RETISIO: Okay. Good day and welcome to Rizzo’s Unlock Digital Commerce Podcast. This is the show where we explore the latest trends in technology in the fast moving world of digital commerce. In this interview style podcast, we ask the important questions of expert digital commerce practitioners and retail leaders to help you gain traction and sell more online.
In our conversation today, we’re delving into a topic that is the backbone of any successful e-commerce venture, and that is integrations. And we figured who better to explore this subject with us than an expert who every day through his company is pushing the boundaries of e-commerce integrations. We are honored to have Wally Ibrahim, the CEO of Jive joining us today. Wally has been in the e-commerce technology world from front end to supply chain for 20 years. He believes every business is an e-commerce business, and his passion is providing them with practical technology solutions.
Welcome Wally, and we appreciate you being here and are looking very forward to your insights
Wally Ibrahim: today. Thank you, Brent. I’m looking forward to the
RETISIO: conversation. Awesome. We also have, as always, Tony Morris lending his expert insight and opinions on all things related to digital commerce. He’s the, currently the Chief Technology Officer at RETISIO and has deep experience in leading retailers and brands and digital commerce tech implementations to operate more efficiently, reduce risk, and grow their businesses. Hey, Tony, welcome back.
Tony Moores: Hey, thanks Brett. I appreciate it. Looking forward to talking
RETISIO: Okay, let’s we’ll start the conversation today and I’ll just start with Wally with a question and we can see how the conversation goes.
So maybe just to give us a, from a high level perspective, Wally, why are, and for layman in the audience and which include myself, why are integrations and e-commerce so important to online businesses?
Wally Ibrahim: Yeah, great question. I think it’s always been important. You know that back in the day when e-commerce was new, maybe you had your e-commerce platform.
You still had to get that order. Down to a warehouse at the very least. Fulfillment system, so on. And I think this is not so much an e-commerce problem anymore in terms of how things are changing, but as things have regressed more, Software is being provided in two different ways. I shouldn’t say two different ways, but in, in mostly through a SaaS product, online offering.
And the other thing that’s happening is that there are more point solutions out there. So it used to be that you would buy. One large software package, maybe two or three. You’d integrate ’em together, you’d buy servers, get your IT department to get everything together and figure out how to fit it together.
That’s really not what’s happening anymore, right? A lot of the really best solutions that are out there are. Or point solutions to solve one problem, right? So you’ve got shipping solutions tax calculation engines, order management systems, marketplace management systems, e-commerce systems, marketing systems of all sorts in e-commerce.
And they’re all provided as solutions that are independent. Individual silos online. Now that’s great in one sense that you can sign up in many ways and use ’em all right, or for each one of them, but they don’t always work together. And even when they do, they may not work the way that you intend it to.
So one of the reasons integration is so important is just simply because if you really wanna build a best of breed system, you have to have. A selection of best of breed software, and that selection has to work together, and that’s where integrations comes in. You have to have it all work together, whether it’s your customer management platform, call it crm.
E-commerce system, all of that. If you really want best to breed across the board, you’re probably gonna be selecting point solutions these days. And those point solutions have to work together and they have to work together with however your business model works. So integrations is critical in the sense that you’re always gonna need the order to go down to some system, right?
So those are the business, what we call business critical integrations. And so that’s complex enough when you’re choosing a bunch of point solutions. But as you’re choosing different systems that go beyond just what’s business critical what is, on the critical path of getting an order down you need them to all be integrated in order for any of it to make sense.
Tony Moores: Yeah we, we’ve got some listeners some are, relatively new. Some of this, this kind of stuff is all hat. In, Integra, I totally agree, right? Integrations have always been the keystone of a successful implementation. It’s, there’s no such thing as a system that has only one part.
Sure. I think, back in the day, we had systems that were made up of small numbers of very large parts and now that’s changed to large numbers of small parts. Exactly. But back in the day, most of it was. Data driven, right? Processes were serial first you did this, then you did that, then you did the other thing.
And the thing you were trying to get right is that, the data was appropriately extracted, transform, load, etc., right? On or between those steps. That landscape, I believe has changed, right? The nothing’s serial the speed at which people, happen, you expect things to happen.
The amount of choices you can make or the branching between I’m curious tell us, what’s your opinion on how. The focus has shift or changed? Were any new challenges that, the rise of composable commerce, the shrinking of best of breed, the, the popularity of M.
Tony Moores: Does that kinda change the way you guys approach integrations and the importance they have?
Wally Ibrahim: Yeah. That’s, wow, there’s a lot there. So I think you hit the nail on the head, first of all, saying it was a small number of large systems and now it’s a large number of small systems. To play into the entire idea of mock composable commerce that’s all great.
And it, I’m not saying that sarcastically, that is great. That is, I think from a technical perspective, The way technology and industry needs to go, you need, if you’re gonna have a large number of small systems they, they’re only gonna work together if they are composable, but you need a composition layer, right?
A layer that integrates them all. That’s how we view the world, is we’re not there to solve your individual business problem on one individual fulfillment system, one individual digital delivery system. Marketing system, content management, so on, where they’re to say, you have a large number of small systems, you need to compose them and get them together, right?
Composable doesn’t mean that they are composed, it means that they’re composable. How do you compose them? And the answer to that is integrations. And there’s a few levels of integrations, of course, right? But classically, you’re gonna say there’s backend integrations and then there’s integration of the screen.
Where, you call it single sign on as one part of that integration to the screen, meaning data coming from multiple systems on one screen. But then there’s backend integrations of course, which, the typical example of that, of e-commerce, of course, is like setting an order from your e-commerce system to your order management system or warehouse system, inventory management system, synchronizing inventory at the backend.
And that’s more necessary because again, To, to use your phrase again, it’s a large number of small systems. And if you’re gonna pick best of breed systems, which I think is the right way to go you’re gonna the best way you’re gonna do that is, is to pick point solutions that are great at what they do and then compose them to work with each other.
Tony Moores: Yeah. That, Really important and I understand that, but like I have this is a personal issue, right? It’s sorry for venting, right? But like I have, I use a Mac and I have an iPad, right? And I don’t use the iPad quite as much as I use the Mac. I’d like to use it more because it’s more portable and yada yada.
But for those of us who are old enough to remember the portal days, yes. I feel like, I have this device. Yes. With lots of little independent portlets or apps on them that might do what they do. And I’ve got the option to select from multiple apps that do the same thing better or worse, or however they fit my needs.
But I still struggle daily with the interoperability between those apps. Which is why. I prefer to work on my back, right? Because I can leverage the operating system and I yeah, I I like to explain people, the, when they ask about mock I say, look it’s really the natural, shrinking the best of breed, right?
He used to take the best e-commerce, the best o m s, the best search and put ’em together. Now you go into, just one of those systems and you can say, I’ll pick on e-commerce, right? You can take, the best the best cart, the best promotion engine, the best pricing engine, right? So you’re doing the same thing.
You’re making them small, but you still run the risk of. Can you leverage that individual solution to its fullest potential if you’re missing the quote unquote operating system? Exactly. And when I first heard about Jive, like I got excited about it because I won’t describe it as an operating system, but it’s, it seemed to look after some of those things I’m worried about, Can you
Wally Ibrahim: talk to us about, it’s funny that you say that term operating system because internally so obviously we’ve been known as in an integration platform, specifically focused on e-commerce.
But internally what we like to talk about is being a business operating system. So going beyond just moving an order, going beyond just calculating tax. Not calculating taxes to say, but, making an API called integrate tax engines and to say we’re an operating system for business. And so what’s the difference there?
And it’s really to say, there are nuances in the way that, that different point solutions work and integrating them and just moving messages between them is really not enough. And I’ll give you a really simple scenario that we’ve run into a few times. You’ve got two systems, right?
You’ve got your e-commerce system, you’ve got your order management system and you need to run a refund. All right, so you can run a refund either. In the order management system because, Somebody who’s on the order management system realized that this item’s not in stock. You might run a refund on the e-commerce system because that e-commerce system for some reason says, Hey, something went wrong.
Or, the user went in and said I’ll issue the refund through here. You might issue it through a logistics solution where, The user themselves. The customer is going in and requesting a refund. All right. That’s great. But the problem that you run into when you do these things is realize, There are nuances in even what the definition of that is, right?
Somebody might say, one system might say, look, a refund is you bought a hundred dollars order. You want a $20 refund. Another system will say no, you’re returning. This one item will issue this refund. All sorts of little nuances of how things are calculated and. You can go through each one of
Tony Moores: those at the order, at the line item.
Exactly. Do you include the shipping? Exactly. Do you include the tax? Was there a promotion? Do you have to split that up across three different Yeah. All of that nonsense.
Wally Ibrahim: So now we can integrate all of that and we do and make it, synchronize between three different systems and reconcile the order properly.
But at the end of the day, if the date is not being captured from the user that is pro that, that’s proper, right? One system might capture what the total refund is, but really does, it captures the order level, like you said, but not at the light item level. Another system captures line item level. And so how do you reconcile that between different systems?
We can work all sorts of magic to do that, but the real answer is to say, look, none of these systems fully capture what a refund means for this business. Because what it really means is capturing the full details that all three systems need in order to operate properly. What really do is issue another screen where the user can then go in and say, and this is what we talk about when we say business operating system, but another screen where the user can go in and say, I’m gonna issue a refund and it’s gonna propagate to those other systems.
So when we talk about integrations or, internally, like I said, business operating system, we, I like to talk about the two is the APIs. And the UIs, right? User interfaces and application programming interfaces. And classically in integrations, you’re only talking about APIs. We like to include all interfaces there.
Because an integration’s not gonna be complete unless you actually a lot of times capture some other either extra information from a user through a UI or sometimes. Have to initiate it through a UI because nothing in those systems that have to work together, for example, to go back to the refund example, to capture the details of that refund, none of those systems capture all the details that all of them need.
So you really need a third screen where the user can say, I’m issuing a refund here now. So this actually works really well. If you look at it from a composable commerce perspective, right? I’m taking a bunch of APIs essentially that all need to understand something, some piece of what that refund means.
Either one of them is gonna allow me to enter it into their native screens. I’m adding another interface, a user interface, not an API interface, but a user interface where the user can then give me the details of what that refund is, and then that integration will then propagate that refund. Properly throughout those systems.
Tony Moores: So yeah. Pardon, pardon me for being cheesy, right? But the two is make an us right? And there’s no I in team. And at the end of the day, we’re building a system, right? And it’s about how those parts. Interact, cooperate and really align toward a mission that determines whether the system is good at what it’s doing or not.
Wally Ibrahim: Absolutely. And in some ways that’s a new problem, right? Because to go back to your description of, it used to be a small number of large systems, it wasn’t as big of a deal, right? Because you had a small number of large systems. Maybe you had 1, 2, 3 systems. You had one system, old monolithic way of building software that executed whatever you needed.
But if I have 10 point solutions that now execute that, where do I go?
Tony Moores: Where do I start? Yeah and look, I the other trend. Maybe it’s not a separate trend, maybe. Maybe they go hand in hand. But I feel not only are there more point solutions, and I don’t mean any disrespect, there’s all, oh, you’re just a point solution.
Because as things get smaller right. But three opinionated people can probably decide where to go to dinner. A hundred opinionated people probably can’t, right? Yeah. And systems are increasingly becoming more opinionated. And I know, like the larger software houses out there, they used to say of computer associates, we built software the old fashioned way we acquire it.
They struggle. They, they go out and they take, they find these best of breed or best for them breed and try to put together suites and they have this collection of. In some cases in a very opinionated subsystems that they’re trying to create a system out of.
A long time ago I put pen to paper cuz I was, just drawing stuff up and thinking about, Hey, wouldn’t it be cool to design an API gateway that used machine learning to kinda watch what’s coming in and what’s coming out to, to watch the payloads to learn over time.
What this thing is doing rather than asking it, what it’s doing. And could we eventually, suppose how to do a new thing, by combining the different APIs that ran through this gateway. And, I get I kinda stopped at the time. I don’t think it was. Plausible to pursue that.
I won’t lie, I did have a Sarah con moment and like maybe I don’t want to be a part of that generative AI kinda stuff. But one of the things that I, I’m looking at it going, is it ever, would it ever really be possible to just observe these things from the outside and say, I.
Oh, I understand not only what you’re doing, but why and how you wanna do it as a subsystem and connect the bigger picture, which is what is the whole system really
Wally Ibrahim: trying to do. Yeah. Yeah. So I think there’s two really good points there that actually we’re, I think, core to our founding, right? One is that you hit on, which is.
How do I look at the entire system and not the subsystem? But the other one was the machine learning AI aspect. And that really came out of a frustration for me of seeing these same or similar integrations being built, reused not reused, I should say, should be reused, not reused, and built over and again, and customized over and over again, really in the same way.
Tony Moores: yeah. Like I’ve never seen two order records that were the same, and I’ve never seen two order records that were so different. They warranted being different. Yeah. You know what I mean? And yet we seem to integrate, especially like in the marketplace space where you know what’s important about an order.
It totally depends on whether you’re a buyer or a seller or an operator.
Wally Ibrahim: I think it, it even goes beyond that, right? It depends on your role. Are you in accounting? Because if you’re in accounting, you’re looking at things like cost of goods sold and line item level things. If you’re in marketing, you’re looking at different things.
So you know what is an order, right? An order means something different to a marketing person. An person in order management, a person in logistics, they’re looking for different things out of that order. So this is also why I think integrations is actually surprisingly hard, right? Because it seems really simple.
All right, one to one. I have an order point A. I wanna move it to point B. Okay, we started with an order. We defined conceptually what that is, but what is point B? Am I moving to a marketing system that’s going to analyze. That order and look at it in aggregate, look at a bunch of orders in aggregate and try to help me with something.
Or am I really concerned about it from an accounting perspective or am I looking at it from an inventory perspective, a shipping perspective. An order is not, an order is not an order across those systems, in fact, some of them may not even call it an order. And then, but even if you step back, even two systems that are competing with each other, as in two soft pieces of software, let’s call it two point solutions, that really execute the same function.
Are still really not one-to-one. And if they were, it would be commoditized and there really wouldn’t be a point in, buying one over the other wouldn’t really matter. You’d say who cares? It’s commoditized. But that’s not really the case. And that’s really almost never the case with software.
Outside of things like HTB servers and virtual machines on Amazon or whatever when you talk about business applications, they’re never commoditized. There’s no two e-commerce systems that are the same. So what offers an order is not an order is not an order.
Tony Moores: So obviously there’s insight in putting these things together.
And maybe at least today, those are beyond machine ntelligence or automation. So there, there’s some human touch there, but like what opportunities do you see in this integration space? For M L A I. Which of you at Jive kind of capitalized on? Is it just in the making it faster, more secure and less and less redundant work?
Or have you gone beyond that?
Wally Ibrahim: Yeah. All the above in a way. So I think there’s a lot of buzz around AI right now. Especially with ChatGPT, generative AI. And even, slightly before that, by a few months, there were things like starry AI and all the, stable diffusion, all of the image generation stuff.
So it’s all really impressive. Having said that, what’s it good for? And so I see a lot of companies approaching, when I say a lot of companies, mostly technology companies, approaching AI by saying, we’re gonna take our product and we’re gonna slap the letters g p T on the end. And that’s our new product.
And what are they really doing there? They’re just putting a new user interface on. So instead of clicking a button to track my shipment, I’m gonna ask the AI where the shipment is. Cool. That’s actually harder, right? I’d rather just click the track button outta my email than log in and talk to the AI about it.
I don’t need that. I don’t need a chat. AI there, but a lot of, but it’s
Tony Moores: very Star Trek, right? You can hear Montgomery Scott say, yeah, there’s a lot factor.
Wally Ibrahim: Yeah. Yeah. I don’t look at AI that way. I think AI is a tool. It’s a very generic tool. And so there’s something to be said there, but it’s a tool.
So the way I like to look at it is this, is that AI does not do away with any of the processes that we have put in place as an industry for years now, right? Whether it’s a software industry, whether it’s commerce, whatever it is, however you wanna apply to it, it doesn’t do away of those. All those processes are in place for a reason.
What we can do is look at that life cycle. Say, where can we apply this tool? And there are some ways you can apply it today that’s better. But not, so to give you a few examples of how we’ve applied it so for example, one way is just data management, data scrubbing, right? You’ve got a bunch of vendors that wanna, that you’re, that are feeding you data.
So obviously that just goes into plain integrations, right? You’re gonna sell a bunch of website. Yeah, translate. And they’re sending it in different formats. And that can change day to day. And one of the problems that happens when you’re in that scenario is that, you say, okay, I’m gonna, I’m gonna add a new attribute to my product catalog.
I’m gonna say gender. As one example. All right, so I add that attribute, and now I’ve broken every integration, or at least I’ve, in the, in the best case I’ve made, I added a new optional thing and I’ve an optional field. And now those vendors have to now consider whether they wanna use that field or not to populate whether their product is gender specific.
Tony Moores: So whether it’s Boolean or Freeform or liberated or, yeah, exactly.
Wally Ibrahim: Yeah. So when you have. Thousands of integrations and that’s not an exaggeration to say thousands. We dealt with it, and I’m sure you guys have seen it too, E-commerce, that’s normal. You that’s not manageable, right?
You can’t call everybody. And say, Hey, we have this new field. You gotta do this. You can’t, you can do all sorts of, little software management tricks and say, all right, here we go. Here’s an upcoming update, whatever update by now. But the way we look at it is say let’s just anticipate that’s gonna change.
And every one of those point solutions that is involved in those integration points may change at any point and involve new attributes. What we say is, Let’s use AI to map those, right? And our anticipation is always that there is a review process. So my advice to anybody looking at using AI today for anything is to say, look, use it, but don’t expect that you’re gonna use it and it’s gonna solve all your problems.
Always have somebody there. Real human reviewing it, which happens already, whether you’re using AI or not, you’re always gonna review something for you, a product for you published your website. So now we’re saying, yeah,
Tony Moores: That’s part of the, that’s part of the issue, right? There’s also the sense of wonder and magic that comes with it.
Look if you were operating a coal mine and you had a new coal miner, you’d probably wanna watch what that person was doing. Because they don’t know the operation, right? Yeah. We talk about machine learning and AI in terms of training and everything else, but at the end of the day, you’re still putting a new agent in the mix.
Wally Ibrahim: Exactly. And that creates, and it’s, to this day, as impressive as AI is, it’s not it’s not a trustworthy agent. You talk to these chat bots like Jack, G p T, it’s not a, and they don’t claim it’s a hundred percent accurate, not even close. In fact, when Open AI released Jack G p t was whatever, eight months ago now, I think they explicitly said we were surprised.
People liked it as much as they did, and, it’s not that accurate. There’s always gonna be a reviewer in place, at least for the foreseeable future. And so the way we look at it is, let’s look at the existing processes that have been in place for a while. Let’s structure those and let’s figure out where AI applies and let’s apply it there.
Maybe it’s data scrubbing, maybe it’s integrations, which are two things that I mentioned we use it for. So those are the ways that we look to use it. But also even in the world of integrations in terms of, a lot of, there’s been a lot of buzz lately about AI can write now, write code.
Okay, sure. We use that, but we review it. We look at it and go, all right, did it do what I wanted? Did it pass the test that we needed to pass in order to move this data accurately? So I think it’s really a matter of just, I think there’s a lot of magic around it and a lot of just people looking at it and saying, how do I use this?
And they’re trying to fit it in everywhere they can. And the obvious choices to say, put g PT on the end of the product title and and we’ll make a chat interface. But really, I think the way, and look there’s marketing. Yeah, it’s some wow factor there when you market it that way. But if you really wanna make it useful, I think look at your existing life cycle and say, where can I plug it in to automate some of the things that I’m doing now?
And now I’m just reviewing what it’s doing versus doing it manually.
Tony Moores: Isn’t it funny, 20, 20 years ago I was doing, machine learning and data science. Before data science was coined, right? And when people would ask me about AI and how that differed from machine learning, my, my canned response was, Hey, yeah, machine learning is the all the parts you know about AI that aren’t bullshit.
And now, There’s this reverse kind of trend where, you know, people who’ve been doing statistics and machine learning for a long old time are like, wow, let me slap AI or G P T in front of it and capture the buzz. Yeah. Funny stuff. Hey. Look, Wally, this is I love this conversation.
I’m enjoying it. I’d really love to invite you back and And also, invite our listeners to, to send us some feedback because there’s a lot more to discuss about when and when not to and how to use machine learning and AI and our systems and integrations and such. But I think we’re getting a little short on time.
Brett, how we doing with that?
RETISIO: Yeah. It’s probably good to wrap it up here. As he’s mentioned, we can. You guys, I’m sure I could keep going for a long time. So we would love to have Wally back.
Wally Ibrahim: But Wally, thank you very much for five been
RETISIO: conversation. Yeah. Thanks for joining us and being here.
It was very good conversation and thank you listeners as well for being here and for your support of the Unlocked Digital Commerce podcast. Be on the lookout for a new episode soon. In the meantime, be sure to check out rotiz.com for more resources and solutions to help you unlock digital commerce for your business.
Talk to you soon.