ZJ: Flowfuse is the main contributor for Node RED. It's more than a decade old, it was started at IBM. They have decided to make it open source, which kickstarted the community around Node RED. And without it being open source, we would never be at the maturity of adoption we are at now.

Announcer: You're listening to Augmented Ops, where manufacturing meets innovation. We highlight the transformative ideas and technologies shaping the frontlines of operations. Helping you stay ahead of the curve in the rapidly evolving world of industrial tech. This week's episode is hosted by Rui Metres, Head of Ecosystem at Tulip, the frontline operations platform.

Roey: Welcome back to Augmented Ops. Today, we're joined by ZJ Van de Weghel, CEO of FlowFuse. Welcome to the show, ZJ. Yeah, thanks for having me. Excited to be here. Awesome. So, I think we have a fantastic conversation today. So, to kick us off, I think you're coming from a very rich software development background, so it would be cool kind of to understand where you're coming from, what led you to FlowFuse as a background.

And specifically, I think the end gate of open source technologies is kind of interesting for our discussion. 

ZJ: Yeah, sure. I was always that techie guy that built his own computer from parts he could find or get from friends or family. And then after a while I started. to make a decision on where do I want to go with my career.

First, I chose economics and I practice, well, not practice, but I studied law for a year, big, big mistake. And then I finally came to my senses and, uh, started with computer sciences in Utrecht, which for me was. Right where I had to be. It was a great feeling to build software, to learn how to build sustainable software that's maintainable, but also like what's the theory behind computer science and how does these algorithms work during my last six months of computer sciences.

I applied to be an intern at a company called GitLab. GitLab brought software development tooling to the world, basically. Well, at least a quarter of the world and the other three quarters of the world is using GitHub right now. By all measures, this is a very successful period of my career. The company did very well, but also as an intern, I went to junior engineer all the way to senior.

And at the end, I was leading the Git integration team at GitLab. And from there on out, I started to look where else I can bring software, but also open source software to a different industry. And I met up with Nick O'Leary who started the Node RED project. And with him, I'm building the FlowFuse company around Node RED.

Roey: Okay. So let's start with that because many of our listeners obviously know Node RED and I would assume that many of them even have hands on experience with Node RED, For those who doesn't, can you give us kind of an introduction to Node RED? 

ZJ: Yes. So Node RED basically has two great components. One, it's a low code development platform.

And the second one is the connectivity, which means that in essence, you can hook up any virtual data source or physical data source to any other. And in industrial use cases, this is applied for data extraction. So if you're on a PLC, how do you actually extract data from analog sources or from digital ones?

How do you normalize the data and publish this to the rest of your data infrastructure in your company? But then also there's a visualization layer, uh, Node RED dashboards, so people can build dashboards on the data that you just extracted or request the data from digital services or a database to test their hypothesis and go from 

Roey: there.

Okay, so that's kind of a fundamentally technology open source project that is being highly adopted and you're one of the main contributor and following that project for now, what, 11 years, I guess. 

ZJ: Yeah, it's, uh, more than a decade old. It was started at IBM the first year, if I recall correctly, wasn't open source.

And then IBM decided to make it open source, which kickstarted the community around Node RED. And the community is highly important to supply feedback and direction of where Node RED should go. And without it being open source, we would never be at the maturity of adoption we are at now. 

Roey: Do you see the correlation between the fact that it's open source and kind of the highly adoption and success of it?

ZJ: Yes, especially if you are looking to have adoption from bottom up in your organization. Three is so easy to adopt for an engineer. There's no legal, there's no procurement. You can just install the software, get going, show value and solve your problems right away. Instead of, yeah, it can take months to just do the legal parts for some companies.

And connect for us the 

Roey: dots between that open source, you know, IOT, NoCAD connectivity aspects into manufacturing. Do you see Node RED kind of for manufacturing as the main vertical or it's highly deployed in other areas as well? 

ZJ: It's very, very broad. The main and the common denominator for all these use cases is you have many different sources of data.

And all have a different format, or it's unstructured even, and you want to connect them either way. That's where Node. js shines. And manufacturing has this problem as a prime example. So that's where we mostly sell into. As a company, our ideal customer is in manufacturing. The problem is so large for them, and we can add so much value for them.

That's where we start. But we also sell to banks and insurance companies. The news and media, Sky UK is a great customer to work with, and they actually have the same problem. So news comes from many sources in different formats. You have to compile and extract all the, all the data from those sources. And after editing, like you also need to publish this on your websites, but also to aggregators.

So it's the same problem that manufacturing has. So you have the data now you need to normalize it. And generally you also have to push that data elsewhere. Nice. 

Roey: With Tulip, we see quite a lot of manufacturing customers, obviously, using it for connectivity around, around the shop floor. But it will be interesting kind of to get your angle when it comes to FlowFuse.

What brought you to fund FlowFuse and how FlowFuse work with Node RED, I guess. 

ZJ: Yeah, so FlowFuse is the main contributor for Node RED. We don't fork Node RED in any way. So the Node RED that is open source is also the Node RED we use. This is a key decision we made very early on in the life of our company.

We did identify with our customers that there's a lot of questions around scaling Node RED on both an organizational level, but those are on a technical level beyond two or three, like, what do you need to get to a 10? 10 instances of Node RED, what do you need to go to 100? And our biggest customer right now has nearly 2, 000 Node RED instances across 11 production sites.

Just to make 

Roey: sure that our audience understand what you're saying. So when you say you're not forking Node RED, you mean you're using the original open source version and you're not kind of creating a better version for your premium customers, rather using the normal and providing additional features.

ZJ: Yes, we provide a platform around Node RED, the Node RED anyone can download for free is the same software we run in our production systems for our customers and the same software we will install for and with them on their hardware devices, their PLCs or their servers. Yes. And I 

Roey: guess that's connecting the dot to what you started with, it's like the GitLab model on top of Git, basically.

ZJ: Yeah, there's a lot of things that we took from GitLab that just made a lot of sense on how to approach open source as a company around an open source project. So GitLab did the same thing. They, they use Git the same, actually, as GitHub does. There's an open source package. There's some extensions here and there in both GitLab and GitHub where they extended Git, but there's nothing that you couldn't do yourself or is, is not something they wouldn't port back in, uh, in a version or two.

So the core of our company and the core for the community is the same Node RED and that allows us to work with them as well. So give us more details, 

Roey: kind of, we did a side tool kind of on forking, but what are kind of the core value proposition on top of Node RED? 

ZJ: So we have four value pillars. The first one is the management layer.

We identified as first. area of improvement. How do you get up and running really quickly? So the server infrastructure, what do you need? How can we get you within a minute from running to a production grade Node RED? Uh, that means the security set up. Uh, that means the settings are like, it's a template, it's standardized and normalized and accesses, uh, and control is set up.

So by default, if you start a Node RED, there's for example, no HTTPS to the editor. That's obviously something we, uh, we help out with. We set up the authentication offers and the authorization layer. Uh, we have a role based access control system and that hooks in with your authentication and authorization layer you got in your company.

So there's, for example, technologies like LDAP and single sign on. That's just something on the backend we have smoothed out completely. So as a user, as an engineer, you go into. Clove uses a platform and within a minute you're up and running with no dreads and we take care of all the details. And then the second pillar is scaling.

So like I said before, our largest customer has 2000 instances of these. So how on scale, how do you run these? How do you operate these? How do you roll out security patches? And also like, how do you even know what all these things are doing? When? Why? Also the remote access. So if you're at your desk or if you're at one Node RED instance in one site, we provide access to any arbitrary Node RED you want access to so you can learn from the functions and the features you built in another site.

So you can just copy paste and templatize your data extraction methods or your dashboards for that matter. And with team 

Roey: manufacturing the classic, let's say Customer is that a large organization with many assets give us a bit color kind of the way people are using Flowfuse with the manufacturing.

ZJ: Ideally, they have tens of thousands of assets, obviously, because of our licensing model, that's more interesting for us as a company, but then the open source adoption already starts for very small companies as well. We are agnostic in terms of how many, obviously the value for a platform to help you scale is much more profound.

If you have a thousand, then you have 10 because the scaling needs are lower. So it depends on the customer for us. The ideal customer has just a need to extract a lot of data from a lot of machines. But then if you look at manufacturing as a, as an industry. This is generally already the case, like compared to software companies, pure software companies, manufacturing is so much more involved.

You have all these parameters, you have all these sensors already, even medium or small sized factories are on a data level actually much more interesting than most larger websites. Ideally, obviously, the bigger the problem, the cooler to work on from an engineering perspective. I am personally most engaged by Customers that just get started with the data acquisition from all the machines and not the subset that is digitally native.

I feel it's very motivating just to get the data out of a milling machine from the 1950s. It's just so cool to understand what that means for organization. 

Roey: So what you're describing is kind of Flowfuse is bringing fundamental, let's say DevOps principles into a modern digitalization journey for a large, large manufacturer, right?

When it comes to obviously access to 

ZJ: data. Yes. So collaborative engineering. Indeed. So that brings DevOps as well. Uh, snapshots and how do you roll back all those strategies? We, uh, we help our customers with, and also help them mature in their DevOps practices. So even if they start out and they're testing and their production instance is the same, we can add a lot of value and then also make it very seamless to migrate to more mature practices where you're not testing in production.

Roey: Side question. So, you know, when we meet customers and, you know, start the journey with Tulip, many times we see alongside, let's say the technology adoption access, you see the change management access that need to comes with adopting of no code, et cetera. Do you see kind of same trends? Who are the people that use FlowFuse and is that kind of, you know, different mentality that requires?

ZJ: The mentality is also organization provided. So if you're in life sciences, traceability and leaving an audit trail of what you're doing is top of mind for engineers as well. If you go outside of life sciences, generally that is something that is not top of mind and you need to show the value of having a rollback feature.

And the value of hashing a change and them understanding what the version exactly is of what's running now, they need to actually run into a bug and then show them the value of like, Oh, you have a rollback. So you're good. And then within five minutes, you still get your data and the feature you just implemented, you just need to redo.

And from there on out, it's a much easier conversation as well. Like, Oh, you could have prevented this with a DevOps way of working. So you extract all your logic, you test your logic with QA, and then you promote it to production when you know it's all good. But we found that if you go in with a DevOps mind right off the get go, it's a lot of education where at first, generally the Node RED customers are much more looking for the auditability, the collaboration tooling that we provide as well.

And the more mature they get in their adoption, automatically they adopt the DevOps suite. 

Roey: So let's talk about the user for a second. Who is usually the user of Node RED and FlowFuse? 

ZJ: Yeah, so the end user are generally all types of engineers. There's some IT, there's a lot of OT engineers as well. And giving 

Roey: that, so if there are many OT engineering that are using this ecosystem, that mean that it's a citizen developer, uh, model in a good way.

You feel that you're empowering the engineers to use these tools with little to none 

ZJ: background. They empower themselves. I'm not going to take that honor away from them. Yeah. Yeah. And that's one of the advantages. Or the community. Like, yeah, exactly. Like the community, the forum of Node RED, like there's so much knowledge already out there.

I did find that engineers are very pragmatic. They Google stuff, they try a lot of things. So saying that we enabled them as a company, I would feel that's too much honor for them and it's mostly just the engineer and the engineer mindset of, I'm not where I want to be yet. But I will get there and yeah, it's basically like software engineering as well.

Sometimes I feel extremely stupid for two hours and then I fix my bug and I feel like a champion. And that feeling mechanical and electrical engineers also have when they're building software. 

Roey: Just to give some numbers, uh, Chet GPT claimed that, uh, Node RED has more than 200 contributors and I guess more than 10, 000 stars, which is probably a, you know, a community of 100, 000 people.

Yeah, we're talking about that scale. 

ZJ: Yes, I would not be surprised. Just the slack alone has 12, 000 members. Then in the forum, we're literally talking about millions of page views a month. Uh, so it's a huge community where there's a lot of vibrant action going on. And then you didn't even talk about the connection library, which is mostly community maintained with over 5, 000 plugins for.

Very large protocols to very small ones all the way to home automation ones. Yeah, there's a lot of broad adoption and broad plugins available. 

Roey: Let's talk about that double down for a second. So you're saying that besides kind of the core capabilities of Node RED, there is a plugin app store or whatever we want to call it that enable users just to download connectors and to, I don't know, assets providers to publish a connector to the community.

ZJ: Exactly. And there's giants, uh, PLC manufacturers, which publish their own nodes. I think Julia, for example, also published their nodes to integrate the Node RED ecosystem with their own platform. And then there's community contributors. There's, for example, uh, a suite that did the original OPC UA node. And now there are two or three.

One is now maintained by a system integrated in Germany. So in true open source fashion, like this is a vibrant ecosystem. That's, uh, goes where there's a need to build plugins. 

Roey: That's very cool. You mentioned before that within the user based, there are IT engineers, people that, let's say, have some coding skills, but also lots of OT engineers that have little coding skills and are probably more closer to the problems usually.

Do you see Node RED and your walk and flow fuse as a IT OT convergence mechanism? I don't like the IT 

ZJ: OT convergence story because it already puts IT and OT at odds with one another. If you want to combine two forces, you shouldn't tell the story of how different they are. I do see a lot of stakeholders implementing their own logic and then it helping out and explaining like, this is what you could, uh, could have done better on an IT skill, but then a stakeholder that knows the ground reality of where the data's coming from.

What are the gorge of these datas? D we for example, filter the resolution of this data on the edge because. The sensor produces a 120 Hertz, that's way too much for what we need. So I did an average, for example, of those values and publish it every one second. So yes, there is a lot of collaboration between IT and OT, and I do think like longer term, you'll have more.

Hybrid teams. So you'll have it embedded into OT engineering teams and, and vice versa. To the extent where I suspect that in a decade from now, it will be a very weird storyline to talk about it and OT convergence. And it feels like there wasn't a split to begin with a bit of like DevOps and how that went in the last 10 years.

So first coined about a decade ago, actually, and the first two years, I wouldn't say there was a lot of discussion around devops and how different developer and operations were. And like a light switch out of the blue for some observers that switched around and developers and operations on the digital side.

Went with DevOps, uh, I feel like IT and OT conversions, I guess it will be, uh, hopefully debated in the next two years. And then when you just have a more broadly skilled team, it will be so obvious that this, this was the way to go. You'll never discuss it anymore because it was the obvious way to go. 

Roey: I'm in favor of that messaging.

So take us to the future for a second. Where is the flow fused in five years? 

ZJ: Yeah. So right now we're very good at data extraction, normalization, and visualization on dashboards. We're getting better and better at closing the feedback loop. So the dashboards become interactive and have an effect on the production line or the scheduling or other influences of the production systems.

From there on out, I really want to help customers with more prediction and analysis, quicker usefulness of the data. It is a very fragmented space right now, and there should be a transformation layer based on Node RED where predictive analysis is just much easier. And so data modeling is very important to get that done, uh, schemas and then the predictive algorithms making those available in a low code fashion is where we can add a lot of value to production systems.

Roey: Okay, good luck in that vision. I think you have a lot to contribute to so many industries. So yeah, thanks a lot to see the progress. I guess to finish the discussion. A question that is kind of outside of, uh, let's say manufacturer cooperation. SaaS company that is built on top of open source. There are, I guess, many companies that follow that model, but it's not super common model.

It comes with some risks and challenges. Can you kind of give us some lights on how that model looks like, the value to the customers, and also this interesting relationship that you have with The community of open source developers. 

ZJ: So FlowFuse follows an open core model. So we built around Node RED and the core remains Node RED.

The core is open source and free of use, and everyone can download the source, read the source, contribute to it. FlowFuse itself. Has partial open source as well. So if you get started up to five, no dread instances, so you get started with no dread in your factory, it's completely free to use. And then as you scale and you get more serious about your no dread adoption, that's also generally when the buyer changes in a company.

So an engineer might adopt FlowFuse, install it on a server, start with one, two, three, four, five, no breads. And then. Their manager starts to get interested and asks like, okay, so now what's, what's the approach here? Can we professionalize? Do we need support? So you see a lot of bottom up adoption. We see both, but in terms of where the core, which is open and free ends and the licensing starts, that's what we use as a differentiating factor.

So it's buyer based as soon as the buyer in the hierarchy of an organization and the requirements gets tougher to even deliver on as an open source projects, for example, support, for example, certain liability. Those things you need a license for, and that's where we also have boundaries around what's open source versus what's no longer open source.

So as soon as the buyer shifts, the model shifts for us as a company. 

Roey: Super interesting model. You, you see this model kind of being adopted more and more in manufacturing. 

ZJ: Yeah, we're not the first one. I completely copied this from GitLab and our earliest invest in the company, uh, is a open core ventures and this model is an buyer based open core.

So there's a lot of thinking. I just stole from people much smarter than I am. It's not unique. Like if you look at, uh, for example, Redis, if you look at confluence. Uh, Redis with Redis Labs, Confluent and, uh, Kafka, then you have MongoDB and now you have Node RED and FlowFuse. You have Git and GitLab. All those companies have an open core and then there's license sets of features around that core to extend the core, to improve on the core and to make it adoptable for larger organizations.

Roey: But correct me if I'm wrong, most of the names that you mentioned are hardcore developer tools. While you're trying to bring something that is more, let's say, easy to use, a bit more citizen developer to the kind of operational team as well. which I think is interesting because I totally understand how that model works for software developers that kind of love these models, basically.

But it's interesting to see how that kind of being adopted in the B2B, enterprise, SaaS, let's say even traditional way of consuming software. 

ZJ: That's exactly right. Uh, so we are probably not the first, but it's sometimes feel we're the first that takes open source and applies it to manufacturing specifically with their own buying models.

The long procurement, they're very risk averse. Mm-hmm . So in some sense, yes we are trailblazers, but then as we just discussed around it, LOT convergence, there's more and more IT in manufacturing and they bring the IT way of thinking about software infrastructure with them. If you look at the data layer and how they built their data lakes.

Yes, there's proprietary software, but there's also a lot of post grass. There's a lot of timescale DB influx DB where it is open source manufacturing focused because they're event driven, for example, and then building on top of that with a super set of license features. 

Roey: You mentioned security, I think, a few times, uh, of reliability as well, I guess.

Do you see these kind of core fundamental reasons when people move kind of from the open source into the FlowFuse package? 

ZJ: Yes, it's highly important for strategic assets, also for mid markets and manufacturing. If you just start out as a manufacturer, it's less important, it seems, and it's mostly engineering driven.

But as soon as you have a mature organization with a security department, there's a lot of checks and balances before you bring software in production. FlowFuse will elevate Node RED for an enterprise. 

Roey: Okay, so yeah, thank you for uh, joining us DJ. Yeah, it was a really interesting discussion and looking into the power of open source.

If I'm not mistaken, I think, you know, it's the first open source discussion we had in the podcast. 

ZJ: Oh, if you want to discuss open source, uh, some more, you know how to reach me. I, I love. talking about it and I've practiced it for the last decades. And I do think my whole career will be focused around bringing open source to more people.

So, uh, always happy to push that messaging in the world. All right. Chills to that. Thank you very much. 

Announcer: Thank you for listening to the Augmented Ops Podcast from Tulip Interfaces. We hope you found this week's episode informative and inspiring. You can find the show on LinkedIn and YouTube or at tulip. co slash podcast.

If you enjoyed this episode, please leave us a rating or review on iTunes or wherever you listen to your podcasts until next time.