n8n on Building the Universal AI Automation Layer

What happened?

Jan Oberhauser: First, thanks for welcoming me here. Great to be here. To answer your question, what happened? I think it comes down to things that we actually did quite a few years ago—actually almost two by now. And I think there’s been two things. One was obviously our focus on AI. It’s like when this whole AI wave started, we kind of—honestly we were worried a little bit at the beginning, we were not sure what does it really mean for us. And we knew it’s probably going to be one of two things. Either it’s going to be a huge opportunity or actually the demise of the company. You obviously want to be very sure it’s going to be one and not the other.

And then we kind of looked in the market. It’s like what are other companies doing? And what most companies did is they kind of added nice AI features to the application, and we kind of realized that it’s probably not the kind of thing that makes sure we’re going to stay around long term. What is really important is to kind of become actually part of the value chain.

So what we did is exactly that. We not just added AI features, we actually allowed people to build AI-powered applications with n8n. There’s obviously nothing that the user base or people kind of realize overnight what n8n has become back then. But it obviously takes some time to catch on.

And that’s where the second piece comes in. Historically, we kind of focused on mainly two things internally at n8n, especially kind of in the marketing department. They always had this more focus on obviously kind of creating leads, but also kind of making sure with the inbound we get more adoption. Which kind of always caused a problem in the sense of we’re always very good in adoption, but at the same time leads normally fall short.

So what happened is that the marketing department then put everything from adoption in leads just to make sure we make that goal as well. And that was obviously problematic because all the long-term planning actually went away. They couldn’t really have the impact they could have had. And we realized, hey, we have this amazing community, we have this amazing bottom-up adoption, so instead of actually kind of forcing something that’s not really true to us, we just really focus on what was already working with this really focusing on this kind of top of the funnel.

That’s what we did. We kind of removed the lead goal. We actually exchanged it with an adoption of large organizations, went all in in the community, making sure as many people as possible are using n8n. And they kind of—also it’s another thing that you can decide and you make the change, but nothing happens overnight. You actually have to fight. You have to have a lot of trust in that, and you have to kind of really double down and kind of wait because these things kind of take some time to mature.

And that’s actually the thing that actually happened. Like, it matured to kind of really double down. We kind of empowered our community more. We created more events, we created more content. And that is in the end this thing that materialized, especially in the beginning and at the end of last year in December, is where the market finally realized that we have become this AI tool, but also kind of the community, especially around on YouTube, where more people created more and more content. And obviously, the more content people create, the more other people want to create content, the better it gets ranked. And then everything kind of started to explode.

George Robson: I heard an amazing story at our board meeting a couple of weeks ago, Jan, that you have “Jan focus time” on long flights, and you actually coded the first version of that AI nodes product flying back from San Francisco. Is that true?

Jan Oberhauser: No, actually I created quite a few things on planes in the past. That thing I didn’t create on planes or trains. I think I lost that time. Honestly, right now I’m—I always love that, to be honest. But now I always feel like destroying more than actually being helpful since I kind of create more MVPs or something like that. I play around rather than actually adding production code on planes or trains anymore, which is sad.

George Robson: On dodgy WiFi. That makes sense. Maybe, Jan, if you can take us back to 2019. I think it was just you sort of founder/entrepreneur when you went out and raised your pre-seed round. What was kind of the aspects, the origin story that really stands out to you? What has really stayed the same from day one of n8n? What has really changed over that time frame?

Jan Oberhauser: I think what always stayed true is for sure that kind of focus on the community. Like, actually the first people, like, in the first week in April, a few people started, and one of the first people that started was a dev rel. Because I always knew, like, if you want to make sure that n8n becomes meaningful and kind of becomes the company like I wanted to build, I really have to kind of make sure I invest in the community very early on. I think that kind of always stayed true.

I think another thing is probably something around values. We always—especially I always try to be as honest and upfront about things as possible. You also see that for example in our license. You probably have seen, like, we never call ourselves open source because we don’t have an OSI-approved open source license. What does it mean? It means our source code is available, everybody can use it totally for free. They can even use it in production, no matter if it’s somebody at home privately or somebody in a large organization. Everybody can use it.

What is, however, different in our license is that people cannot commercialize our code. It means nobody can just take our code and offer, like, a hosted version of n8n, for example, or kind of create XYZ automator in the product.

Why is it important? Because I saw a lot of organizations in the past, like open-source organizations kind of changing license, and people obviously got very angry and they really hated it. And what I thought was quite interesting is people didn’t hate it because of the license that the companies chose. People were mainly angry because people—like this company changed the rules. And that’s never nice. And so what I thought, hey, I think it’s a very reasonable thing. So I’m just very honest and upfront about it from the very beginning. I said, “Hey, I’m not building n8n and giving it away for free because I’m a good person. Like, I actually want to build a business around it. I want to make sure I can get paid. I want to make sure all the other people can get paid as well, because I actually think it’s in the interest of everybody.” So that is why I chose a SUL. We always have stayed very true around that.

Again, now getting to the second part of your question is like what changed? The thing is definitely around, like, what we started at is like I said, as I mentioned before, like, we started as this kind of automation tool. Honestly, I never would have thought that we would get AI ever. It’s like it never even crossed my mind that we would go into the AI space. But I think in the end it’s like sometimes you see opportunities out there, and you just have to take them. And also, I think it makes a difference between building a big, sustainable company that really matters, and probably building a kind of a startup at some point that just literally goes out of business, sadly.

Pat Grady: Jan, you’ve done a masterful job on the community-building aspect of n8n, and you mentioned the values, which I’m sure were really critical to earning the trust of your community. Can you also talk a bit about the tactics? It’s kind of a constant push and pull of trying to kind of shape or curate the community while also really trying to listen and—you know, and see where they want to go. Tactically, how have you managed the community over the years?

Jan Oberhauser: I think from the very beginning it was very important for me just to kind of include people in the sense of, for example, in the beginning we didn’t have any telemetry data at all, that we didn’t collect any telemetry. Which was obviously fine at the very beginning, but obviously as a company grows you kind of literally need data to kind of improve the product very fast. Again, this kind of makes a difference between a company that survives or one that doesn’t. So we had to add telemetry data.

Another thing we had at the edge was at some point we obviously added paid features. And every time we made a bigger change, we kind of posted it to the community forum first. We kind of—like, we shared it with them what we wanted to do, gave them the reasoning why we wanted to do it, and then kind of listened to their feedback. I think that is just kind of taking them with you when you’re doing something. I think that’s a very important thing.

Another thing is definitely kind of trying to kind of empower them. Actually, like, one of the earliest n8n employees was Ricardo. Ricardo, he is this amazing guy working in Florida, and he was one of the first contributors to n8n. He discovered our product on Product Hunt, and he started to kind of create one note, then he created two notes, created three notes. And every time he contributed something, I said, “Hey, that’s great. Thank you very much.” And then I kind of gave feedback, I improved things. And then he learned and he was excited and created more notes. I think at some point he created, like, 50 or 60 notes, like, integrations for n8n.

Then obviously, as soon as I had the first money in the bank, I hired him. I think it’s really just very important to kind of show people that you care and take them with you, and kind of show how much you value them. I think—I really hope people, the community really knows how much you value them. And we try to give back a lot. I think that’s generally—I think this kind of—even though we’re not open source, we have this open source ethos where you kind of really give a lot first. And I think you normally get so much more back than you actually receive. We see the same thing again also with our community today where we just say, like, we empower you. And then they create all of this amazing content, no matter if it’s on YouTube, on LinkedIn. They create tutorials. I think that worked out very well, and that’s something we’re still doing to this day.

George Robson: Jan, has there ever been a shift in—I mean, given the scale of the community today, right? It’s hundreds of thousands of members in size. Has there been a change in how you try and surface the right information or the right ideas, and how you choose which to prioritize coming from that community base?

Jan Oberhauser: Yeah, in the beginning it was definitely easier. Like, which in the beginning, like, I didn’t really worry about monetization at all. So I just wanted to make sure I built the best product, and I just built things that people really wanted. So I could—literally from the very beginning, we had a community forum where people could add feature requests and people could upvote the ones they wanted to have the most. And that is literally kind of how we chose very early on what to build. Unless this was something totally out of whack, we normally build it because we knew people really valued it very much.

And that worked actually for quite a while. At some point you have to be a bit more opinionated, where you’re not just kind of building what the community wants, but we actually have to kind of think about where the market is going, also how you kind of build something again that’s sustainable, that again, especially also something that also kind of large organizations use as well, and also just certainly where you see the market going. And I think that’s what we’re doing right now. We’re trying to find a good mix between what are people really asking for and what do we think we have to build to be successful in the long term, both because of something like AI, but also again, what, for example, enterprise organizations need.

The make-or-break moment

Pat Grady: You mentioned earlier, toward the start of the conversation, that you had this sort of make-or-break moment in the early days of AI, where you realized that AI was either going to be the future of n8n, or I think the word you used was “the demise,” the demise of n8n. Can you take us back to that moment? Like, what did you see that made you realize how important an inflection point in the trajectory of the company this is going to be? And then what did you do to figure out how to position n8n so well with respect to these AI tailwinds?

Jan Oberhauser: What did I see? An interesting take. One thing I actually saw is the funding announcement of Pinecone, where—so I think they raised, like, a $100-million round or something like that. I also saw that they kind of raised around a year before, at some point before, and I was wondering, like, what changed? It’s like why are companies suddenly caring so much about them?

And I realized again, it was exactly that. It’s like, what they were in the past is they were the vector database. What they’ve become, they became a deep database for AI. That’s where I say we have to do something very similar within n8n as well. And that’s exactly what we did. What we also did on this in the past and what most of our competitors did is they kind of created an OpenAI node, a node that kind of connected via HTTP, get requests to OpenAI. And you could do quite some nice things with it. And you can still do it up to this point in time.

But what we have realized is that this is honestly not enough. Like, you can do quite nice use cases, but the really powerful things are not possible. You really have to be able to kind of create proper agents, like change auto prompts, add tools, add a vector database and output parsers and all of these things. And that is when we then kind of built out our advanced AI functionality, where we allowed people to do the things in n8n in this kind of low code, no code way that was previously only possible by writing Python scripts. So we kind of drastically reduced the entry barrier for people to build things, and also reduced the speed because the stuff has historically always been quite finicky, honestly. Like, connect A with B, and it didn’t work, and it’s taking away all of those pain points and it takes care of it underneath the hood, and you just click a few times, add an agent, add a model, add the memory and so on, and it’s working magically.

How is open source changing?

George Robson: Jan, talking about companies that share, I guess an open-source ethos across AI, obviously the role of open source I think is changing across the industry at large. We’ve seen obviously a lot of headlines coming out of Meta, out of Mistral, OpenAI released their GPT, open source version, et cetera. What are just some perspectives you can share on how you perceive the state of open source in AI more generally, and maybe how do you see its role changing in the future?

Jan Oberhauser: I think the great thing we always had with open source is it’s like it’s kind of—it’s driving very rapid innovation in the end, because you have this kind of huge army of people that really, really care very deeply, they’re very smart, and very often have a lot of time. And they kind of do things that organizations can very often not do because again, they have very different incentives, or have very different timelines or have…

And then I think the nice thing about open source is like, you can do whatever you want, and you can explore things that maybe wouldn’t even make sense in a company setting.

So I actually think, like, open source is, like, really this amazing thing that kind of opens up the world to very many different possibilities. In the end, like, some of them wins, and a lot of open source projects never make it to anything. Obviously, that’s the ones nobody talks about, but some of them, they kind of hit something, and those are the kind of the ones, and kind of really shape kind of way often the whole industry in the end.

George Robson: Maybe a follow up question to that, Jan. I mean, do you see a shift in the role of open source? I mean, do you see that companies, I mean, today might be using some of these open-source technologies to save money, to save costs, but maybe actually having more control inside of organizations over the performance of models might be a value proposition that evolves in the future?

Jan Oberhauser: Definitely very, very much. I think, like, first, I’m not sure if open source is very often it’s the cheapest thing. I think this is a kind of misconception; it’s actually very often not true. I think you get a lot of things like being able to kind of self host, knowing where data is stored and what’s happening.

I think there’s a lot of value there, but very often it’s actually not the cheapest thing. Very often it’s actually the more expensive thing, because you have hardware that’s probably idle for 99 percent of the time, and this actually kind of turns out to be more expensive. So actually, I’m not sure if that’s actually true. But the second part I think is true, and I think people care about it more and more. Like, we actually work now with a few organizations, and that’s actually the main reason why they wanted to use open source. It’s not because it’s free, it’s because they actually care about the data privacy and data security angle. And I think that’s also what we always historically have seen within it. And very often it’s like people want to kind of self-host and want to know where the data is stored, and obviously it doesn’t mean they want it on the computer underneath the desk. They still run it on the cloud, but obviously in their own private cloud.

George Robson: Jan, one, I think, noticeable shift over the last couple of years, something that, of course, n8n, I’m sure has been a beneficiary of, is just improvements in the communication protocols between different systems and different models. And obviously MCP is a high-profile candidate that has really driven the industry forwards. Can you talk a little bit about that, kind of how you see the changing role of some of these protocols in the future and where you think they might evolve?

Jan Oberhauser: I think that if you kind of standardize, in the end you kind of accelerate things. I think that’s very important. Even interestingly if the standard is maybe not perfect, I think it still adds so much value. And I think if MCP is going to be the one we’re going to use in a few years—I have no idea, but I think it’s great to have a starting point that you kind of can build on top of. In the end this MCP is like the HTTP of AI workflows, and I think that’s really amazing. I think it kind of will enable—it already enables things like agent-to-agent and it’s a kind of the building blocks you actually need for this kind of more powerful use cases like for example, you had marketplaces or like plug-and-play automations. And I think n8n can act as the kind of orchestration layer between this kind of service, MCP agents and tools. And also obviously n8n is also the best kind of platform to build also tools that you can access with MCP as well.

What is n8n?

Pat Grady: Actually on that Jan, we haven’t really done the straightforward what is n8n? So for somebody listening who has a passing knowledge of n8n but wasn’t sure exactly what it is, what is n8n, and when should people think about it when building in AI?

Jan Oberhauser: I think by now it’s probably the easiest to use, most powerful way to build AI agents right now. I think it allows you to kind of again, build things you previously never even thought you were able—you could build. I think the nice thing about it is you can go very fast from a first idea you have to a first prototype, and then bring that prototype into production.

Pat Grady: What is the stack that you see people using within n8n?

Jan Oberhauser: It’s actually quite diverse. I think it actually probably does make sense, because the nice thing about a system like n8n is you can just connect everything to anything. It’s like there is probably not kind of this default stack people are using. Like, they use literally any LLM, they use any kind of memory or vector store or any kind of applications, et cetera. It’s just the thing that makes anything great, because honestly, like, I don’t know what LLM is going to win the race in the end, who’s going to be the best one, if it’s going to be one or if it’s going to be a million different ones, it’s going to be a small one, a big one. I have no idea. But that’s the great thing about n8n is like, we don’t have to care because the nice thing is you can use whatever is best for your use case, and I think that is what makes it so powerful.

Pat Grady: Any interesting observations on what’s been trending, positive or negative in the n8n universe?

Jan Oberhauser: What people definitely use quite a lot is honestly still like all the kind of Google tools. Google is definitely still, I think, the one kind of application you can use. People use it privately in a large organization, and I think that’s an amazing thing. That’s why we definitely see this as probably one of the strongest ones. We definitely also see a lot of these kind of communication platforms, if it’s a Slack or Telegram or anything like that. It also kind of makes sense especially in this kind of AI world, because you obviously need an interface, how you kind of interact with the agents you build, and I think that is probably a few to call out. And apart from that obviously, also things like just databases, any kind of—and again, the nice thing about n8n, you can connect literally anything. So it’s not just kind of applications, it’s also kind of low level things as well. And also honestly, very often also internal tools as well.

The changing user base

Pat Grady: One of the nice things about n8n is you struck a good balance between the sort of flexibility, customization, control that a developer might want with the ease of use that somebody who’s less technical might want. Has your user base changed kind of the complexion of the n8n user base? Has it changed as you’ve gotten more and more AI adoption?

Jan Oberhauser: It’s quite interesting, like, when actually our user base started to explode, like, since the last eight months, you are really wondering if those people are really going to be the most successful, like if the quality actually dropped. And quality doesn’t mean obviously it’s bad people, it’s just like the kind of people that are successful within n8n. And it didn’t, which was quite surprising. Really, people that are either already quite technical or people that really have a use case and they care very deeply about what they’re building and they’re just willing to put in the work. Like, maybe they’re not as technical when they kind of enter, but they have a problem they really want to solve.

And we have actually quite a few users that actually started to learn to code because of n8n. And why do they do that? Because they can realize, “Hey, I can do that much, like if I’m technical, but I cannot code.” But again, this can literally go anywhere if I can code. It’s kind of this whole new world that opens up that is actually—it didn’t change that much, and I think that’s really amazing. There’s so many technical people out there and people are really interested. I think our community, it’s really amazing. It’s just the ideas they’re having, how driven they are, and how much they also want to help each other out. I think that’s just great.

How AI has changed things

George Robson: Jan, do you think with the proliferation of adoption of many of these different AI technologies, the pressures are different on being a founder in 2025? I mean, you’ve lived it over the last eight months.

Jan Oberhauser: Yeah, I think that it definitely changed things. I think the main thing is like, the past, it was easier to plan and know where you’re going. And right now you’re always living in this constant—I wouldn’t call it fear, but kind of constant state of uncertainty where the world is moving, and you have to be really willing to kind of—like, you have to strike a very good balance between not missing anything important, but at the same time not jump on anything that’s coming up your way. And I think that is definitely much harder, it feels like, than it was in the past.

George Robson: Given that, how do you think the signals that you get as a CEO have changed in terms of how you weight them? I mean, even things like, you know, the usage you see across the business or the revenue being generated, you know, in different parts of the company, how do you sort of assess the sort of durability of that and make sure you’re investing behind the right things?

Jan Oberhauser: And this actually probably talks about something I think is generally quite important. It’s definitely harder in kind of this open source world we are having, as obviously we have some use cases that generate revenues like with enterprises, and we have a lot of people that use the product for free and we never generate revenues with them. But honestly, none is more important than the other one because one is obviously going to help us directly, but the other one is the thing that drives all of it. So it means we have to strike a very good balance between both of them.

And honestly, normally I’m leaning much more heavily into kind of—for example, I guess you obviously asked about signals, but I think it’s an important point is like, I normally lean much more heavily in kind of giving away more for free because again, everything we give away for free kind of makes the whole product better for literally everybody, and drives more adoption and also drives more revenues in the long term, versus everything we built for the enterprise organizations obviously just kind of helps a subset of the users. I think it’s still important to do both and I think you have to listen to both very closely, but also I think it’s around timing in the end. It’s like right now I think it’s really about capturing the market, and we have to listen to what do people really want to build? And I think what is important is to really capture the market generally, literally capture the smallest builder out there. Because what we already see is that through the smallest builders, we get into the largest organizations out there. What I think is impossible is if you now focus very heavily on enterprise only and I think you can never go downwards again. I think nobody went from enterprise to kind of owning a space. And I think that is—again, also Google started the same way as well. They maybe own quite a lot. The same with Microsoft as well. But you have to start in the right direction there.

Pat Grady: You know, right now we’re seeing this explosion of people building with AI capabilities. Over the last few years we’ve seen an explosion in the AI capabilities themselves, and so the question is, is that technological foundation, the AI capabilities themselves, do you think that that is still innovating at the same rate it has over the last few years, or are we finally asymptoting in terms of the capabilities coming out of foundation models or out of the open source world? Do you have a point of view on that?

Jan Oberhauser: I think probably looking at GPT-5, I think it definitely feels like it’s slowing down, which kind of makes sense. I think it’s not very often what happens there is more low-hanging fruits in the beginning than the later point in time, and you can throw more compute power and more data for quite a while, but at some point you kind of max out there and at some point it just becomes too much. So I think definitely the pace is kind of slowing down. However, I just think that’s more or less a kind of a temporary thing, because there’s so much money in the market right now where people are exploring a lot of different things. And some of them are still very early and I think they’re not at the right stage yet, but I think with some of them we’re coming to the right stage, I think we’re going to see probably another acceleration.

George Robson: Jan, maybe building on Pat’s question, we’re seeing more code being written by machines, right? And more end-to-end, agent-to-agent automation. How does that change the positioning of n8n? How do you think about areas you’ll invest, maybe given that future?

Jan Oberhauser: I think one thing is, I think it probably talks a little bit to the kind of role as a developer. Like, in the past, a developer was somebody that kind of built something for you. It’s like they said, “Hey I need X,” and they built X for you. Now with tools like n8n and all of the vibe-coding tools, that’s changing. You see developers more as the people that kind of generate, kind of create the guardrails for you, the people that empower you to build the things you actually want and need. And I think that is really amazing, and honestly it also says exactly what n8n was always about. It was always about empowering people. And I think it’s great. I think literally every developer is kind of going into the role of empowering other people to build the things they actually want and need. Yes, developers obviously still have the vibe coders and other people like that. They’re not going to build everything. There’s still going to be a role for the classical engineer. But I really love that kind of world where people are empowered, and the people that have the problems are the best equipped to kind of solve them themselves. And I think that’s kind of a road where we’re going. And I think that the engineers and developers are definitely a very important part for that.

The need for orchestration

George Robson: I mean, Jan, something I think we’d love to know your opinion on. I’m sure a lot of people listening are struggling between building, you know, verticalized applications that solve a very acute use case versus kind of having broader platform visions for their companies. n8n is the horizontal of horizontal tools in many ways. How do you see the positioning changing given some of those dynamics of some of these very vertical applications, and then some of the more horizontal tools coming out of the labs and the like?

Jan Oberhauser: Yeah, I think, like, vertical tools are obviously amazing. If you had, like, one very specific use case that does this one thing perfectly and it can do it much better than every horizontal tool. Obviously, that’s exactly what happened with SaaS.

George Robson: Hmm.

Jan Oberhauser: Like, in the SaaS world, they have a vertical application for literally everything. That is again why we had the need for something like n8n, because you had to kind of bring all of them together again. I think there’s also a limit, but it also probably happens more and more in AI right now where people build these kind of very vertical tools as well. But as this is happening, the complexity increases again, and you either need kind of another orchestration tool or you need a horizontal tool that kind of builds everything for you again. So either kind of, again, you bring all these tools together or use a more horizontal tool to kind of build this thing in this one tool rather than kind of a million different ones. Again, I think that’s again why we obviously feel like we’re in a very good position there because again, I think no matter which way it’s going to turn out, I think we are very well positioned for this world.

The Excel for AI

Pat Grady: If everything goes right, what will n8n be in five or ten years? What role will n8n play in the world?

Jan Oberhauser: The idea is always like—for very many different reasons, I always compared n8n to Excel. I always thought about, hey, if people 15 years ago they heard “spreadsheet,” they thought about Excel. And if people in a few years think about, “Hey, I have to build AI, have to do anything with AI,” the only thing that should come to mind is n8n. So I see it more as kind of this default orchestration layer, like what is kind of the platform from everything to building to deploying and where you find your agents, anything like that. I think that is where I think n8n is going to naturally evolve, and I think we are very well positioned as we are already kind of starting to be like the default building tool already.

George Robson: That’s a future we’re very excited to live in. Thank you for sharing that. So Jan, you build a remote-first company, obviously centered in Europe but with global ambitions. Just talk a little bit about thinking about going into the U.S. and kind of building the team and scaling the organization over the last year.

Jan Oberhauser: Sure. Obviously we started in Europe, and the whole team was based here, which always was quite interesting. Even though we were based in Europe, like, our user base was always very global, and even for quite a long time Europe and the U.S. had the same size even though we honestly didn’t provide our U.S. customers the best experience. That definitely shows this very, very big need in the U.S.. That’s why we also right now are expanding to the U.S., and we are actually just right now opening our office in New York. And this is also why we’re hiring quite a lot. It’s like, especially obviously in the U.S. but literally like also I guess we could say worldwide, literally everything from engineers to people in support and especially a lot of people in the go-to-market org. I think it’s also the kind of the thing that gets me excited is like I think not many European organizations kind of have the possibility to really build something really global, something that really matters. I think n8n is probably one of the orgs I think that has quite a lot of potential, and it’s nice to see that we are now finally kind of going that direction and kind of also kind of capture the U.S. and take that market over as well.

Lightning round

George Robson: And maybe to wrap we love to ask some kind of rapid-fire questions just to get your take on a couple of key themes. I think one, just from your reflections of the last six years, what’s maybe one of the hardest truths you’ve learned about Jan as a founder/CEO?

Jan Oberhauser: I guess one thing is probably I really don’t like to say no. [laughs] There’s so many different things out there, and I think very often it turned out to be okay, like, to kind of do a lot of stuff in parallel. Sometimes in hindsight this actually didn’t turn out that well. Interestingly, I think that most things very luckily turned out to be quite well, so I’m very lucky there. But I think it also could have turned out very differently, and I think for most companies it actually turns out very differently.

George Robson: We want you to keep taking those big swings.

Jan Oberhauser: Yeah.

Pat Grady: What is one must-read or must-watch piece of content in AI—blog, book, show?

Jan Oberhauser: Like, I still love Her. It’s just a movie. I think it’s just—yeah, I think it’s just this thing that is especially amazing about it is like a few years ago it was sci-fi, and it’s now suddenly this thing that is just around the corner. And that’s why I just love it. I think it’s also a great movie.

George Robson: Is there a tool or a product, Jan, you’ve been playing with in AI that you would recommend or that really interested you?

Jan Oberhauser: I think I just very recently and probably very late I came there, but I started to use Granola. I think it’s such an amazing tool. It’s so great, so simple to use, does such an amazing job. So I think that would probably be the one that comes to mind right now.

Pat Grady: What AI application or application category do you think is most likely to break out in the next six to 12 months?

Jan Oberhauser: I think it’s probably more this kind of AI-powered internal tooling. I think there’s obviously like this external part that I think is a little bit hard. You have to be very careful there. But internally you can take much more risks. We also see that already at n8n quite a lot. That’s obviously why I’m also excited about it.

George Robson: Yeah. Needless to say, we appreciate you sharing some of your story, looking into your crystal ball for us, and kind of sharing some of the perspectives over the last six years. It’s been a privilege for Sequoia to be a part of it, so thank you for having us on the journey. We look forward to the future.

Jan Oberhauser: Thank you very much. Thanks for having me.

Mentioned in this episode

Mentioned in this episode:

• Model Context Protocol (MCP): Open protocol that lets AI models safely use external tools and data that is used extensively by n8n for orchestration.
• Vector database: A database optimized for storing and searching embeddings, now widely used for RAG systems in AI.
• Granola: AI productivity tool mentioned by Jan as a recent favorite. 
• Her: A film that Jan says, “a few years ago, it was sci fi, and it’s now suddenly this thing that is just around the corner.”