Engineering Transformation and Culture

Tullio Siragusa (00:12):

Hey, everyone. Welcome back to Tech Leaders Unplugged. I'm your host, Tullio Siragusa. I'm speaking today.. I'm getting unplugged today with Ebenezer Schubert, who is the VP of engineering at… who are you with? Tell us about your company a little bit.

Ebenezer Schubert (00:32):

So, I'm with a company named OutSystems. We are a low-code application platform company. We are actually the market leader in this space. We've been doing low-code before, low-code was a thing. So basically low-code platforms provide three big things, right? It provides a, it, it ultimately improves your productivity, right? Developer productivity significantly, where you can focus on building things that matter, right? Not at the nitty gritty details, but able to send build the applications or build value for the customers. And it provides three things. One is visual development. So, customer developers are actually working on the end product, not typing in the text that eventually might be visualized, but it's on a visual environment. The second thing is we do an integrated lifecycle. So, for every application that you build, you can see a feedback loop very quickly. That improves productivity significantly. And the last one is our hosting platform where we host applications that customers build on our platform and show them that we can host. And it's on a cloud-native architecture. So there is no limit to the type of applications that you can build.

Tullio Siragusa (01:56):

Excellent. All right. Well, let's dive into our topic Today. We're talking about engineering transformation and culture, specifically about adaptation, software-focused generative ai, cloud migration, and agile methodologies. I mean, we basically have mentioned all the transformational efforts that most organizations are going through right now. So let's dig a little bit about this in terms of, for example, generative ai. Everybody wants to do it. Everybody wants to implement it into the organization. But when it comes to the rubber meeting the road, what does that look like? How do you make that happen? And how do you balance that with other things that are going on like a cloud migration or implementing agile methodologies? I mean, my, most companies have a lot of these things in motion, so gen AI is like, let's make that happen. So in your opinion, what have you seen work best on balancing these things in terms of prioritizing transformation?

Ebenezer Schubert (02:52):

Yeah, so one thing, one thing that needs to be very clear when you're looking at generative AI is it's not going to be the answer to everything learned. Not every problem can be solved through the generative ai. It's very important to understand the type of use cases and the capabilities that you can solve with generative AI. And that's, that's very important. And the second one is, you know, AI is very easy to demo. I always say this, AI is very easy to demo, but it's very difficult to productize and get value out of it. And I, I, I think, you know, getting clarity on these two will help you drive the initiatives or make the business case for driving these initiatives very well.

Tullio Siragusa (03:32):

So let's talk about that a little bit. How do you go about prioritizing these things? What, what are some of the steps based on some of your experience?

Ebenezer Schubert (03:43):

So you know generative AI right now, if you look at ChatGPT and other interfaces, they're very textual in nature. Now, there are other generative AI models that could, you know, create visuals. And, you know, things like Dali can create a lot of visual things through generative ai, right? So, you need to understand which part of your system you are going to build with modification with generative ai, and it's important for you to build, whether it is a core competency, understand whether it is one of your core competencies or something that your business is going to have as something that's supporting, right? So, if it is a core competency, then you need to build the capabilities in-house, right? Whereas when it is supporting, you could partner a lot with a lot of these companies that are doing it through APIs and others, you can integrate it into your workflow, right? And it also depends on the maturity that you are in, in your digital transformation, because generative AI is going to come into play in different parts of your digital systems, whether it is selling, whether it is marketing, whether it is you know, customer support and contact places that you're going to use it in. So, it depends on the segment that you're going to apply. Of course, there is usage, myriad of usage across all of them, right? My view is some of these technologies are going to rapidly commoditize, and the differentiation is going to be on data. So anytime you're talking about AI transformation, it actually starts with getting good-quality data about the use case that you are attacking, right? So, you need to be able to get good-quality data that you can use to fine-tune these models for your use case and your system, right? Even if it's not fine-tuning, you are using the data to augment and protect these systems from going crazy. So, you can put all the grounding around it.

Tullio Siragusa (05:49):

So let's talk about this. In a real-world case scenario, a CEO of an organization hears about generative AI, it's an enterprise company. Maybe they're a manufacturer and think, “wow, we could gain some automation and some improvements to level up the team”. Let's put it in motion, right? CTO looks at it and says, we're just in the process of integrating all of our data. We're doing a cloud migration, and it's going to take us another six months to complete that. So how does one effectively communicate the realities of implementing a generative AI solution when the data is, you know, up to snuff yet? Like, how do you balance that out when you're getting the pressure perhaps from the top saying, go, make this happen, but the real, reality is completely different, right? So, talk to us a little bit about how you even manage through that.

Ebenezer Schubert (06:43):

No, this is, this is a very common scenario, right? Where CEOs attend the conference or read it in the news that everybody is doing it. So we need to do it right away. And this is where it is very important to understand the limits of it, right? I think about what it can do and what it cannot do. And today, generative AI is very good at understanding languages and being able to, you know, construct something around language. So, you look at, your problem space and see where you can improve from a language perspective. There, you may not be able to ha you don't have to find unit too much, right? You can use some kind of a human in the loop to ensure things are fine. So, a classic example would be a marketing group sending out you know, marketing emails, right? Or a support team sending out support responses. All right? This response, if, if you need to handcraft it for individual support cases and to write that response back, you're going to use templates that are going to be more standardized. It won't be customized, and it takes a lot of time for somebody to build that up. You could use generative AI to generate some of these things from very small templates very easily, and that might be a good starting point to understand the technology and its limits and what it can do, but at the same time, protecting it from having to deal with large amounts of data. But you can have some human in the middle who can look at it and say, okay, everything is kosher, but it just improves their productivity substantially, right? I think that's a good way to kind of put your toes in, get, get familiarity with the, what the technologies boundaries are right before you jump on it, right? And definitely working with partners and solution providers who can cater the solution to you would be another way of dealing with it. But when you come back to prioritization between it, in my view, data always proceeds AI without having data that is clean and good and cater to the use cases that you need. Jumping right on ai, unless the problem is very generic that you can have general data dealing with it, like language generation for example. It's, it needs to be customized for what you are doing.

Tullio Siragusa (09:07):

Well, at least you have to inform the system on where to, you know, the boundaries of where to get the data to formulate a response. So, here's what I'm, what I've heard so far, is you could break it down into site bite-size chunks, you know, work with the business units to identify where are those areas. We can have some immediate lift that doesn't require tremendous transformation or, or data sources alignment, but certainly do the work to get it to be more valuable. So, my question along the lines of this is, do you think that that also transforms the CTO's role a little bit or the CIO's role a little bit? Because now they got to get really close to the business lines and understand the specific challenges that they're facing at a much deeper level, because how could they develop some kind of generative AI solution without fully comprehending what customer success deals with, what sales and marketing deals with, what, what the fulfillment deals with? So what are you seeing in terms of you know, making that happen? What needs to happen there to create more connections and empower these CTOs or CIOs or otherwise to be effective at their job? Because it does require a deeper level of business integration, I would think, with those units. What do you, what are your thoughts?

Ebenezer Schubert (10:25):

No, I think that's definitely something that needs to happen, right? Because as things move more into digital, right? Where everything happens, every line of business will require significant support from the digital or technology teams, right? Whether they are embedded within the line of business is separate, but the understanding and the success requirements should be kind of very intertwined, between them, right? And it's, it's becoming more and more critical, right? I know that if you look at the last 10, 12 years, I think Lev in 2011 or so, Jeffrey Moore wrote this paper about systems of record was a system of engagement. While in the last 12 years, if you look at it, those have remained more theoretical, right? Where people said, okay, your entire ID system is going to be more tuned towards systems of engagement rather than putting a form in front of a system of record, right? So, if you, if you look at a lot of the enterprise architectures today, you have a core system of record and you, you put a form or some kind of a workflow on top of the system of record. So that dictates pretty much everything, right? But pretty is, we are seeing a lot of examples nowadays where it is engagement down rather than system record outwards, right? Where do you define your engagement with the customer or your end user, right? And that dictates everything else behind the scenes. Now, generative AI and other AI technologies have helped you frame that problem much better. By that I mean, you know, your system of engagement is a chat interface. It's not a natural form of how your data model is. It's not a web form that they click and fill. It's, it's going to be like a conversation that from that conversation, you extract information and fill and update your systems of records, or your engagement is a video call, right? From there, you extract information. So you are going to see more and more in the future things that are more driven by the engagement model, right? Rather than the system of record outwards and the companies that adapt to it are going to win. And as you said, if this is the paradigm we are going to move towards, the line of business is all about that engagement, right? And they, they are going to define all these engagements and if they are not doing the stuff that needs to be done in terms of technology and the interfaces that are required, it's going to become, the separation is going to be very hard to maintain it. They need to be much more intertwined.

Tullio Siragusa (13:06):

You made such an interesting point, and as a design thinking practitioner, I can't help but realize that really it's all about the experience. Whether it's the employee experience, the customer experience, the partners' experience, it's all about the experience they're going to have interacting with you as a company, interacting with the various systems. So you, you, you, you make an interesting case, these two need to come together. So in essence, the CTO's job becomes more of a designer of engagement models or experience models across the organization. Most are not equipped for that today because they're mostly technologists, right? Solving the technical problem. But this is more about solving the business problem. And yes, technology does solve the business problem, but, but this is a different mindset that needs to be adapted. From what I'm hearing from you, what are some of the things that people in that seat ought to be thinking about to level up their experience or their resume so that they can be supportive of this, of this kind of environment? What are your thoughts in terms of what gaps are in place right now that perhaps people need to learn to think differently? How can they do that?

Ebenezer Schubert (14:18):

I think they need to segment the technology in two pieces. And I think this is, this is happening in big SaaS companies a lot more. One is that platform, what I call the platform thinking, which is the infrastructure and capabilities that are needed at the backend, and then be able to understand that experience is a separate track by itself, right? And they move at extremely different paces. The underlying mechanics of how to do one versus the other is very different. For example, if you are building a cloud infrastructure that needs to be reliable, it can't go down all of that and it needs to fit into a cost envelope that you need to plan out. There are vendors involved. It's going to be a little bit longer term, much more predictable. It's the traditional engineering route if you will. But when you look at experience, right? Experience is all about experimentation. Figuring out which works, which, which type of workflow is a customer will more willing to adopt. You know, how can I com make it composable for different types of users, right? So those kinds of aspects are going to be, you know, closer to the customers and the line of business, if you will. And then you will see that adapt to the business outcomes that are needed, right? So one of the big things that will change is instead of treating these things as projects, they've become outcome driven, right? So, there is an outcome and a bunch of hypotheses to get to the outcome. So, your outcome might be revenue growth, or your outcome might be customer adoption, but you got to look at it as, and the projects that you're doing are just hypotheses or ways to get there, right? And you should have enough data to be able to make this distinction to switch as you, as you are developing. So, the teams at the front end of it have to be much more, it much more agile, more nuanced to the understanding of the customers, and be kind of co-located with the line of business or the function, right? And tied very closely to their outcomes.

Tullio Siragusa (16:31):

I heard a lot of lean canvas in there. I've heard a lot of wireless strategic mapping in there. So, let's talk a little bit about what types of skillsets or people the CTOs or CIOs need to surround themselves with to be effective at this. Because a lot of this isn't necessarily how well, you know, technology as it is how well you can design and put together solutions and from a hypothesis perspective, how well you can go wide and then go deep and validate and follow those kinds of cadences to get to an outcome. So, what kind of skillsets that are transferable perhaps into from other areas now need to be part of the CTO's organization that perhaps didn't have to be part of in the past? In your opinion, what do you think needs to happen there?

Ebenezer Schubert (17:22):

So, you're seeing this kind of converging in other groups, and like typically in software product companies, you're seeing the product managers kind of get closer to the engineering as close as possible. And some companies, they are either a two-in-a-box or even in the same organizational structure. So, you have product managers thinking about the outcome and the hypothesis around the outcomes and the projects that need to be delivered. You need to fill yourself with strong architects who can see this picture very well, right? Who straddle three things. You know, they have an insight into the business, they have the insight into technology somewhat around people. And in my view, I've found three in a box working very well. You have an engineering manager who has a good sense of project management and people you have an architect who kind of straddles technology and business. And then you have a product manager who can understand the market and sometimes you need to add a product designer to it who can bring the design aspects of it, a four-in-the-box, kind of working on a particular stream of solution. I've seen that to be working extremely well if they are stronger, you know, have mutual respect than what can work as a team. So you need to bring all of these things together into your organization to be able to deliver on these capabilities.

Ebenezer Schubert (18:43):

You know, what excites me about this Ebby is that as I hear you talking about this foreign inbox and the various folks with the different skillset that are needed to make, to be effective at, at all, this, it really does also require a cultural shift in the organization. You can't accomplish this in command and control structures. You just simply can't because it's not team-oriented. This isn't a one-person makes-decision kind of effort, it's a team effort and a collaboration. Would you agree? And how, what are some of the challenges around that? Especially for more of the old-school companies that might still have very hierarchical structures, you know, are they going to fail in terms of being able to collaborate around this kind of thing? Who's going to win based? Yeah, I know this, the folks, the folks that have these people assigned to it will win. But from a cultural alignment point of view, what changes need to happen there in your, in your opinion?

Ebenezer Schubert (19:36):

Yeah, so culturally it's a, it's a dramatic shift, right? I've been through multiple of these transformations. We've done the agile transformation, all of those. There's a way, there are two ways to do it. You do the standard way of collecting going, getting a consultant, do a top-down, have a big organization top down for enabling this, and they drive this initiative top down. What I've seen when that happens is teams tend to build resistance. You know, in software parlance, they try to treat this as an API. Okay, this is another set of things that I need to put check marks on, I'm going to do them. But fundamentally nothing changes in how I operate. So there's no cultural change, but there is a process change that happens. You know, the model that I've seen working, it's a little bit longer term, but I do see this as not a short-term change. It's a shift in how we are going to operate. So, it's worthwhile investing, investing time in the longer-term view of things, right? You take a cohort of a small set of groups, you make them evangelists, okay? So that they fundamentally see, see the impact of the change that is happening, right? Where you are able to work with them to make sure that they see the effects of their change, right? And solve real problems. Because a lot of times we take cookie-cutter approaches and drive it down without a full understanding of the problems that the specific teams are going through, right? Because there is a difference between reading a book and saying, yeah, this makes sense. I'm going to adopt it without understanding fundamentally why we are doing it, right? So I can give you a few examples of projects that start without understanding why we are doing it and then fail miserably, right?

Tullio Siragusa (21:27):

It's… it sounds interesting that you know, I'm, I'm going back to when shared services started many years ago, and a lot of companies have to learn how to leverage that in order to maximize outcomes. But this is deeper than that. This is really if you're a functionally driven organization, you still have somewhat silos, right? This requires cross-functional thinking and cross-functional teams and the, the authority needs to be shared across those functional teams in order to be effective. And ultimately you don't really have one decision maker. It is a collaboration, for these things to work. And I you know, that's a big challenge, you know? So, let's go back to the first conversation about… here's a CEO reads about [INAUDIBLE]. AI gets excited about it, let's go make this happen, right? So, what is a good CTO could say to push back, well, you know, we need going to need, we're going to need the X, Y, Z and we're going to need to change the operational structure. How do you navigate that? Because that's a big shift. Well, what we, based on what we just talked about.

Ebenezer Schubert (22:34):

So, so, you know, you can always do it in smaller pieces and get familiarity with it before you roll it. I'm a big believer in it. Get this, get the understanding of the problem extremely well before you saw to it, because people just jump in without understanding the problem in full, right? Going back to the previous one, one of the things that, for example, I'm going to give you an example of things that have fundamentally shifted in business that requires you to alter, right? So some of these practices are evolving because the feedback loop with the customers is shortening, right? So I used to work in hardware companies that built large hardware systems. So, your… the time you envision the project to the time you can get feedback is like two, two and a half years out, right? You build the hardware platform that takes a year to build, and then you release it, and then it takes another year to adopt, and then you start seeing feedback flow in. So in this cycle, you need to be extremely good at predicting what that will look like, right? Meaning if you don't get that right, you've wasted two and a half years and then you're out of the market,

Tullio Siragusa (23:45):

Right.

Ebenezer Schubert (23:46):

In the software world, in the cloud-based SaaS kind of software world, you release it and immediately all your customers get it right? And you are the one operating it. So you have good ways to collect data. So the feedback loop is extremely short. Right? Now, imagine if your development cycles involve multiple handoffs, right? You are handing off from a person defining the requirements to development, to testing, to release, and then you have customer success, and you get feedback and then do it. Whereas you could do it a lot faster, right? Because you release and you get feedback. So, you can, you can iterate that feedback very well faster. And this fundamentally has to shift everywhere else, right? This is kind of the transformation that I was talking about unless you know the nuance of why you are transforming, right? You transform for no reason, right? And then you say that this is the modern way of doing it, so we all should invest in it because you have to see the applicability of that in your business, right? So when you talk about generative ai, it's first and foremost important for you to understand its applicability of it to your business and what your business fundamentals look like. Otherwise, you get into this mismatch of everybody is doing it, so I got to do it, and I don't know why I got to do it, but I got to do it, right?

Tullio Siragusa (25:15):

It sounds as though you need to fall in love with the problem first. Yes. And then do a lot more listening and listening and taking notes and seeing how that matches up to solving the problem. And then of course, we've also talked about caring about the kind of experience you create along the way for everybody. And all of it has to be tied to some purpose or outcome that has been clearly defined that ties back to solving that problem. So, you know it, it's very, it's been very, it's been great to have you as a guest today. You've given us a lot to think about. You know yeah, you want to adapt generative ai, you want to do the latest and greatest, but is it really going to help you? Do you understand what you're solving? What's the purpose? What's the outcome that's going to come to that? And do you have a culture in place that allows for collaboration to make these kinds of things happen? The shared authority to make these things happen is so much to consider, but it's definitely a brave new world that's being introduced, and it's going to challenge a lot of organizations to be more collaborative, which is actually good. It'll be good for the organization. Thanks for being with us today, Ebe. It's been great. Stay with me as we go off there in just a second. And we just announce what we got coming up tomorrow. We've got Reza Raul, who's the CTO of Real Networks Wednesday we're going to speak with Preston Greeting, who's the founder of PAC Form. So looking forward to seeing you all back tomorrow. Same time, same channel on Tech Leaders Unplugged. Take care, everyone.