministic logic as well and again, human-in-the-loop step, um, and

7. 22:46 – 24:47

   What reliability and auditability actually mean

   1. JOJan Oberhauser22:46

      so on.

   2. AGAakash Gupta22:46

      So I wanna make sure I define for everybody watching those key terms we talked about. So starting with reliability, one element of reliability you just talked about and you showed us is, hey, Claude, they barely have four nines of uptime right now. [chuckles] So you can put a back stop model. If Claude is not available, use this other model. Is there any other component of reliability that people need to make sure they understand?

   3. JOJan Oberhauser23:12

      Um, like there's probably multiple ones. Like another one to call out is literally it's running on, on your own infrastructure. So it means, like, if you talk about, um, you can literally, like n8n can run on your infrastructure. You can literally self-host it. So if you want to make sure, like, it's close to your own data, like there's no internet problems that can really, um, cause any problems there, you can run it there. Um, you know, like for example, like all the t- all the tools, like for example, this get email tool or anything else, is literally code we created, that's literally tested, and we, we, we, we, we, we maintain. So even if the API change or anything like that, literally what all we have to do is we change the, the, the kind of the code one time. It gets changed for everybody. We know it's the kind of this reliable, uh, reliable piece of tool that, that, that works for everybody and not something that literally any person worldwide has to re-implement every time and certain edge cases are forgotten. Like, we can really think very deeply a-about the quality, um, there. On top of obviously we have like, um, this kind of whole platform. They not just, again, create workflows, but you can also, um, deploy them on. So again, like if something goes wrong, you can just have retries, um, you know, where the data gets stored, and so on. I think like there's a lot of things where it simply takes out- A lot of the, the, the worries and the thoughts you normally have to, to think about when you deploy, um, something else or just get code written by another solution for you.

   4. AGAakash Gupta24:37

      Mm. So another word we talked about was auditability.

   5. JOJan Oberhauser24:41

      Mm-hmm.

   6. AGAakash Gupta24:41

      And I think the way I'm seeing it, although I'm curious if this is, like, technically accurate. Here's

8. 24:47 – 29:40

   Ads

   1. AGAakash Gupta24:47

      what I did last week that changed how I think about AI coding tools. I built a full internal dashboard in Bolt.new. Took me about two hours. Looked good, worked well, had all the functionality the team needed. Then I did something you can't do in most AI coding tools. I sent the actual source code to my engineer. In Base44, in Lovable, in most of the tools, the code lives inside the platform. Your engineer can't pull it. They look at your prototype, say, "Cool," and rebuild the whole thing from scratch. Bolt.new is different. The code is pullable, real production-grade front-end code. My engineer opened it, reviewed it, and started building on top of what I'd already made. No rebuild, no start from scratch. The prototype became the product. I use Bolt.new every single day. I built my entire paid cohort platform on it. Seventy-five product managers log into something I made in Bolt.new. Try it yourself, bolt.new/aakash. That's bolt.new/aakash. Today's episode is brought to you by Jira Product Discovery. If you're like most product managers, you're probably in Jira, tracking tickets and managing the backlog. But what about everything that happens before delivery? Jira Product Discovery helps you move your discovery, prioritization, and even roadmapping work out of spreadsheets and into a purpose-built tool designed for product teams. Capture insights, prioritize what matters, and create roadmaps you can easily tailor for any audience. And because it's built to work with Jira, everything stays connected from idea to delivery. Used by product teams at Canva, Deliveroo, and even The Economist, check out why and try it for free today at atlassian.com/product-discovery. That's A-T-L-A-S-S-I-A-N.com/product-discovery. Jira Product Discovery, build the right thing. Do you know how to take an AI product from idea to development to evaluation to deployment and eventually to scale? That's exactly what Product Faculty's AI PM Certification helps you do. I even took the course myself. You'll learn directly from Rohan Varma, the product lead working on Codex at OpenAI. You'll go deep into AI prototyping, evaluations, agents, AI-native workflows, Claude Code, OpenClaw, latency, cost, guardrails, RAG, routing, fine-tuning, and production systems. You'll even build your own AI product as your capstone with unlimited one-on-one support. So if you want to stop just learning AI and actually build AI products that work, join Product Faculty's AI PM Certification on Maven. Five thousand plus students have graduated, and they have one thousand plus reviews. Use code AAKASH150 to get $550 off your enrollment. When you build these things in Claude Code, the fundamental sort of unit, the source of truth is, is code. It's a bunch of files that kind of live in some Python script, some random markdown. It's kind of like, kind of like built on a brittle house.

   2. JOJan Oberhauser27:46

      Mm-hmm.

   3. AGAakash Gupta27:46

      Versus here, what we have, I think, and how I experience the product-

   4. JOJan Oberhauser27:50

      Mm

   5. AGAakash Gupta27:50

      ... as a power user, is the core is this workflow, like, editor.

   6. JOJan Oberhauser27:55

      Right.

   7. AGAakash Gupta27:55

      And everything is thought of in terms of more stable components, where each of these lines and nodes in the workflow is something that you guys really put your foot behind.

   8. JOJan Oberhauser28:06

      Mm.

   9. AGAakash Gupta28:06

      It's stable, it is working, and it's something that somebody can see versus just in code.

   10. JOJan Oberhauser28:12

       Exactly, especially like this auditability piece. I think that, again, there are, there are multiple things that like next to, to what you can see what has been built. Again, it is the workflow itself. Again, you can also see how did it run in the past. Like say, hey, like, here are the last execution step, for example, went through. So you can see literally here step by step what, um, how did it actually run. Again, it started here, it called an agent, it called those different tools. Again, you can see the information went in and out of the product, um, and you kind of get so, um, like you can so, uh, literally identify everything that AI really did for you. Again, with code, it's, it's the input and you see the output of the whole thing. But what really happened in there, you have no idea. Like, why did it take certain decisions? For example, we could now add here, for example, between, um, this, um, the output of the agent and the output, like an additional node that adds, for example, the deterministic logic and say, hey, if it outputs like, again, something that's smaller than 50 or larger than 50, do something very differently. Like you can be again 100% sure it's always doing this, and you can then also, uh, after it's inspect, why did it take a certain path? Because like that number actually got most output, then it went there. And if it didn't go, um, that was a certain thing, you know exactly again how we can debug it and change it, and then you can literally rerun half of the workflow from that data point, uh, from that, uh, point in time, um, and can then, um, get it to the kind of state and the quality you actually require.

9. 29:40 – 32:21

   Handing a workflow to a teammate, and where n8n shines

   1. AGAakash Gupta29:40

      So another thing that you had mentioned that's important about n8n is this idea of being able to pass it to somebody. So can you-

   2. JOJan Oberhauser29:48

      Mm-hmm

   3. AGAakash Gupta29:49

      ... show us, like, what is the-- If I wanted to give this, let's say I'm a manager and I want to give this-

   4. JOJan Oberhauser29:54

      Mm-hmm

   5. AGAakash Gupta29:54

      ... to somebody on my team, "You should also have this workflow," how would I do that?

   6. JOJan Oberhauser29:58

      You can literally, um, maybe add them a few pieces there. You can literally share your workflow with other people. In case I just in-invite them there, and then they can, um, access it and kind of change it themselves. Um, also, um, few other important pieces, you can literally have a version history. You can just kind of see how it changed over time. Um, again, means a person can actually go through and kind of see, hey, okay, who made changes, um, um, what went on there. You can even have, um, things like, again, publish certain versions where you kind of describe what changed between them, similar to Git. Um, another important one is, um, actually where did it go? Um, you can even ve-very easily export them, so you can literally send another person an email, uh, a workflow via email, uh, if you want to. And we also something that's not in this version yet here, because I think it's an older version, so I didn't update it, is what we have is kind of review steps. You can literally say, "Hey, I, I made this change. Please review that workflow, see if it does the right thing for you." Um, and then you can approve it, and only then it gets, for example, um, published.

   7. AGAakash Gupta30:58

      Mm. Okay. What is making enterprises like the model companies themselves use it? Here we've shown the email and productivity assistant.

   8. JOJan Oberhauser31:09

      Mm-hmm.

   9. AGAakash Gupta31:09

      I guess for me, I feel like I-- other products could also do this.

   10. JOJan Oberhauser31:12

       Mm.

   11. AGAakash Gupta31:12

       What are those use cases that n8n is so uniquely equipped for?

   12. JOJan Oberhauser31:17

       C-certain ones are, for example, again, like the, the more important reliability and security is obviously more it shines. Again, we have definitely quite a few companies using that for use cases like, uh, like as a survey of like, uh, s-security, uh, security, um, orchestration, meaning like, don't know, every time an email arrives with an attachment, scan it, um, then archive it. Um, i-if there's something is going wrong, um, inform certain people or, or make something certain gets, uh, started, or anything that, um, relies on like more sensitive information. It could be employment data or could be also be even things like on and offboarding employees where you don't want to, like you really don't want to have anything go wrong because you don't want to have an half onboarded employee. You don't want, you don't want to have even less than half offboarded employee either. Um, we have people using us for, for DevOps use cases. Um, there's definitely like a wide variety, but I think in anything, if you think about anything where you really want to kind of want to be 100% sure that, that it really gets done and not just half done.

10. 32:21 – 41:44

    What PMs should build after their first agent

    1. JOJan Oberhauser32:21

       Um, so means well as anything that is not a private use case, um, that is very often why people normally choose n8n, um, over other solutions.

    2. AGAakash Gupta32:29

       I feel like you guys have so much power that sometimes it's almost hard for people to figure out, at least when I talk to them about it, about like what is the specific thing that I should go do next? So we've showed them the email and productivity assistant.

    3. JOJan Oberhauser32:42

       Mm-hmm.

    4. AGAakash Gupta32:42

       We've given them a preview. This is amazing for security and compliance use cases. If you're a product manager, you've built the email productivity assistant, what is the like second, third, fourth thing you should build in n8n?

    5. JOJan Oberhauser32:53

       And I think this is probably more a question for the person. It's like, think just the, the problem with a very hard sound apart from like hours because you can do everything. In the end, like you want to do the thing that really most impactful for you. Like if in the end, you want people to think about like, what's the thing I'm doing literally every day? What is something where I'm using multiple applications, where I'm very often copy-pasting things be-between applications, um, where I don't know, maybe also get information from one place, put it into like a ChatGPT or, or Claude, and then do something with that, create a PDF report. Um, anything where, where you think like, "Hey, I'm, I feel like I'm wasting time," um, then at least you can get it started for your personal use cases. Um, for kind of company, uh, wide use cases, I think that is, uh, very often, again, uh, they normally suggest people to start with something small. Like we, we have a lot of companies that just want to start with something huge to say, "Hey, how can I totally transform X?" And then they, they, they spend literally building and building and building for, for, for weeks, and it, it's a very hard thing because at some point, uh, if you don't have experience yet, um, y- y- you maybe think build things the very wrong way. And I think what, what we have seen that sometimes workflows that literally just contain like an, an agent, a, a few other nodes, they're actually very often the most impactful ones. And again, not always do you need actually AI for everything. Again, AI is one tool of very many. Like we have one company that literally used n8n to kind of do password resets of employees. They have a lot of employees. Um, a, a lot of them, um, for-forget their passwords regularly, and they were locked out for literally hours. They cut down the, the time like immensely, and it saves them like literally multiple full-time employees a year in time saved. So like it, it-- I think it's not about like the craziest, um, one. It's really the one, "Hey, the h- where can I get started easily? What do I think is the most impactful one?" And then build, build your way up from there. Especially are these simpler ones are a great kind of to kind of show off to other people. Just say, "Look what, what I built. See what's possible. I spent literally two hours on that, and that's the impact." They get other people involved, and then the very interesting thing is like, especially in org is like you-- the more you talk with other people that actually build, you get other ideas as well. "Ah, like I built this." "Ah, yeah, that's a very similar use case I have as well." And you adopt it. And on top, we also have, um, our, our template library online. There's over ten thousand workflows for people, um, there. So you can just go in there, say, "Hey, I'm using tool X, Y, and Z." Um, or you can say, "Hey, I'm in sales," and you get literally for each use case, uh, for, for each use case, you see, literally see hundreds of them, uh, u- um, workflows and, and agents other people have built. But it can either use them to get inspiration or literally take them, um, um, simply as they are, add your own credentials, and you can get started there.

    6. AGAakash Gupta35:42

       Can you show us the template library and maybe give us some inside information about some of the popular ones and the ones that PMs should be looking out for?

    7. JOJan Oberhauser35:51

       Yeah, you see, um, the templates page. As I mentioned before, you can get started in saying, "Hey, I want to use Google Sheets." Um, um, and then you see like a lot of use, um, in this case, for example, four thousand, um, workflows that use Google Sheets. You can then say, "Hey, I want, um, I'm in sales." Now you see all the sales use cases that use a Google Sheet. Um, but we, for example, uh, see very often is a lot of people use also n8n for web scraping. Um, just, um, get information from web pages, get information from different data sources, um, enrich it in, in a certain way, save it, in this case, for example, to Google Sheet, and then send, um, that information to, to Slack. Um, or we can go over to-- So remove everything here. And go, for example, um, marketing. Um, maybe actually let's, let's get in, in IT ops. Maybe there's, um, one interesting ones. Um, yeah, here for example, you have, um, you can say-- the, the nice thing about n8n is, again, going back to the reliability piece, like if something goes wrong, um, you, you can get informed in whatever way you want. You can get informed, um, via an email. You can inform this case as an example how you can get informed via Telegram. Um, um, again, here we see like, um, somebody's used n-n8n for monitoring here. They, for example, track, hey, are SSL certificates on my server still up to date? And then it kind of, uh, alerts them on, on Disc-Dis-Discord. It sends it, uh, also to Notion. So there's a million use cases, um, out there, um, as you can see, um, that you can use for, for literally anything. We definitely see a lot of monitoring use cases, um, in, in there as well. Again, very often also in combination with AI, because I think that's again, where you want to react very fast. The, the reliability piece again matters, um, where you can say, hey, again, if, um, by default just inform me and then maybe try to auto-fix it, then use AI. But you can really kind of very clearly define again the different steps that should be taken first, and then you can, um, um, make sure again, again, for example, the inform- the informing pa- piece always happens to a hundred percent. But this, uh, kind of auto-fixing part with AI, for example, only happens later. And again, um, think that's again, one of the things where n8n really shines. Um, and I think I just advise anybody to just look through there and kind of get some inspiration of people, what people are actually building with n8n.

    8. AGAakash Gupta38:12

       Yes, a lot of scraping use cases. I also have used n8n a lot to teach AI. Like it's a really good-

    9. JOJan Oberhauser38:18

       Mm-hmm

    10. AGAakash Gupta38:18

        ... place where you can very easily fine-tune a model. You can very easily set up a RAG system. You can set up a vector database. So if you want to learn the fundamentals of AI, I think it's also really good for that outside of just workflows. So one of the areas we talked about was the ease of use. So can we see how the AI assistant has done on extending our personal productivity workflow?

    11. JOJan Oberhauser38:40

        Surely. Okay, here we see the prompt. So this is thought how I thought about it, and then it said extend it here. It actually also ask a question in between. Um, and now we can maybe go into the workflow. We can see on the side, um, let me extend it actually a bit more. Um, and now you can see how it extended it. So what it actually added now is, for example, this lookup. So means like it can now look up contacts on, on Google. Um, you also can see it, it actually executed, because what it's doing as part of the process already tries to kind of test the workflow for you and runs it. Um, you can also see here this approval for one-on-one. So if you want to book a one-on-one, um, you can also see here the same thing again. Says, "I'm about to book a thirty-minute one-on-one." Um, it will include a, um, a, a Google link, and do-- are you okay if it gets added to the Google Calendar? And only then, again, it, it does that.

    12. AGAakash Gupta39:33

        Okay. So it took five or six minutes it looks like, but in the end, you're gonna get an updated workflow that actually makes sense, that works with your tool.

    13. JOJan Oberhauser39:44

        Exactly.

    14. AGAakash Gupta39:45

        So your episode is coming right after Wade Foster, CEO and founder of Zapier. So we have to ask the obligatory question, Zapier versus n8n. What can n8n do that Zapier can't?

    15. JOJan Oberhauser39:57

        I think like the, the thing is like the, the great thing about n8n is, is like we focused from the very beginning about power and flexibility. I think what you really... Again, when you, when you get the most value out of n8n is you kind of, if you don't want to be kind of locked, uh, into kind-- this kind of automation where you kind of have something, um, where you want to be sure, like, uh, you build something today. You think, again, I mentioned before, you want to start with something simple, and then kind of you want to kind of build on top of that and, and, and, and kind of build it up. I think that's again when n8n shines very strongly, because again, we have this power and flexibility built in. We have, um, uh, you have things like code nodes where people can fall back to code anytime. We have this very, um, um, uh, complex, um, agents can be building. Again, it's not just, um, basic agents. You can literally add your own memory. You can put out, and you can add outside powers to make sure the output is always a certain format. You can have, again, as I mentioned before, you can have different models. You can have fallback models. You have this human-in-the-loop, loop steps. I think, um, a lot of the things that, that I think when it becomes a little bit more complex rather than, than simple use cases. Again, like, uh, uh, uh, Wade is, is a great guy. Um, I, I've always enjoyed talking to him. I think Zapier is a great product. But again, as mentioned before, there's, uh, always the right use case, um, uh, for the right product. Um, and I think like, um, especially again, if it's about where you need power and flexibility and a very deep, uh, integration to AI, it's when n8n really, really shines.

    16. AGAakash Gupta41:26

        Mm.

    17. JOJan Oberhauser41:26

        And again, obviously, I think it's probably not worth mentioning, but I think obviously if you want to self-host it, um, and, uh, and things like that, um, that is where, where obviously n8n, um, is, um, uh, is I guess where, where Zapier I guess cannot compete considering their, their SaaS solution there.

    18. AGAakash Gupta41:42

        Mm. That's an important differentiator

11. 41:44 – 44:23

    Why sprinkling AI on top only gets you 10 to 30%

    1. AGAakash Gupta41:44

       too. So I wanna learn more about your guys' growth, 'cause you have a pretty fascinating history. We talked about how many times people have called you obsolete. You guys actually existed before ChatGPT. n8n has been around, and I think that might have been one of the first times. And you talked about in a podcast how you said you were scared a little bit. [chuckles] But sprinkling AI on top would only give you ten to thirty percent growth, not the 10x growth that you guys have been seeing year after year since. How did you make that call, and can you break that down for us a little bit more? What does it mean to sprinkle AI on top versus make it core?

    2. JOJan Oberhauser42:23

       Sprinkle AI on top is, what I see is like somebody tells you, "Add AI," and you say, "Hey, here, I add this AI button somewhere." It does something with AI. Where I see again m- become-- but, but we have with it is like really think about, like, how it can really become part of the value chain. How can we really kind of not just add AI to the product, but really kind of make sure people actually build with n8n? Um, meaning like when, when the idea was like how when people think about if I want to build an agent, like there should be built-- want to build an agent with n8n. So we're not just-- So we're actually seeing like you can actually, um, get the, the real value of-- get real value for people out there. And I think like the, the thing is probably like when people get asked, add AI to their product or build an AI agent or kind of sprinkle like, uh, like we want to be the solution that kind of helps them to do that, because that's where the real value lies and where you can under-- again, see a, a ten percent or a thirty percent, uh, growth. But literally, like we-- like last year, we grew, grew literally 10x exactly for that reason, because we were part of the value chain, and we kind of empowered people, and we kind of provided real value for them.

    3. AGAakash Gupta43:29

       I think a pretty crazy stat you guys released is eighty percent of workflows on n8n now use AI agents.

    4. JOJan Oberhauser43:36

       Mm-hmm.

    5. AGAakash Gupta43:36

       Obviously, that would have been zero [chuckles] at the beginning of twenty twenty-three. So is that your twenty twenty-two users adopting AI? Is this a new crowd? If so, what happened to the old crowd?

    6. JOJan Oberhauser43:47

       It's actually people adopting. It's like I mentioned before, like our users are the, the tinkerers. Like they're, they're, they're, they're, they're very interested in, in, in, in, in, in kind of checking out the latest tech and kind of really making their day-to-day more efficient. And obviously, AI is the tool to do that. And all of them were super interested in actually adopting AI and kind of really making it part o-of, of, of their, their, their daily usage. And I think that's, uh, the, um, how we became so successful because we just made it, um, super simple to kind of really, um, get kind of people into this kind of-- to g-get people started with AI much faster and easier than with other solutions

12. 44:23 – 54:05

    Killing the lead gen target and per-seat pricing

    1. JOJan Oberhauser44:23

       out there.

    2. AGAakash Gupta44:23

       And how you've grown this thing is very different. Like the average company doing this type of growth, they would embrace per-seat pricing. They would have lead gen targets. You killed your lead gen target, and you refused per-seat pricing. Why?

    3. JOJan Oberhauser44:37

       The thing is like we are in this very lucky situation that we can think very long term. It's like I think that most AI companies out there, they, they maybe grow their AI, ARR very strongly, but at the same time, they're just losing a lot of money. So it means like you have to kind of show a lot of growth to kind of get more money from investors. Um, and it, it means that-- But what forces them to kind of really think very short term, like how can I get more ARR now to g-get more money tomorrow? Um, and then the whole cycle repeats. Um, because of the way n8n is built, like we are actually sustainable. Like right now, we, we are actually creating a profit. So it means like we don't rely on more investor money, uh, anytime soon or at all, uh, if we keep on doing what we are doing right now. Rather we can think about like what sets us up for success in the long term. And I think right now, like being able to kind of just capture a lot of the usage and not think about again, how to in-increase revenues as fast as possible, but how do I really kind of create the best product out there and how can we create a lot of value for our customers, and how can we get to capture a lo-- uh, as much as possible, uh, of the, of the opportunity out there. Um, that's why it's obviously it's a very much easier, um, decision for us to do something like that than for many other AI companies out there right now.

    4. AGAakash Gupta45:49

       So the lead gen and per-seat pricing, those are pretty public for anybody who studied n8n like me. What might be some other metrics you've deleted or a metric you're going to delete that people don't know about?

    5. JOJan Oberhauser46:00

       Metrics we deleted, maybe I think we can talk about, uh, um, internal thing we are, we are focusing on very strongly is like one thing is obviously like what we want to achieve like in, in certain goals regarding users, um, and AI in the long term. But the other thing is like how we want to kind of build the company, um, how do we think n8n should function. And we set this internal goal like we want to reach a billion users with less than a thousand employees. Means like we don't, like we, we don't want to grow our headcount. Like we, we obviously grow it because we have to, um, but it's the opposite of, of what we're interested in. We kind of we want to build this very efficient org. The, the, the idea behind it is very simple, like organizations come to us to help them to kind of transform the organization to become more efficient, to use AI better, to automate more, and so on. And if you're not kind of doing it internally, like you're literally hyp-hypocrites, like that, that's horrible. [chuckles] Like I hate nothing more and especially I think we cannot do a, a great job. Like we can only help them transform if we actually transform internally as well and kind of had our own-- learned our own lessons, know what worked, what didn't work, what was impactful, what wasn't impactful. Um, and, and for that reason, again, we, we are not, we are not talking about like how fast we have grown our headcount, um, over the last years because there's literally nothing we think is, is, is a positive thing right now. We are actually doing the opposite, where we kind of try to stay as lean as possible, um, and really try to kind of build the organization of the future to a-again, help our customers and our users do exactly the same thing as well. Maybe another thing to point out is like what, what we're not, what, what we're not caring as much about is like obviously as, as kind of this freely available product, is obviously like you have obviously pro-- users that pay you and users that don't pay you. Um, and we obviously could always kind of, for example, focus on how do we get people that currently don't pay us to use our free, um, free version that self-force us to our hosted solution that we actually, uh, earn revenues. That's nothing we're doing at all because it doesn't matter for us. Like all we care about is that they're using n8n, and we know it like we, we're generating value for them. At some point, like they either-- or we work for an organization where they're gonna want to pay us at, at some point in the future, but again, we don't care about if they're paying us right now or not. All that matters for us is they're, they're happy n8n cu- um, users, um, they keep on using us in the long term.

    6. AGAakash Gupta48:18

       Mm. Fascinating. For somebody who was a VP of growth at Apollo.io, where I was just focused on free-to-paid conversion, free growth, that's very interesting and very different for people who might not realize way to approach those free users. You mentioned the employees, and I actually wanted to talk about that. I think at some point you had said you want to be the first billion-dollar company under five hundred employees. I think your career page puts you at around a hundred and sixty employees. But LinkedIn, it seems like a bunch of people are wanting to associate themselves with n8n, and you have like over a thousand people listed working at- n8n on LinkedIn. So how many people really are working at n8n right now?

    7. JOJan Oberhauser49:00

       Currently, we are around three hundred and seventy people, um, in, in the org, and we, we are hiring. We're probably gonna be around five hundred by end of the year. Um, you're right. The, the reason why the number is so inflated because we obviously have a lot of, um, partners and agencies working with us, um, and they don't very often appear as n8n employees. Um, also, the other thing you mentioned is, like, our old-- I, I talked about right now a billion users less than, within a thousand employees. Our old goal was a billion ARR with less than five hundred employees. Um, we changed it for two reasons. First, uh, we became, um, like, we, we did much more on the enterprise side of things, and we realized, hey, as long as people-- as long as the, the buying side is not done by AI, we cannot do the selling side by AI either. Like, it's still a people game. You, you need... Like, people don't si-sign a half-a-million-dollar contract without talking to anybody. So, like, it, it's gonna stay very people-heavy for a long time. And the second thing, uh, was also that we realized, like I mentioned before, like, what matters in the end is fast adoption. And kind of talking about, like, the, the, the one billion ARR was, was not set because we said, "Hey, we want to make a lot of money." It was set to kind of as, as kind of my sort of being, like, a, a big impactful organization. Uh, but at the same time, it still confused people internally, uh, even, or even externally saying, "Hey, like, all it is it's just about money." And we kind of learned to be very clear saying, "Hey, it's not about money. It's about, about adoption." That's why kind of this, this, this value switching away from ARR to really, um, um, actually users, uh, made much more sense for us.

    8. AGAakash Gupta50:28

       I have one more question about how you guys grow because it really is different from everyone else.

    9. JOJan Oberhauser50:33

       Mm-hmm.

    10. AGAakash Gupta50:33

        Like, let's say, like, Whisper. Like, I think they're like a company that I put in your similar category. Like AI, they're just taking off like crazy. So they launched their note taker. Their entire focus was viral growth. They even bought like rickshaw ads in Delhi and things like that. You guys have pers-- been perceived as much more quiet. Is that like a strategy? Is that your personality? What's going on there?

    11. JOJan Oberhauser50:56

        Um, we actually... Yeah, I think it's probably also part of my personality. I'm not a person that has to be literally anywhere, and, um, I, I, I'm more the, the introvert kind. So, um, this kind of things don't come naturally to me, and I, I'm not a person that really has to be anywhere on stage or... So I think there's probably, I think there's definitely also a component a-about me there. But generally, like, as an org, I think, again, we are also a European organization. I think we are probably more the kind of, uh, underpromise, overdeliver, uh, kind of people. So we don't really, uh, in, in the business of kind of really kind of buying any kind of viral growth there. Again, like, we think more long term. Again, like, we became not viral because we paid people to talk about us. We became viral, uh, last year because we had people being excited about n8n, getting a lot of value out of it. They wanted to share it. We had people that kind of created, um, their business around n8n, and then, um, and kind of generated, um, own money, like, um, m-money like that. And I mean, this obviously kind of makes it even more interested in, in actually talking about it because it's this nice win-win situation. It's like the more they talk about it, the more visible they are. The more visible they are, the more visible we are, and then obviously they get, get more contracts. So it's this nice, um, piece there as well. Again, we, we had from the very beginning this very strong community focus. I talked already before about, again, having an event literally every day, uh, on the board somewhere about n8n. I think that is, again, this kind of more long-term thinking. We know, hey, like, investing in the community is nothing that kind of gives you growth tomorrow or next week. It's, again, a thing that takes many, many years. But once you have it and once it works, it's amazing. And also by, by, even by, by now, so, like, most of the enterprise users actually comes from our community people. Like, they literally use n8n privately. They get a lot of value out of it. They say, "Hey, I actually work for this huge org number X," um, whatever, and then I have an also use case there. So I, I can actually automate things there, and then they do the first, uh, internal use case. Then they get other people excited. We have literally communities inside of l-large orgs that are, like, over a thousand people are, uh, big, and that have their own internal community events inside of large organizations, inside of the system integrators. I think that's just amazing. It's nothing you can, again, do very fast, but I think this is kind of more long-term sustainable things, um, that, again, worked out very well for us.

    12. AGAakash Gupta53:09

        Very European indeed. Although if you think about like a Lovable or something, they are more viral. So I think there is also a component to your personality and how you have grown this thing. So that's how you've grown this thing, which is just amazing. I want to talk a little bit about product. You said three hundred and seventy employees. How do you structure your product team?

    13. JOJan Oberhauser53:29

        Right now we have, we have like squads, um, which are like between three and five engineers, have one to two PMs, um, and one designer. Um, they have a, a VP of, of, of, of product. Um, underneath we have a director, and then we have, um, uh, underneath our PM teams. Um, and then, um, the engineers report to an, an engineering leader again, and, um, and, and designers to theirs. And I think that works actually very well for us. We have this very small units that can kind of, um, work, uh, very fast and efficiently and kind of, um, get things, uh, pushed out quite fast without a lot of the overhead.

13. 54:05 – 1:08:55

    How n8n builds product and hires AI PMs

    1. AGAakash Gupta54:05

       When did you hire your first head of product, and how did you make that decision?

    2. JOJan Oberhauser54:10

       We probably hired him around April twenty twenty-one, so it was around the same time we raised our Series A. By then we had, um, multiple engineers already, um, but, and I was obviously still totally involved in the product. Um, but I also realized, like, it's, it's important to have somebody who has the experience to really lead a product org, who can spend more time on it, who can go deep on the problems. Um, and, and I, I met him, and from the very beginning, he like... I think he, he mentioned the past n8n is a product he would love to have thought about himself. Like, he's literally a productivity nerd. He has like a shortcut for literally anything on his computer. Um, how he does everything from, I don't know, shopping to meetings, all it, it is super efficient. He's probably one of the smartest people I know, [chuckles] and it was just very, very clear from the very beginning that he is the, the right person to kind of lead a, a product company like n8n, uh, on the product side of things.

    3. AGAakash Gupta55:07

       Mm. So what did you hand him and what did you refuse to hand over?

    4. JOJan Oberhauser55:11

       Originally, I stayed very close. Like, again, I was still very involved. Again, also on the pro-uh, um, pro-product side, uh, product development side of things. I literally still, um, merged every PR, so kind of really, kind of sort of reviewed everything that wa-kind of was happening there. Um, but that obviously changed more and more over time. Um, there I kind of, kind of very early on kind of, um, owned the roadmap. Um, but again, we, we always talked about all the things that were happening about literally every single feature in the very beginning, uh, and, and discussed things, how we think they should work. Um, and again, as the product team grew more and more, actually I, I became f-for a lot of the things, uh, more, more removed. Um, I still have, for example, like, for example, check-ins with the design team. Um, we had this, this, this case where after at some point they, they, they built a, a feature, um, and they, they made certain decisions that honestly went a-against what I thought was the right direction to go in. And then just realized, hey, actually, I, I probably have been too far away from the product at that point in time. And then I said, "Hey, now actually I have to kind of go back there." I literally, I, we have now still to this day, we have regular meetings every week with the design team where they kind of show me what things they're exploring, what things they're thinking about. I can give, uh, feedback there. They can hear me what, what, what I'm thinking about things. Um, I can give f-uh, feedback very early on, and it just gives me a better, uh, idea where everything is going. And just apart from that, obviously like, and it is still my baby, I just want to be very close to it and kind of understand what-what's going on there. And again, just, uh, make sure it's, it's always on the right level. And then there's obviously certain things where it matters less, and other things where it, it's worth to kind of really go a little bit deeper and understand how exactly they're thinking about things and, and, and what direction it's moving.

    5. AGAakash Gupta56:56

       So when a PM is talking to you these days at the CEO level, what are the ways they should be talking to you? What are the mistakes they might make?

    6. JOJan Oberhauser57:05

       It's like it has to be obviously o-on, on, on the right level where it really matters in, in the sense of like, obviously it comes... That, that hence what they're coming, uh, to me about. Is it about a certain direction they want to be going in? I think that's, that's exactly the, the, the right kind of level where it says like, "Hey, we think we, we want to go in that direction. We want to explore those things." Because then very early on I, I can give feedback. I think it's right, um, or we can obviously say very early on it's not the right direction and, and can also give feedback there. Um, and we don't, um, waste, waste time and resources there. But again, everything is always happening very closely with David because, like, he's the product leader. I, I, I, I trust him literally unconditionally. He's just a super smart guy. But I think it's always, uh, good to kind of, uh, bounce his ideas off. Um, obviously where it's less helpful very often is, like, if it's obviously just in the detail. I think, like, I still like the des-design demos because again, that's literally where I want to be in the detail because, like, I think that's, is, is, is important for me. But if the last thing, again, any, uh, senior leader wants if they... For, for you get involved for every small decision, because most of them don't matter. And I think, like, if it's about that, "Hey, should we pu-put a button here or there?" Or any kind of smaller things that don't really kind of make a substantial difference for the product experience or the opportunity, I think that's probably not, uh, what you want to be talking with any senior leader of.

    7. AGAakash Gupta58:24

       One hundred percent. So I went and found an opening [chuckles] for n8n. I wanted to just see what are you guys looking for for your PM. And a couple things stood out to me when I was looking at this. Obviously, it's a very empowered job. You own the p-strategy and roadmap. But where I wanted to go was, what do you require? So some of the things you talked about here, two to three years on platform. Well, that's-

    8. JOJan Oberhauser58:48

       Mm

    9. AGAakash Gupta58:48

       ... because it's a platform role.

    10. JOJan Oberhauser58:50

        Mm-hmm.

    11. AGAakash Gupta58:50

        You talk a lot about technical depth.

    12. JOJan Oberhauser58:52

        Mm.

    13. AGAakash Gupta58:52

        Some of the words we were just defining earlier in this podcast, reliability and scalability trade-offs, holding your own in an architecture discussion. It seems like a very technical role. How technical do you have to be to be a successful PM at n8n or more broadly, an AI PM these days?

    14. JOJan Oberhauser59:13

        Like the experience we made is the more technical, the better. And honestly, almost every role we hired, um, like also our designers are very technical. We also expect them to kind of, kind of, um, u-use Claude Code, uh, create a function, demos, and just get a, they, they get a good understanding there. So like, I think the more technical, the better because you just get a... I think you, you can ma-have much more impact. You also kind of understand, especially, and recently our user base was quite technical, so it was even more important than now. Um, I think also now where, again, we talk about power and flexibility, the thing we never want to give up is this power and flexibility. Ne-people should never kind of feel locked in. And I think, like, technical people are, are great for that, especially because they kind of really understand very deeply, like, what's possible, like how much work is really something we need to do there. Generally, we always look for the kind of rising stars. I think that worked very well for us. Um, also we look like for really passionate builders. Like one of our values is literally like a, a culture goal, like we are builders. Like we want people that kind of very often that are tinkerers. Like we have people that do like have home automation running at home and kind of get very excited about those things because again, they are very close to, to our users there. And also people that kind of understand what it really means to kind of build something that scales and something that really is again, reliable. Um, again, that's why we wanted people actually not just build a basic agent, but actually understand, hey, you want to probably go deeper. You also want to think about evaluations. You want think about how does it scale, how to kind of, uh, improve it, um, h-how do we kind of make sure it's, it's really reliable and all of those things. And I think, um, all of those, um, if you find that this kind of trait, traits in people, we just understand, hey, they're not just able to operate very high level, but can also go deeper if they have to. That is normally a very good sign for us, um, and that are good, are good PMs. Um, so again, we, we found some very amazing ones, but we also had- It's always quite a, a hard time finding the kind of more technical PMs out there, um, over the last years

    15. AGAakash Gupta1:01:12

        I keep hearing this from AI CEOs, even less technical products than you, that they want these pretty technical PMs. One thing I don't see on this list is evals. What do you think about evals? It's one of the hottest topics in product.

    16. JOJan Oberhauser1:01:27

        Mm.

    17. AGAakash Gupta1:01:27

        Should PMs be owning the evals or at least defining, you know, the golden set? What is the role of evals for PMs at n8n?

    18. JOJan Oberhauser1:01:36

        Um, we actually have an own team. It's called the AI Trust team, which actually owns them. Um, and they literally also have built an internal product around it as well, um, to, to actually make sure that evals are run the right way, like especially like a, a com- like a complex product like n8n where we, again, where you want to test against, and not just a simply input output, but if you want to see in ca- in case, for example, n8n like to have the right tools been called and so on, it becomes even more important actually not just get something that, that just again works for, for, for, for half the people out there, but we really need something very specific. So again, people have to understand evals very, very deeply. And I think actually evals are... I think it's, it's kind of sad. People talk about them a lot, but people don't really [laughs] uh, use them as much as they should, because I think they're, they're probably not, not very fun to create. Um, they're very hard to create. I think there's definitely still a lot of opportunity in that space as well. And honestly, also still a lot we could be doing much better at n8n. We also have obviously an eval product, um, uh, in there as well, like you can create your own evals, um, in n8n. Think they're very valuable. But honestly, I would love if more people would actually ask for them and actually create more evals, because I think that's again, been talk about being able to kind of run things reliably. Evals are the way to actually ensure that, especially if you want to switch, for example, a model or other things like that, um, for, for no matter re- whatever it is, just a ti- time reason you want a faster throughput or price or whatever, you need evals to kind of just get very fast, a good understanding if the performance is the same way or, um, uh, how we can change the prompts very easy to kind of make it, maybe make it work for those models as well.

    19. AGAakash Gupta1:03:13

        So a lot of people who are watching this podcast, they're seeing this role [laughs] and they're like, "I want this role." So if they want this role, I have a two-part question.

    20. JOJan Oberhauser1:03:24

        Mm.

    21. AGAakash Gupta1:03:24

        How do they get noticed by n8n? Like, what are you looking for to interview people? And then how are you interviewing people?

    22. JOJan Oberhauser1:03:32

        Notice, um, I say I'm, I'm not directly involved at that level, but I think like what's definitely always helpful if you just see people being active in the space, maybe having built something like that already, just, just understanding what it actually means. Again, especially, again, we talk about them having to be technical. Again, like if you can literally very easily see that again by, by s- by, by having split something or just seeing, um, how they kind of present things, how they think through those things, um, it's definitely very helpful. And the second part, sorry, the first thing was how they get noticed. And the second part was-

    23. AGAakash Gupta1:04:05

        And then how do they succeed in the interviews? Like there's so many different types-

    24. JOJan Oberhauser1:04:08

        Mm

    25. AGAakash Gupta1:04:08

        ... of AI PM interviews these days. What are you guys running? How are you guys separating, you know, the pretenders who just use Claude Code to ship some product on GitHub, but they don't actually, they aren't actually technical? How are you differentiating, okay, this is the PM in the interview process that could actually succeed here?

    26. JOJan Oberhauser1:04:26

        We found that they kind of really kind of... I found it was mentioned before that they're really kind of able to kind of go deeper and not just it feels like they're repeating things they ha- heard online without really understanding what it actually means. And I think this happens a lot. Like again, you can by just asking the right questions, um, like we, we are very lucky in, in the sense of that a lot of the, the, the things that actually become problematic, um, at, at a certain point, we actually also experience, um, internally already. So we know what kind of questions we, we can ask and what kind of answers to expect there, which actually kind of a, a person would answer if they actually thought about a problem more deeply versus the person that actually just repeats what, what they heard and, and it seems like it, it sounds good, but actually there is nothing behind it. Um, kind of I think like, but, but even then, like I think with, with, with the AI, I think it, it definitely becomes like we, we'd also kind of have tasks for, for, for each role. I think it be- bec- becomes even more and more important there. Um, it definitely seen in the past, um, that a lot of tasks, if they're especially take-home tasks, it is, yeah, obviously if they never know is, was AI involved or other people involved. Like even pre-AI, we had cases where some other people simply helped out there. So like something like a, a live session where you just go through a problem and kind of try to kind of see like how would you work through the problem together, it actually tells us much more. We also had problems in the past that, like you kind of have to, of- kind of obviously have people that kind of worked in the same space before and people that never heard about a problem. And I think it's very hard to, to understand, uh, sometimes if, if you have some- a person already worked on the problem before, um, if they actually at the right quality, did they just again, are they actually at, at the right level? Um, or did they just again experience it before this, right? They perform much better than people that never thought about it, and this is, it's just a very hard thing. Honestly, very often you, you have to still work with them for, for, um, a month or two to really understand, hey, uh, did we really meet the bar? And I mentioned before, like we, we're having this goal, reach a billion users less than a thousand employees. Like that obviously makes very clear talent density is really important. If you have only a thousand people at, at a certain scale, you can only have the best ones. And that kind of really en- ensuring that, like the quality is as high as possible obviously always means you kind of really have to always kind of, um, um, keep the hiring bar high. But again, if somebody doesn't work out, I think it's the best for both sides if you also kind of sadly let them go, because I think it's obviously not gonna be great for n8n, but it's also g- gonna be bad for, for the people actually to leave them in that role because they're not set up for success. Doesn't mean that they are bad in any way, but they're not the right fit for the role if we need them in, in the organization.

    27. AGAakash Gupta1:07:18

        Wow, what a wide-ranging conversation. We covered n8n's growth, n8n versus Claude Code and Zapier. We showed you guys how to use n8n to create a personal productivity assistant and get other use cases from templates, as well as how they grew. Jan, thank you for all of the alpha you dropped today.

    28. JOJan Oberhauser1:07:38

        Thank you for having me. I really enjoyed the conversation. See you.

    29. AGAakash Gupta1:07:40

        All right. And if people wanna find you online, where should they go?

    30. JOJan Oberhauser1:07:43

        Honestly, I'm only active on LinkedIn. I, I left, um, X, um, and so only on LinkedIn. [laughs]

Episode duration: 1:09:05