ISV Talks
ISV Talks
AI Is Not Replacing Platforms. It Is Elevating Them. An Interview with Flowgear.
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In this ISV Talks podcast episode, Carol Livingston, owner of Dynamics Connections, interviews Daniel Chilcott and Jon Milner (JJ), Managers and Co-Founders at Flowgear. They share how AI is reshaping the integration landscape and why trusted platforms with real engineering depth matter more than ever.
Daniel and JJ explain how AI is helping non-technical users build clearer prototypes, improving communication between business teams and engineering, and making it easier to connect data across multiple systems. Their message is clear: AI is not replacing integration platforms. It is strengthening them and expanding what organizations can accomplish.
They also discuss the growing importance of trust. Many companies are discovering that quick build tools are not enough when integrations touch revenue critical systems. Customers want platforms with proven connectors, deep domain knowledge, and the experience to handle complexity at scale.
This episode is a great resource for anyone exploring AI, data connectivity, or the future of cross system intelligence.
Learn more about Flowgear: https://www.flowgear.net/
Welcome And Meet Flowgear
SPEAKER_00Welcome to this episode of ISV Talks. I'm Karen Livingston of Dynamics Connections and your host of ISV Talks Podcast. And on this episode, we have Daniel and JJ from Flowgear. Welcome, guys.
SPEAKER_02Thanks for having us. Hey, Jim.
SPEAKER_00Hey. Well, can you both tell us a little bit about yourselves and how long you've been with the company?
SPEAKER_02Yeah, Daniel Chilcott. I'm co-founder at Flowgear. I'm one of those nerds who grew up coding. I got into a tech support job as my first role. That business started a software role, started writing CRM as part of that, and then realized that every time we shipped that product, what our customers really needed was to be able to integrate the CRM into their other products. And that's kind of where the idea for creating an integration platform came from and the company that eventually became Flogio, which we started in 2010, and we've both been involved since the beginning.
SPEAKER_00Great. And JJ, introduce yourself.
SPEAKER_03Also a bit of an IT geek. Started when I was 17. I'm 52 now, so showing uh a little bit of gray, been in the industry since it actually started in the garage.
The Cloud First Origin Story
SPEAKER_00Great. Well, Daniel, you want to tell us a little bit about how the company started? I think you kind of started talking about that, but go ahead and give us a little bit of the company history.
SPEAKER_02Yeah, I mean, so I I was seeing that every time we try to take our product to a customer, they needed to integrate it. And so, you know, didn't really know a lot around enterprise services buses from you know back in the day, but built some on-premise tooling that would allow integration between different systems. We did a few CRMs, a few ERPs. And that was in a predecessor business that I started around 2007. And actually, JJ, through his other business, became a customer of ours. And after we'd worked together for, I'm not even sure how long, maybe six months, a year or thereabouts, decided to go into business together. And the idea for Flow Gear was that from day one, it would be a platform that was a cloud integration platform. And that was actually quite unusual at the time. You know, a lot of software was starting to move to cloud versions, obviously, Salesforce in 99 being the first such product. And so many industries or categories of software were starting to move across that way, but it hadn't really happened for integration yet. And I think I'd I'd love to claim that we had some special foresight, but it was just super fortuitous timing. And around that time, our industry was starting to do the same thing. And in fact, in the early years, we often had to justify to customers why they would want a cloud platform to integrate when a lot of the endpoints were still on premise back then. Um needless to say, that that changed pretty quickly.
Why ERP And CRM Integration Hurts
SPEAKER_00Yeah, I bet. Well, you were a visionary at the beginning, it sounds like so well, tell us what's the basic description of your products for dynamics and kind of what are your unique features. JJ, do you want to take that question?
SPEAKER_03Yeah. So I mean, there are a lot of options you have when you need to integrate more than one system. Of course, you can hand code, and that comes with its own challenges, or you can use some kind of integration or import export tool normally is what you find from a specific vendor. But the challenges of ERP systems and CRM systems is they really go to the heart of companies. And every business is unique. The CRM and the Dynamics products, uh, there's the AB20 principle, there's a whole lot of components that are common to multiple customers, and that's why they choose a product like Dynamics. But then they need to connect that into the rest of their business, and they'll have a whole myriad of systems. And so Flogia's a unique advantage is number one, that it has a really, really rich ecosystem of connectors, over 250 connectors to the various platforms that companies would need. And then because it was been around as long as it has, and because it actually originated in South Africa at a time when bandwidth was a problem, we developed a very rich capability called the drop point, which allows us to reach into on-premise systems, systems that were never developed with cloud and integration in mind, potentially non-internet routable, and whether that be a warehouse system or an asset management system or a ticketing system or an HR system, it doesn't really could be flat files. This really allows you to get the most out of a your dynamics platform because it doesn't matter where the data sits, whether it's on-prem, whether there's an API, whether there's an SDK, or whether it's an A to the cloud application.
SPEAKER_00Yeah, and what other products are you compatible with? I know you mentioned dynamics, but yeah, 250 connectors. There's probably a lot of ERPs that you connect with.
SPEAKER_02Yeah, I mean, you name it. Yeah. So, you know, all the usual ERP players. Our objective, and I think this is pretty important as a as a differentiation, is you know, a lot of vendors have integration capability baked in, but it never serves their purpose to integrate into a competitive platform. Whereas because we're an independent vendor, it suits us to have competing platforms in every category. And that kind of best serves the customers' interests. You know, we have a lot of customers who are using different ERPs across different businesses that they own, for example. So it's important that you can consolidate that. But of course, we're we're much broader than just ERP. So you know, virtually every software category, whether it's CRM, HR, warehouse management, and so on. And we also have what we call technology connectors that deal with open standards and services. So if you need to get to a database, a file system, FTP, an API, whether it's Graph or REST, you've got generic connectors at your disposal for all of those as well. In addition, we have an SDK. So folks who have uh a software proficiency are also able to build connectors and can often forklift investment that they have. So connectors that they've already built as hand-coded connectors, relatively easy to forklift those as reusable connectors in the platform for their own internal use as well.
SPEAKER_00Nice. Yeah, and it sounds interesting because uh are there any particular industries that need these connectors that you work with or specialize in?
SPEAKER_02Yeah, I I mean from where we're standing, every business has this problem. So, you know, we have customers across all verticals, but I guess there's some common themes. So we we have a lot of customers in the technology industry, MSPs, for example, telcos, SaaS vendors, in financial services. We work with a number of credit unions, we have several insurance customers, a good spread in things like manufacturing, retail distribution, so so quite diverse, but they all share similar issues. And while there might be quite a spread of different integrations, many of them either start or end with the ERP. So anytime data needs to move across different systems, that's kind of with respect to internally in the organization. And then another key aspect is the ability to integrate you know, up to suppliers or down to customers or perhaps to partners. And the advantage is that you can do that in a mode that suits your organization or suits your supplier customer. So on the one hand, if your supplier mandates that you engage with them using files over FTP, the platform can do that, it has the connectors. On the other, if you want to be able to offer an API for uh billing purposes to your partners, you can actually use Flogia to create that API and then allocate access to that to your to your partners so that they can in turn consume that. So it's kind of all things to everyone, and it allows you to interact with third parties in whatever way is appropriate for that business relationship.
SPEAKER_00Okay. Still gonna kind of come back to that question. Are there some industries that you do specialize in or that tend to you see a lot of business with?
SPEAKER_02I I think probably the one would be technology businesses.
SPEAKER_00Okay.
SPEAKER_02While the key kind of aspect of our vision is to reduce unnecessary complexity in the integration, right? You know, there's a lot of complexity that we want to abstract away. I think we still also resonate with SaaS vendors because they often feel this pain. They often want to make integration available to their customers in a way that's repeatable. And so we tend to resonate there where it it's sort of grudge work. And so anything that they can do to make that easier and scalable is a big advantage for them.
SPEAKER_03Right. No one really hugs their integration engineer. It's a bit of a thankless job.
SPEAKER_00Right. Well, JG, is there a story that you can tell us about a customer that was successful with your product and why?
Customer Success Stories In Production
SPEAKER_03Yeah, so I'm I'm always impressed with just how far FlowGear can take a business, particularly a successful business with a fantastic reputation, a big client base. One often thinks that you know these businesses have done everything that they could possibly do. And one of the ones that comes to mind is a company called Core Group. They distribute Apple in South Africa. So if there's an Apple product that comes into the country, imagine the volumes associated with that. FlowGear connects SysPro in this case and their warehouse management system, which is called Manhattan. Manhattan's the same way our system Boeing uses for parts. So highly complex, very sophisticated bin management, stock keeping management, lots of challenges around how you'd handle uh discrepancies in the data, inconsistencies. Maybe stock needs to be moved from one virtual warehouse to another. More than one system needs to handle that. Human in the loop, corrections of data. Uh so that's one, and and you can imagine that they're a very demanding customer because the customer experience is so important to them. Uh so it's one that we're pretty proud of uh because it's been a very successful relationship and a lot of amazing work with them. Another one, I won't mention the company's name, but they're a telco and they had a challenge that they needed to execute Flogia within their data center. And they needed very, very low latency on the connections, doing CDR management, basically telco billing. They can imagine every record in a telco is money, and any record that's lost or that is incorrect is money lost. So it's harder than dealing with a bank. Banks don't work in real time, telcos work in real time. And so that's one of the unique aspects that Flogia has is that we can execute in a number of different modalities in cloud and dedicated subscriptions, and in our enterprise customers, if they have a requirement to run in their own environments, we can do that as well.
SPEAKER_00Right. And and I bet security is probably a big concern for those companies. That's pretty sensitive data. They don't want it just out on the public internet or in public AI engines and so forth, right?
SPEAKER_03Yeah, absolutely. You know, Fluge was built from the ground up to be secure. We're one of the few platforms where each client's data runs in their own virtual machines. So while it's a cloud service, you consume it, then it feels like a shared service. We're not sharing memory, we're not sharing disk space, and for you know, further isolation, even in their own subscriptions at the enterprise level.
Security Isolation And Deployment Options
SPEAKER_00Nice. So maybe both of you can kind of tell me a little bit about what's kind of in the future and what are some of the new things you're looking to introduce to your clients as they move to the cloud or just to add new value to the product.
Governance Controls For Safe Changes
Agentic AI That Builds And Fixes
SPEAKER_02So our value is around thinking about what's the maximum amount of work that we can take away when you're trying to build an integration so that you're just left with the absolute essentials to make it functional. So we try and abstract as much as we possibly can. For example, you know, the connectors all work differently. Some of them are based on one type of API, some are another, some don't have an API, they're doing something else under the hood. That's part of the complexity that we're abstracting. Similarly, it's one thing to create a workflow and have it run the first time, but when you need to evolve it over time, you need to be careful that you're not shipping something that's broken into production. You need a control process to make sure it got reviewed and approved before we shipped into prod. Those are what we call our governance controls. And again, that's part of what the platform manages. Now, when AI hit a couple of years back, you know, within a very short period of time, the whole team was using things like ChatGPT to bang out pieces of script and you know start to work on coding. But then 18 months ago, when agentic AI hits, this is a completely new way of working. And so we're immediately evaluating how do we take what's working here in our software development practice and apply it to integration. And it turns out that it's not that far removed. So you know, if code is instructions to perform some tasks, well, the way that the workflows are described, what's actually behind them is a script, and that's a set of instructions that need to run to create that integration. And so we built agentic AI into the platform. It was released earlier this month, and it's proving very successful in a few areas. One is for new users, it means that you can go into the platform not understanding any of the mechanics of how you would visually build, not knowing which connectors are available, and use a short prompt to get you from zero to one. You can watch it build. And then, secondly, for a seasoned user, they are able to prompt things very specifically in a way that is more succinct than creating those steps manually. So, so that's you know, so two scenarios where it's really just a huge advantage to use. But I think the key thing is in the word agentic itself, which is among other things, what it means is that it's able to evaluate the result of something and then take a different route if it didn't work. And we realized that we could do that in FlowGeo. So we're using AI to help you build a workflow. We kind of have the design on the screen, but then it's actually able to run the workflow and test the results of it. So it can see the logs, it can see the error message, and if something's wrong, it can iterate on it and fix it. And so, much like we see in software development today, we you have those kind of gains happening in the workflow design kind of build experience. And it's fun, but also saves a huge amount of grunt work. So we're pretty excited about that. And we think that that's something that's going to be a defining characteristic, it's something that the market is starting to expect as well, but really allows you to build much more competently and quickly than has been possible before. So we're excited about that capability, and that's something that we're continuing to strengthen through this year.
SPEAKER_00That's amazing. Software that tests itself.
SPEAKER_03Yeah, essentially. It goes a step further than that. There's a standard for AI called MCP, Model Context Protocol, which is a new version of APIs, effectively, that's designed for LLMs, for large language models. And the idea behind it is that you point your AI system claude codex, whatever it is, to an MCP server, and it's kind of self-describing. It tells the AI what methods are available, what resources are available, and how to use it. Imagine all businesses have been with FlowGear now for 15, 16 years. They've got a rich set of APIs that connect to their business. And I don't know a CEO today who isn't looking at their board and having discussions and saying, well, how do I get to the forefront? How do I become relevant? How do I adopt AI? How do I do it responsibly? How do I do it safely? And Flogia has the ability to take the existing APIs that a customer has today and represent them instead of a as an API, as an MCP service that you can point an LNM to. So for new workflows, that's obviously very exciting. But it also means that you can take systems that were never built for AI and AI enable them. And Flow will discover the structure of those services. Imagine taking an ERP that hasn't got the API hooks yet that you're looking for. Imagine having the permission sets and all the rules that sit within the ERP presented by the SDKs, and we've wrapped those connectors. And what we've done is we've almost got an integration to an integration layer, a double adapter, as it were, that allows LLMs to speak to Flogia without you having to necessarily rewrite them. They can benefit from rewriting because there's certain aspects of MCP that you you want to take further. But this is incredibly powerful because it allows businesses to leverage their most prizes asset, which is their data. If you talk to any of the leaders in the market today, the real the real gold is private data. The LLMs don't know that, right? They were trained on public data. So if you have warehouse information, stock information, information on your customers, that's your competitive advantage. And now you can securely make that available without having to prototype and do all this sort of what I call AI pixie dust, right? And get into real AI work that's going to yield real results.
Deterministic Workflows Versus AI Hallucinations
SPEAKER_00That's amazing. Wow. And that's really going to help customers in the future be able to leverage their own data, right? And so that's pretty amazing. Any new trends, JJ, that you're seeing on the horizon? What are you seeing that maybe even others or the industry might be developing in the future?
SPEAKER_03So one of the one of the challenges of AI, as we all know, is hallucinations. And Dan already mentioned that Flower has the ability to test its results. And that's useful when building the workflows. But more importantly, from an integration perspective, you need the result of that integration to be the same every time, which is different to a hallucination, right? Both could be correct, but the information could be presented slightly differently. You can't have that when you're integrating transactions or when you're integrating CRM data and you need postal information to follow a certain format, etc. And so the trend that we're seeing is a greater appreciation of how you get the most out of AI, which is to use agentic decision making, but then and that should be like maybe 20% of your process, and then to have the remaining 80% deterministic, 100% verifiable, runs at scale all the time. And so the trend now is to look for systems that can bridge those two worlds together. And I mentioned earlier on that FlowGear can take these hardened, battle-tested workflows that are already APIs and make them AI enabled, so you can now get the best of both worlds. And we're really excited to see how customers can adopt that trend, maybe point customers to the right data sources using AI, and then use their Flow Gear workflows to deliver a consistent experience when it matters.
SPEAKER_00Yeah. And so is this data they don't already have or know of that exists out there?
SPEAKER_03Well, it would be a combination, right? So the deterministic data would be their data, their private data, the gold, and the agentic data would be a combination of whatever they are models we're trained on. And that could include interfaces like voice, right? And video, not necessarily only text.
SPEAKER_00Oh, yeah.
SPEAKER_03So it's bridging those two worlds, right? Imagine a customer who's got a query. Yes, they could type into a chat agent, but there's some amazing tools out there where you can just speak in natural language. Imagine a driver doing deliveries and he doesn't want to use his hands while he's typing. He can just say I'm running late or give me another route or plan a route for me. And the route management system where we've got APIs hooked in directing into a customer's database can do those sort of things. So I think we we haven't even begun to see how the intersection of these different areas can come together. But the real power is in merging that deterministic element and the AI element together safely.
SPEAKER_00Yeah, that that's pretty interesting because there is a lot of traffic data out there. There's weather data out there, you know, so you can use that in a practical way for your field or remote workers and travel time. And that's amazing. Yeah, I love that. I love that. Any other thoughts, Daniel, on that topic of just any future trends that you see?
SPEAKER_02Yeah, I I think everything that's happened in AI has really served to educate the market of what should be possible. And and the disconnect is, well, there are several. I mean, you know, security is a huge issue that needs to be dealt with, but there's there's a growing awareness of the types of problems that AI should be able to solve. And so customers are coming to us because there's this sort of this realization that this is the platform that should be able to do it for them. So there's a far greater understanding of just how broadly the platform can be deployed. Whereas, you know, in prior years, we'd usually hear from folks because they had a very specific use case, help me connect A to B and B to C. But now it's it's much more pervasive. They're thinking about a way to connect everything in the business across all departments, across all systems with all customers and so on. And, you know, part of this is the kind of paradox of plenty, where the more abundant or the cheaper something becomes, the more of it's consumed. And so previously where there was probably a much higher cost to building any single integration. And if you can now do that in a fraction of the time, it just means you're going to get more done. And all of those kind of efficiency gains have a significant cumulative advantage. And I think businesses realizing that. So, you know, our role is to really help them figure out those gains. You know, in some of the examples that JJ gave you, those things can be done, but they have all sorts of kind of gnarly bits to deal with. Like, you know, business logic that no one had realized was going to be that complicated. But we have all the data sources, we have the ability to map this. And we have the acceleration that we can get through AI. And so these are things that can be brought to market and provide huge value for business.
SPEAKER_00Yeah, sure, sure. So any other last thoughts that you have before we start to wrap up here?
SPEAKER_03Well, for me, and when I talk to customers and they're saying that how should they adopt AI, what should they be doing about integration? What I always say to them is this is not just an IT department opportunity. You need to expose these tools to uh your execs, to your leadership teams, to your customer service teams. And that is actually possible now with AI. They don't have to have this deep development bench in order to identify a problem in the business or an opportunity to do something faster, quicker, cheaper, and make some mistakes, right? But that's also important. And what I say to customers now is that there's such a compression algorithm happening. If I I remember this when I started IT, as I said at 17, it was possible to know everything. Right now it's possible to know everything about AI in three months, six months, a year, I don't know that you'll catch up. Right. So it's a really unique opportunity to get on the bus and start experimenting and start understanding where it works for you and what uh I've even I've even changed the way I speak to people because I'm used to structuring my prompts. And when I talk to you know staff and colleagues, and I need to, I think I've become a better communicator for it. It's almost forced me to be more logical and not to assume as many things. So you know, maybe it's rewiring my brain, but I think there's an opportunity for those that are bold enough to get started and to start to learn and to feel comfortable being a little uncomfortable and start to learn you know the techniques to take advantage of the opportunities that are now there.
SPEAKER_00Yeah, that's a good point about prompts. Like when you start working with you, you have to be more specific and intentional about the words and the description when you're interacting with AI. So that's that's a it kind of does rewire you. And even I like how you said that's affecting your communication just interactions, right? So what about you, Daniel? Are you seeing what kind of last thoughts do you have about kind of this whole evolution that we're in?
SPEAKER_02Yeah, I I I think there's a lot more expectation from being being in the software industry for for many years now. You've always understood what's possible and what you get done is is a trade-off of like how much time you're willing to put into it. And that ratio has just changed so profoundly. And I think what's exciting is that it's always kind of been this profession where no one outside the industry industry really understood it. But for better or worse, everyone's learning real quick. Um, and the fact that you know someone in finance can knock together a front end that is a pretty good substitute for the BI tool they were using, uh that's great. Because when they come with a you know a really complicated data transformation issue, that's something that our platform can handle. But they've been able to build to a level that they wouldn't have even been able to start before. And one of the huge disconnects around, I guess, any technical field, but specifically in view is software, is how does someone who does not have a technical background communicate what they want? They intuit what they want, but it's often hard for them to specify it. Whereas if they've already had a stab at it going through some sort of vibe-coded approach, that is a great way to specify and get exposure to the decision points that you didn't realize were there. So, in terms of kind of closing the gap between kind of call it the engineers and those who are not in the software engineering industry, much, much easier to communicate. And I think that's going to yield a lot of good as well.
SPEAKER_00Yeah. Yeah. I wonder if data warehouses will become a thing of the past. Because now you can get access to the data.
SPEAKER_02Yeah, I mean the hard part's getting all the data there at the right kind of timing. You know, if if you've got everything there, then for most types of reporting, I think vibe coding is a pretty good way to do it. Um, maybe if you need to speed it up a little bit, it might be better with a platform. But essentially, you know, if you can express what you need in English and the data is sitting in one database, it's certainly a lot easier than it used to be to put that together.
What Drives Customers To Flowgear
SPEAKER_00Oh my gosh. Yeah. Well, thank you both. I I have one last question I want to ask both of you. So kind of our wrapping up here, is there one trend that you're seeing that's driving customers or partners to come talk to you?
SPEAKER_02Yeah, I think for our case, it it's certainly the sort of general awareness in the market of what is possible. Whereas, as I mentioned, you know, previously customers would come to us because they knew one specific thing. Now they're generally coming to us because they have a general understanding of the value that the platform could give across the whole organization. So that's that's a big shift that we're seeing. And I credit AI for uh that trend, at least in large part.
SPEAKER_00And DJ, what do you think? And what's kind of the one trend you're seeing that's driving customers or partners to come talk to you?
SPEAKER_03They want something they can trust. They want a platform that understands the data sources that matter to their business. It's very easy to fake this for a minute, but as soon as you actually get into the weeds and you have to move data around, it becomes very apparent when you're a platform that's built for what we would call citizen integrator, uh, with some patterns that you know grab an email, put it in the box, send the email. It's very different when you're responsible for something that is the earning power of a business. And so more and more we're seeing an appreciation from customers that pedigree matters and that experience matters and domain expertise for the endpoints and the industries that you know matter to them matters. And uh, we're fortunate with Flogia that we've got a very rich ecosystem of partners. So they can take the experience that we built over the years and then just add another layer, another force multiplier for particular verticals. And we've seen the partner community grow from strength to strength because of that.
Contact Details And Closing
SPEAKER_00Great. Well, that's amazing. Well, thank you both, Daniel and JJ, for joining me today. I think this was an amazing conversation. We learned a lot about Flow Gear and the future. And here's their contact information if you want to reach out to them. And Daniel, want to go over your contact info?
SPEAKER_02Yeah, Carol, thanks very much for having us. Uh, you're welcome to reach me directly. It's uh Daniel at Flowgear.net. Our website's also Flowgear.net.
SPEAKER_00Perfect. Well, thank you again for joining us, friends. We'll see you on the next episode of ISV Talks. Bye bye. Cheers. Cheers.