Connecting Enterprise Systems for Agentic Work With Flowgear

[00:00:00] Scale your business with agentic AI with limited risk. With Denodo's AI data layer, your agents are provided with real-time company data and guardrails for company protection. Create the business you always dreamed of with Denodo. And you can do that by simply visiting denodo.com to learn more. But now, back to my guest.

[00:00:27] Every company has systems that refuse to speak to one another. Then we've got that small problem of data silos. A quote is approved in the CRM, somebody needs an invoice in an accounting platform, and suddenly a simple business process has become a relay race involving spreadsheets, copy and paste, and possibly mild despair. Well, today I'm joined by Daniel Chilcott. He's the CEO and co-founder of Flowgear.

[00:00:56] Which is an integration platform as a service provider that was funded in South Africa and now serving customers globally. And Daniel is going to explain today how AI is changing enterprise integration. And instead of adapting a rigid template or building every connection by hand, users can now describe the outcome that they need, and then allow an agent to create a test, to create, test, and refine that workflow.

[00:01:25] And Flowgear's builder MCP can bring the capability into an AI chat or development environment. So I want to learn more about this, discuss why complicated business rules still require experienced people, and what AI agents need from ERP and CRM systems, and how automation can give employees additional time for valuable customer conversations. Well, enough from me. Let me introduce you to Daniel right now.

[00:01:54] Thank you for joining me on the podcast today. Can you tell everyone listening a little about who you are and what you do? Yeah, thanks for having me, Neil. I'm Daniel Chilcott. I'm managing director and co-founder over at Flowgear. We are essentially Africa's first iPaaS platform, a bit of a mouthful integration platform as a service. And what we do is help companies to automate integration of their data across their different apps and services so that they can streamline things operationally.

[00:02:23] Well, thank you for joining me today. One of the things I always try and do with my guests is find a little bit more about their origin story and how they ended up on this path they're on. So for yourself there, tell me a bit more about your background and how you got where you are today. Yeah, so I grew up coding pretty young. My father actually gave me a ZX Spectrum, which you may be familiar with. It was a microcomputer, I guess, from the 80s. So that's what I grew up on, copying program listings.

[00:02:52] But my first real job was as software developer number one in a very small business. And originally it was just a fight to survive. So we'd take anything that we were asked to do, so custom software. Over time that evolved into a CRM. And it did kind of become a situation where it didn't matter what you wanted. The answer was always CRM plus and then we'd build a module. But it worked pretty well. Our customers were happy.

[00:03:19] But what always happened is they needed our product to integrate into the other systems that we're using. Their other business systems, their ERP and so on. And so we were building the integrations for that by hand. For example, you have a quote and that gets approved. Well, that needs to now become an invoice in your accounting system. And so we were doing all that by hand. And that's kind of what spawned the idea for there's got to be a better way to do this.

[00:03:43] I left that business in 2007, went solo and built a small on-premises integration platform. But that was really the predecessor to Flowgear. The idea with Flowgear, which we started in 2010, was to take a capability like that, but build it as a cloud service. And back then, actually, we spent a lot of time trying to justify why you would even want this as a cloud service. Because at the time, a lot of folks were still running a lot of things on-premises.

[00:04:11] And we sort of had to say, I promise you, you're not going to be doing this a few years from now. This is the way you need to do it. So delivering it as a cloud subscription is very useful. It takes out of a lot of the hassle of a customer's hands when you're managing it as a complete solution on their behalf. Wow, what an incredible story. I love how you've gone from planting those seeds to building this. And fast forward to present day, Flowgear now serves companies globally.

[00:04:37] And I believe you've also got plans for growth in the U.S. market as well. Tell me about that. Yeah, we started to pick up customers pretty much organically in the late 2010s. We hadn't really focused on it prior to that. Today, the U.S. accounts for about half of our customer base. And we're really focused on, I think, product-led growth.

[00:04:59] And AI is the huge kind of paradigm shift for us that's enabled us to take that to the next level. What I mean by that is that to resonate with someone who might be interested in your product, you've got to show them something that's really specific. And this is a challenge that a platform play has versus a product play. A product play does what it says on the tin or it doesn't. And the customer buys if it does what it says, right?

[00:05:24] But a platform, the implication is that you have to still build on top of that to get to a solution that's useful for you as a customer. And so the way that you would mitigate that previously would be in our space to provide templates or kind of starter packs for the types of integrations that you think your customers would want. And the problem with those templates is they were kind of a one-size-fits-no one, right? So they gave you a very kind of false illusion of having a solution.

[00:05:54] But when you actually try to apply it, there'd be all sorts of things that didn't fit and steep learning curve. And you'd have to do a lot of work to make them actually work for your business. Well, AI has kind of obviated all of that. You can now use an agent in our platform to kind of prompt the specific integration that you want into existence. And it's able to do all of the testing that an expert would do.

[00:06:16] So it's not getting the design right immediately out the gate, but that doesn't matter because you can test it, see the problem, and then get to the next iteration. So that's why I'm very confident around a product-led growth strategy. But in addition to that, we also work with partners, particularly in the ERP and CRM space. These are typically companies who are involved in implementing and supporting customers against specific ERPs, CRMs, HR tools, that sort of thing.

[00:06:44] And they often have this problem that their customers present to them, which is they needed these products to integrate into other systems. And they're often struggling to find a right-sized solution. On the one hand, there are a lot of first-party integrations that are highly productized. They're kind of plug and play. And if they work for you, that's great. And they're usually very inexpensive. But if they don't, there's just zero customizability. And then on the other end of the spectrum, a lot of the offers tend to be oversized.

[00:07:12] We see this with competitors that are very kind of enterprise-facing. So Flowgear meets that sort of sweet spot, gives them an opportunity to have a solution for their customers that's right-sized for them. So in short, the focus is product-led growth as a main way to drive adoption of the platform, but also strong partnerships in the places that we know customers are looking for these solutions. And you mentioned AI there a couple of times.

[00:07:40] And of course, Flowgeo has been around for some time now, long before AI took over just about everything and every part of a business. So over the last three years, just how much of an impact has AI had on your business? Well, we actually realized going back three, maybe even four years that we wanted to do a very big revamp of the platform. And kind of the number one thing with software is you'd never rebuild the product because it's always much harder than you expect.

[00:08:09] And that certainly has been true. But it did allow us to rethink everything for first principles. And the timing, and I'd love to take credit for this, but it completely was unplanned. The timing around ChatGPT and then very shortly after a Gentic AI hitting was just perfect because we got to build support for that from the ground up. And it's much better to do that than to try and retrofit it on the product. So it was almost like we were a startup again at exactly the right time.

[00:08:38] But it's fundamentally changed things in terms of the build experience. You know, we used to sell the platform, as did our competitors, on the value propping around low code, no code or visual design. And that was the way to convey that you don't need to be an expert. You can build something that's sophisticated without needing to know what's going on under the hood, as it were. And now the play is different. It's, well, AI allows you to build beyond your own expertise.

[00:09:04] And so if the agent is able to effectively interact with the platform, a lot of the time our customers aren't even looking at the visual design of the workflows anymore. They're just trusting that the workflow has been built correctly. We still got a great visual designer. But most of the time when people are building, it's actually outside of the platform and it's often a means to an end.

[00:09:27] So there's been a big shift for us there as well, where historically most of the integrations were behind the scenes, automated end-to-end syncs that ran at a cadence or based on some event. And you didn't really know about the platform unless something had gone wrong, maybe some piece of data that you needed to tweak to fix the sync. That's completely changed. Those use cases are still there.

[00:09:48] But increasingly, customers are using our platform to build on-demand integrations that are required by the apps that they're building. So they don't want to have to run a business process that requires them to run multiple steps across multiple different apps. They're building apps that are abstractions on that. They're creating very light front end that allows them to achieve some sort of high-level business process.

[00:10:17] And then they're using Flowgear as the way to actuate all the tasks that that involves. And so a greater audience within our customers is actually exposed to the platform to varying degrees for sure, as you would expect. But the unlock has been profound. Whereas before it was kind of the domain of the central IT, increasingly just about anyone in the org is able to start interacting with the platform and getting the gains of automation.

[00:10:45] And I think it speaks to one of the biggest areas that AI has not really unpacked yet. If you think about where the data sits for knowledge work, the obvious initial answers are things like your cloud storage, your Google Workspace or OneDrive, for example. And obviously, agents can hook into that. But arguably, the real valuable data sits behind the APIs for ERP and CRM and warehouse management and other line of business tools.

[00:11:14] And so Flowgear gives you a bridge to safely connect into those data sources. But it goes beyond just reading from them. It can actually take actions. And so it's a way to realize agentic work for knowledge workers much more broadly. So huge shift. We're barely the same business that we were a few years back. And you've seen so many big tech trends from mobile to digital disruption cloud and then AI almost dropped at just the perfect moment for you.

[00:11:43] But what has been the most memorable thing working for the company so far? You must have picked up a few stories along the way. But what stands out to you as the most memorable moment? Yeah, I think this period of trying to figure out how AI needs to fit in has been a very exciting time to work.

[00:12:08] Kind of the lights went on when we started to see agentic work do increasingly more of the build work in our own code base. So, you know, you still need an expert to kind of oversee it and steer it in the right direction. And there's a ton of different failure modes. But there was just this kind of realization across the business as well. What we're seeing here in building our core product, we can have our customers see these same benefits when they're using our product.

[00:12:38] And that's been super exciting to work on. And, you know, it goes wrong in a ton of ways. And for a period of the past year, we've really been heads down trying to make this robust and reliable. There's been a ton of stuff to fix. It's been a hard grind. But incredibly rewarding because we've come out at the end of it with a profoundly different product that is far beyond anything that we could have done before generative AI existed.

[00:13:04] So I would say of all the things that have happened over the years, this has definitely been the most exciting time to be building. It is an exciting time right now. But I'm sure it's not all been sunshine and rainbows. And you said that things can go wrong. What have been your biggest challenges in your work recently? And how did you overcome some of those challenges? Because I'm sure that story will resonate with people listening as well. Because I think every business leader is going through the same thing at the moment.

[00:13:31] Yeah, I think the challenge is, you know, there's this idea of irreducible complexity that, you know, some things are just complicated. You can simplify them a certain degree. But fundamentally, a lot of business processes are hard. There's like, you know, really nuanced rules that have legacy reasons for being there that can't be changed. And so that's the thing that we grapple with. And that certainly AI is helpful there, but you still need experts.

[00:14:00] And so we rely quite heavily on solution architect when our own professional services teams are delivering to customers. So it's someone who can kind of bridge the gap between the expert in the customer who understands this process, can bridge that gap and understand exactly how that needs to be translated into an integration that the platform is going to be running. So, you know, there's nothing new in that sense. You can't remove the complexity. You can just push it somewhere else.

[00:14:30] And so we're trying to take over that. But there's always that gap where you're trying to get into someone's business and understand how it operates and their way of thinking about everything from the way they manage their customers to their inventory and so on. And I think a lot of the value that we provide in our professional services team sits there. You have to have seen this stuff before to recognize the patterns.

[00:14:54] And sometimes it comes down to, you know, an expert will answer the question that no one ever asked because they didn't realize that that was going to be a problem. So that's probably the fundamental challenge of the space. And, you know, at the end of the day, you still need an expert to help you work through that. And I'm curious, if you were to look back on your journey, you probably don't do this too much,

[00:15:19] but if you look back at your tech stack on day one and where you are now, just how much has the company's technology evolved since you launched? I think with our new platform, there's barely, I mean, I'd be surprised if 5% of the lines of code have survived in the past five years. So we've done a huge rewrite. We did it in two stages. And the new platform that we released this year was about a year in the works.

[00:15:48] And that was kind of the second half of it. So the product is completely unrecognizable. But, you know, that's software. And I am a little nostalgic. I'm in my 40s. And so I like to record these things. So when we're about to shut something down, I actually do take some screen grabs or take a short video. And we've got some of those kind of histories. It is interesting looking back at how far it's come. But I would say it's basically unrecognizable compared to where we were five years back.

[00:16:18] Absolutely love it. And what would you say has been your biggest milestones throughout the journey you've been on there? Any big standout moments? Yeah, I think it's always nice to see external validation of your work. In the early 2010s, we were invited to pitch at an event called Demo. It's actually a Silicon Valley event. And that year, for the first time, they were doing an Africa event, which was in Kenya. So we pitched at that.

[00:16:47] And we won for the enterprise category. And the prize on that was doing a trip to Silicon Valley the following year. And that was super fun. Around that time, we also got onto Gartner's radar. They were starting to track this new category of cloud-based integration platforms and what came to be known as iPaaS, which, of course, was their term. So we had some early validation around that.

[00:17:10] And then more recently, we just won partner of the year with IAMCP, which was very validating and we're very appreciative of. I think it's a good signal to the work of the work that the team has put in on the platform and the success as a result. And when you first started out there and solving a very real-world problem for your customers and you get to help and you get to hear so many different stories and feedback, etc.,

[00:17:39] are there any specific customer success stories that stand out to you? I don't have one sort of specific one that comes to mind. But I think the measure is generally if you become indispensable, you've done good work. And so we've got customers who've been with us. Well, we've actually got a few who've been with us since the start back in 2010, but many who have been with us 8, 10 years.

[00:18:07] And a good indicator is not just that they've retained you, but they keep coming back for more. And that is a strong signal of trust and also the value that they're getting. So usually the ROI is measured in terms of how much person time was freed up through some process. But I think it's important to look at sort of softer measures as well.

[00:18:30] Early on, we had cases where, for example, there's a call center that was established just to check in with customers each week to see what they wanted to order around product that week. And, of course, because of the volume of calls they had to get through, it tended to be a very transactional call. And we put in something that allowed them to automate that over like a WhatsApp channel.

[00:18:56] And so that meant that the kind of run-of-the-mill mundane reordering could be done without human intervention, which meant that the same pool of call center people could engage in more meaningful conversation with those accounts instead of just a weekly check-in. So I like those types of stories because it's beyond just that kind of raw efficiency gain and helps our customers to build better relationships with their own customers as well.

[00:19:22] And as we look to the future or growing the business, any customer growth or trends that you're seeing out there in the day-to-day work, anything you're moving on, anything you can share there? Yeah, I think the biggest thing for us is, again, just being around AI. One of the big things around that is this transition that we're certainly starting to see customers express interest in is that if they can avoid being in the front end of the product, they want to.

[00:19:51] Because more and more, they're able to do meaningful work directly from their agent app. And so for us, the way we've supported that is through our builder MCP server. That's what I referred to earlier, which allows you to build integrations without actually being in Flowgear. And oftentimes, you're not actually just building an integration for its own sake. You're building an integration because that is what's needed for an app that you're creating. And so a user can just say, I need an app that does this, and it will know to automatically create the integration on the back end.

[00:20:21] And broadly, we're seeing, I think there's definitely some nervousness in the market, but companies like Salesforce, for example, have a headless version of their product where they're expecting users to be able to interact through an agent. So there's no front end to this thing. And we think that for certain tasks, especially those that are onerous through the front ends, because the front ends aren't designed for the way a customer needs to run that process.

[00:20:50] Agents are a great way to do that. So I think that's going to continue. And if I just wanted to sort of characterize the stage that we're in, in AI and realization of value, if the past year or so has been around unlocking that value for building products by enabling it to write really good code and test it. So in other words, it's helped software engineers.

[00:21:18] Then I think the next year is going to be those kinds of unlocks coming to knowledge work. And, you know, you're seeing stories every day around how that's starting to happen. So we want to be there to make sure that we're providing value in that story. And as AI tools become increasingly embedded into everyday business workflows, I was also reading recently that Flowgear had launched Builder MCP to bring enterprise integration into any AI chat or IDE.

[00:21:48] So what gap in enterprise integration capabilities inspired this? Is there a story behind that too? Yeah, I think it's the similar sort of theme is that people want to be able to build from their agents and they don't necessarily want to explicitly create an integration. What they want is a solution and they can express that in the chat.

[00:22:09] And if the agent realizes that to achieve that solution and integration is needed, it can create the workflow on their behalf without them necessarily even being aware that that's what's happened. And that's the unlock of MCP. It's a way to bring Flowgear to where people want to work instead of requiring them to be inside the platform. And I'd love to have a bit of fun with you before I let you go today as well. I think we're all in a constant state of continuous learning at the moment.

[00:22:38] It can feel overwhelming. Everything you learn today can be updated within a couple of weeks, especially with the speed of AI tools and how they're continuously evolving. So as a business leader, how do you keep up to speed with tech trends and continuous learning and ensure you don't get left behind? Any tips you can share on that? Yeah, I mean, this is very kind of domain specific, but I find Twitter or X, I guess, as a really good source of information for the field.

[00:23:07] There are a lot of fraudsters out there as well. So I tend to just move very quickly through content and try and pick up a couple of nuggets a day. And then the other thing that I've used almost every day for probably 10, 15 years is Hacker News, which is a very high signal feed of important events, usually related to the software industry. That's been super useful. Other than that, just experimenting.

[00:23:34] I mean, I don't really have a lot of time to do stuff on the side or hobby projects, but where it applies to the platform, I'm still pretty hands-on in product R&D. So that's the place that I'll generally experiment with something. And there have been waves of things that I just haven't really latched onto. I still don't understand crypto.

[00:23:57] But certainly, you know, I'm always kind of closely evaluating whether emerging tech has relevance for our platform specifically. And I like to, I guess, learn by experimenting there. Brilliant. And for anyone that wants to learn more about Flowgear, can you tell the listeners a little bit more about who your customers are, companies' core products, and ultimately how people listening and hearing about you for the first time can find out more information about everything we talked about here? Yeah.

[00:24:26] I mean, welcome to hit me up directly, daniel.flowgear.net. Otherwise, one of the contact options on our website. We'd love to have a conversation. Excellent. Well, I'll add links to everything that you mentioned there and including your LinkedIn as well, so we can keep this conversation going. But thank you for sitting down and sharing your story today. Really appreciate your time. Thanks very much, Neil.

[00:24:49] After listening to my guest today, one of the things I'm taking away is AI can generate an integration, test it, find a problem, try again. But every business contains rules that are also shaped by customers. Inventory, history, and the occasional decision that nobody can quite explain anymore. And complexity can move, but it rarely disappears.

[00:25:15] But the encouraging part here is what happens when automation is applied well. Daniel's example of moving routine weekly orders to WhatsApp allowed call center employees to have better conversation with customers. I think that tells us much more than simply counting automated steps. And I loved his approach to continuous learning. Scan trusted sources. Ignore the fraudsters. And experiment where the technology connects to a real product problem.

[00:25:44] And kudos to him as well for admitting that he still doesn't understand crypto and blockchain. So there is still hope for the rest of us. But a big thank you to Daniel and Flowgear. Remember, you can find him on LinkedIn. Check out the Flowgear website. You can also find me at techtalksnetwork.com. And if you've got a feedback with me, tell me what complicated process inside your company deserves a better connection.

[00:26:10] And while you take that away and allow that to marinate, I'm going to prepare for tomorrow's guest. Thanks for listening. Bye for now.