Join us weekly to discuss the latest and greatest in low-code and digital process automation with executives and experts. Real conversations, no marketing BS. Hosted by Rob Koplowitz, John Rymer, and Ryan Duguid. Visit analysis.tech to get in touch about your personal low-code journey and learn about ways we can help.
John Rymer (00:00)
Hi everyone, welcome to the Low-Down by Analysis.Tech I'm one of your hosts for today's episode, John Rymer, of Analysis.Tech. And I'm here with my partner, friend, baseball buddy, et cetera, Rob Koplowitz. Rob, how are you today?
Rob Koplowitz (00:17)
I'm good, John. Thanks for ⁓ thanks for getting all this pulled together and and we get another friend with us today. We've got we've got Burley an old Burley Kawasaki. Those of you who are real low down aficionados will remember his one as either our pff second or third guest. He's back.
Burley Kawasaki (00:35)
Yeah, I I'm glad to be back. Thanks. Thank you for having me.
Rob Koplowitz (00:38)
Glad to have you.
John Rymer (00:39)
And Burley, you are now VP of Product at Creatio, which you were not the first time we had you on on the podcast. So
Burley Kawasaki (00:48)
I was not. Yeah. I although ⁓ you know, funny story, I I had actually so the first time I was doing some independent consulting and Craycia was actually a client of mine and that led to a project at the time I thought it was just a project and you move on, like a lot of consulting projects were, but I I had the chance to write the Note Code playbook with Creatio and I had such a great time working with their leadership team and we saw, you know, so much common ground on the topics of No Code that that actually led to me joining Creatio. So
Happy to be b back. ⁓ and I you know, I I cover a lot of areas as product does around product product marketing, overall, you know, positioning and thought leadership. And then one of the newest areas I've taken on is around our vertical industry strategy as well. so I am involved in in how do we take sort of a you know, an an AI platform and a code platform and then verticalize it across our top industry.
John Rymer (01:42)
Right, right. Such an important topic, Burley. I I mean people want solutions. And and so for a platform company, that's a bit of a that's a bit of jujitsu that you have to do. So very, very, ⁓ very, very interested in how you guys how you guys do that, how how you guys fare with that strategy. Burley, since we usually start with ⁓ tell us about your journey, but I think you just told us about the latest of your journey. ⁓
Burley Kawasaki (01:44)
Mm. Mm-hmm.
Mm-hmm, they do. Yeah.
Mm-hmm.
Yeah.
John Rymer (02:12)
And you told us about your journey at Microsoft for many years and and Avenon and so forth the prior. So maybe, Rob, we should just cut to the chase and get into
Rob Koplowitz (02:20)
Yeah, everybody go back and watch that first episode with Burley. We'll save you a little time this morning.
Burley Kawasaki (02:22)
Yeah, yeah. Yeah. Yeah. Yeah. At least have your agent go back and read the transcript. Yeah. ⁓
John Rymer (02:23)
Please
Yeah, there you go. There you go.
Rob Koplowitz (02:29)
Yeah.
John Rymer (02:33)
Yes.
Rob Koplowitz (02:34)
All right. So Burley, let's let's jump in then and talk about well, the last time you're on, we talked about low code and ⁓ and the world has shifted a little bit, but not all that much, right? So we've talked to ⁓ gosh, I I mean I really don't think we've missed any of the prominent low code vendors in the show over the course of time. we've been waiting to have Creatio on. Everybody has added some some ⁓ approach around AI assisted development, you know, some kind of you know
Burley Kawasaki (02:41)
Mm-hmm.
It has. Yeah.
Rob Koplowitz (03:04)
vibe coding front end onto the low code platform, all low code vendors are moving in that direction. We think the combination of a platform and you know a vibe coding approach is actually a really compelling set of capabilities. ⁓ but they're not all the same. So tell us how is Creatio thinking about this? How is Creatio approaching this?
Burley Kawasaki (03:25)
Yeah. Yeah, I I you know, if I look back again, you know, two years ago when when we talked, it you know, at the time, like you said, it was just sort of emerging and and so I I I think a lot of the market was was seeing it as sort of you know an add-on to sort of low code, right? Every vendor, like you said, sort of you know, rushed to get some feature out there. But I but I think at least the way we view it is this is much more of a structural change in the way that people think about developing software, right? And and so
It's not really an option anymore. It's it is it is the way you build. And ⁓ you may be still using visual tools part of the time, but but now you're using natural language part of the time. But it's almost inseparable. We we think about how you know AI and and tools work together as part of the development process. and and I think the best ⁓ sort of experiences are when you you do support the use of both, right? Because there are some times when
you know, a AI can generate much faster, but then there's a lot of times where it's still ⁓ is is powerful to be able to visually show this to your stakeholders or to be able to make small fine tuned adjustments using visual tools. So so again, it is just one and the same now. In some ways you can see AI as being the purest form of low code or no code, right? Because you're you're not writing code and then you have your choice of of whether and how much you want to use tools versus natural language.
Rob Koplowitz (04:51)
Great. So we're gonna go a lot deeper into all of that, Burley. So John, over to you.
Burley Kawasaki (04:55)
Mm-hmm.
John Rymer (04:57)
So, Burley, Ratio has ha you've incorporated AI into the platform for some time now for a number of releases. What kinds of of scenarios, what kinds of use cases are you seeing people really lean on on the AI features to deliver?
Burley Kawasaki (05:04)
Mm-hmm. Mm-hmm.
Yeah. Yeah, we we we actually we are adding on AI right about the time I joined. And so we we've we've now had the opportunity to sort of watch our customers adopt and to deploy AI now for about two years. ⁓ and in particular and I granted some of my com my comments will be sort of a little bit biased towards the enterprise and also, you know, ⁓ enterprises and regulated industries. So we we do a lot of work in banking, financial services, ⁓
manufacturing, public sector, these are are ⁓ very sort of complex sort of enterprise ⁓ industries, right? They and what we're seeing is that you know, as powerful as AI is, there's still the workflow, the process is still messy, right? I mean if you look at the way these enterprises perform their work, there's a lot of sort of human and manual touches that I I won't say will never go away, but it they're they're not going away anytime in the first
foreseeable near future, right? And which I think is a bit different than, if I'm a startup, I can sort of, you know, run entirely on an agent or two. And you know, but but a bank or a a a a government or ⁓ a manufacturer, right? They they have a lot of important sort of exception processes, manual processes that AI can help with, but are are probably not going to completely replace again in the in the near future. And and so what that means is we've sort of think about
this and I'll I'll use a specific example from ⁓ you know from one of our industries from banking, but we think about it as the you know as sort of the fusion of both, you know, human led processes and agentic processes together. And you you can't have just an agentic led process. You know, is you know, customers coming into your branch if you're a bank or customers that are going to be visiting something on the factory floor, right? It it's
it's going to have to ⁓ support the the broadest possible process. And and so that's that's sort of the mindset. And when we think about automation, it's about automating sort of the end to end process flow, ⁓ not just the agentic parts of it. ⁓ onboarding, I'll use yeah, go ahead.
John Rymer (07:27)
So does that I'm sorry, does that mean ⁓ that the A that folks are using the AI ⁓ tools for prototyping, for example, or or the front you know, the front end to sort of prove out a concept, but then there's there's more they're using the other tools, more traditional tools on the platform to actually deliver? Is that is that
Burley Kawasaki (07:50)
No, it's it it they're using it it's not about sort of prototyping versus real deployments. ⁓ but but it's even in the sort of full operational processes. There's parts like ⁓ in in banking, for example, ⁓ like like many industries, you you've seen a lot of ⁓ sort of customer self service, whether it be chat bots or you know, even voice bots now become very popular. But there's in banking there's so much ⁓ regulation still about what kinds of questions that
y you can't answer or not. I I I just saw ⁓ a study one of the the banking ⁓ publications had done that talked about the liability of banks who, you know, if if the chatbot responds incorrectly, right? What what does that mean? And there's fines, there's penalties. And so you have to you have to just ⁓ you know combine, I would say, the the power of AI, but still in the context of there is going to be important processes. Sometimes it's
it's regulatory based, but a lot of times it's it's maybe because it's more relationship based, right? You trust your banker, you you know, you don't want to entirely ⁓ remove the human element, right? And so it's it's about finding the way to bring those processes together. ⁓ I was I was gonna use onboarding as an example. If you think about like onboarding a new customer for some sort of like ⁓ you know wealth management or or or small business, right? There's there's a lot of things that AI can do really well, like taking all the
the you know ⁓ articles of incorporation or a lot of the the the documents that you would typically have to provide and a lot of those can be automated more and more by AI to automatically you know reason over documents, extract information, remove data entry. But then there's still so much about the relationship that you have with you know with the customer. ⁓ and there's still important parts of ⁓ just how how how some parts of the process have to perform in a bank because of risk or compliance.
that you you're never going to remove the human touch, but it's about ⁓ enabling them and and having visibility so that every time your, you know, let's say your relationship manager ⁓ has a discussion with you in the branch, you know, on the phone, wherever, they always have full visibility to what AI has been doing as sort of a coworker or, you know, teammate on behalf to speed up the process, but but not replacing the the human who's still responsible for that relationship and responsible for you know everything that happens.
John Rymer (10:14)
Thank you. That's really cool. Let's take a quick commercial break, folks. You're listening and watching. If you're watching, the Low-Down by Analysis.tech, a mostly weekly podcast brought to you by your friends at analysis.tech. Rob and I are two of the ⁓ partners at ⁓ analysis.tech. We do a lot of of consulting and advisory work. We create a lot of content. All that content is available on an analysis.tech, backstroke research.
Burley Kawasaki (10:16)
Mm.
John Rymer (10:44)
for free. ⁓ we also have a sub stack where you can see our latest ⁓ our latest ⁓ research and ⁓ and analysis ⁓ and like i say we're about we're here just about every week hit the like button if you like what you're hearing hit the subscribe button if you don't want to miss any of our our episodes and with that rob let's return to our conversation with burley
Rob Koplowitz (11:09)
Let's go right back to where you left off there, Burley, because I think I think what I started off by saying we we actually find the combination of what used to be called a low-code platform ⁓ and ⁓ an AI-assisted development, as well as AI services, right? We kind of intermingle the the the two terms. We shouldn't, right? There's one thing is is having AI help you develop an application. One is a particular service that actually can make a process better. ⁓ but you know the thing about
Burley Kawasaki (11:12)
Mm-hmm.
Mm-hmm.
Rob Koplowitz (11:39)
The thing about low-code platforms, whatever we want to call them, you've got a platform, you've got a model, you've got a control plane. Not quite sure where that term fits in the model versus the governance, et cetera. But it but you are deploying things in a repeatable manner so that they can be that they can be safely deployed in an enterprise. You keep bringing up banking, manufacturing. These are mission critical environments and possibly highly regulated environments. ⁓
Burley Kawasaki (11:47)
Mm-hmm.
Rob Koplowitz (12:08)
And there's a very there's a lot of interesting reasons to to to choose a platform that allows me to scale my AI development efforts, but do it into something with a with a control plane. Can you walk us through how that would work in the Creatio platform?
Burley Kawasaki (12:26)
Yeah. Well, and and I think, you know, it's it's a super important topic. ⁓ you know, not only for sort of highly regulated industries, but I would say every enterprise is starting to worry about, okay, you know, in in the in the race to sort of keep ⁓ up with the AI innovation pace, like every every morning you wake up and there's a new ⁓ startup that's created some new A AI capability. And so it's it's leading very much to the same AI sprawl that we saw when SaaS first became popular.
where everyone can and and did sort of start to turn on different SAS point solutions. Now if we fast forward even, you know, six months, twelve months, right, I I I believe that it's unsustainable, it's untenable to have that many sort of point solutions. And so the sort of two things that we think about as important pieces of of you know solving that problem, especially when you're in in regulated environments. One is taking a more of a platform led approach, right? You you can't have, you know, forty two different
sort of development tools and frameworks and and so there will be consolidation at a platform level we believe maybe it's not one, right? But fewer than we have today, right? And I I think you you've always seen that in development technologies that, you know, there's this in innovation cycle where there's lots of vendors and then there's you know consolidation. And and so that's why we've been investing so heavily in a platform approach that gives you ⁓ sort of a common set of tools, common set of models, et cetera. But but you're right, I think in above in addition to that
This notion of a control plane, I think, is becoming more and more compelling, right? Because e even though you may be standardizing on fewer platforms, you still want to have visibility control across other agents that may not have been built on your platform, right? And and so you want the ability to support, especially with the the the you know, the most popular Horizon ⁓ you know, AI providers with anthropic and open AI, like you need to be able to
coexist and and govern and control, you know, them because that you know, that is exploding very rapidly. So so we at Cratio have taken this approach where we provide ⁓ a both that platform and the control plane essentially is what we call ⁓ Cratio AI Studio. And and so it is both a development platform that helps you ⁓ build rapidly using no code. ⁓ but you it also has a control plane at the same time that manages not just our agents, but you can register and link other
agents anthropic or or open AI the most common, but also ⁓ you know we support others, especially using MCP. You can just register your agents and we'll give you visibility. and and I'll I'll highlight that you know this we we see in some of our early adopted very strong correlation between ⁓ the investment in the control plane and governance and then sort of the confidence level. ⁓ and th this would make sense. But we we did a a survey recently we we surveyed ⁓ over six hundred
sort of business and IT leaders worldwide. And when you looked at the ones that had established a control plane and governance, ⁓ there there was a high correlation. It was like seventy-one percent of those that had deployed the governance layer were now much more confident about their ability to scale and roll out AI versus those that were still, you know, less mature in terms of of those practices.
Rob Koplowitz (15:41)
It make it makes perfect sense. I want to just do one quick follow-up, Burley, because you've hit on this in the context of a couple of other answers here, but I want to make sure this really lands for our audience because I think it's I think it's really important. An end-to-end sophisticated process. I mean, remember when you were BPM online? I mean, you guys sort of you guys go back, your roots are in long-running, sophisticated processes. And
Burley Kawasaki (15:46)
Mm-hmm.
Mm-hmm, mm-hmm.
Process. Correct.
Rob Koplowitz (16:07)
And they will involve a lot of different types of tasks over the course of that. And some of those will be human-based and some of those will be integration-based and some of those will be deterministic. Most of them used to be deterministic. Increasingly, they and by the way, we've always had AI services injected into those orchestration engines, but ⁓ but boy, there's a lot of opportunity. We're not going to go end-to-end agents, but specific agents at specific parts of the process that are highly controlled could be extraordinarily valuable.
Burley Kawasaki (16:23)
Right. Correct.
Rob Koplowitz (16:37)
Can you talk just just land that for our audience about how important a a model driven archest architecture with an orchestration engine is in this emerging world?
Burley Kawasaki (16:48)
Well, yeah, it's ⁓ it's so critical because I think, you know, we we've all seen that with, you know, different coding agents, coding tools, you can prototype very rapidly and you can you can build code, you know. So ⁓ so people think though, they they can fuse between being able to build a prototype and then being able to run ⁓ something operationally at scale, right, with the right levels of control and governance and is especially as we see, you know, all of our our customers same study I mentioned has shown that
you know, we're we're now well beyond sort of the initial proof of concepts and tire kicking, like organizations even in regulated industries, are deploying it. And when you do, you realize that that prototype I built y using Replit or, you know, clawed code in, you know, ten minutes, now you have to have a lot more around it in terms of the harness or in terms of the the governance to really make it r you know, a actually run in your in your environment. And so th I
I believe in and of course I have a bias a little bit, but is the more ⁓ sort of transparency you have to both the workflow as well as the model, right, allows you to to have much more control and governance over that. And and so if you've used these coding agents and they've built all these sort of blocks of opaque code and that you're now trying to run or deploy, fundamentally you're you're always going to be sort of, you know, only as good as what you can observe around it.
Right, the generated code versus the investment in in sort of a more declarative approach, ⁓ for both your workflows as well as the rest of the model, means that you can enforce control all the way up to the very beginning, right, of the design process, and that you can actually guide the the choices you make during at design time differently to ensure higher levels of control or governance or smarter decisions, right? You can guide them to say, well, look, maybe that's you know, you're using a really expensive model, right? This advanced reasoning.
You could use something small and, you know, efficient and by the way, much more cost effective. Right. And and so these are the things that rather than waiting till you after the fact, after someone has built these, you know, sort of shift left all the way through and I believe models and sort of more declarative ⁓ d development approaches really enable that in a way that you can't with sort of traditional ⁓ pure code.
Rob Koplowitz (19:01)
And Berlie, I'd throw in one more thought before just passing back to John. You can manage and reconfigure your end-to-end process through a model, not through code. John, over to you.
Burley Kawasaki (19:07)
Mm-hmm, mm-hmm.
John Rymer (19:10)
Right. Yeah, Burley, very interesting discussion there. ⁓ I have begun to think a lot of people th when they hear the word they hear the term control plane, think it's after the fact management, as you mentioned, after the fact. It's real it really extends into the development activity. I've begun to think of it as as both. ⁓ it's full life cycle, ⁓ you know, in other words. So ⁓
Burley Kawasaki (19:28)
Mm-hmm, mm-hmm.
And it just John you b both you and and Rob will appreciate this, but if you think back to when we were sort of either at Microsoft or or following Microsoft, you guys both spent a lot of time around SharePoint, right? SharePoint was the classic example where, you know, there was not enough structure enforced during the building of a lot of these SharePoint sites and so after the fact became hugely difficult to govern and control, right? And and so I I think there are plenty of examples like that.
John Rymer (19:57)
Yeah. Yeah.
Burley Kawasaki (20:01)
you I just pick on that one, but it it has the same ⁓ potential to get really messy sometime down the road, even in six months, if you haven't, you know, really ⁓ enforced those at design time.
John Rymer (20:12)
Yeah, good great point. So Burley, you've been at it for Cratios been at it for a couple of years. ⁓ what do you find are the are the big changes that customers need to make in order to have success, in order to take advantage of the the facilities that you guys are putting out there? Maybe top three that you've observed, top three changes?
Burley Kawasaki (20:18)
Mm-hmm.
Mm-hmm. Top three. Yeah. ⁓ so I'd say one of the changes we're seeing is, you know, ⁓ a lot of you know, AI development started in sort of centralized innovation teams, right? And whether it be, you know, in IT or under the, you know, chief AI officer, but they were, you know, th the these were more ⁓ I call it experienced developers. Maybe they're using code, maybe they weren't using code, but they they still understand sort of you know, ⁓ understood a lot.
⁓ of the design trade-offs and characteristics of how you build an agent, for example. ⁓ but the problem is there's only you know so many of those individuals inside a typical organization and that in some ways becomes a bottleneck, right? because there's only so many of these agents and go through that sort of centralized process. And so I'd say one of the most important changes that we are starting to see is a push towards more decentralized ⁓ ideation and development. Now
Normally you would you would say, well, you know, that's again shadow AI run amuck, and you would worry about all these people that ⁓ are are building AI using ⁓ again ChatGPT or they're using Replit or other tools. But this is an area where we are we are actually trying to embrace ⁓ sort of empowering the end user with something that we call AI twin. And so AI twin is our you know is our way to sort of help give the user the ability to vibecode and to build agents. ⁓
But it's still based on the same governed and managed platform. So there's not this, I have to, you know, throw it away later on. This is still fully managed. It's built on the same sort of foundational components and the same platform as my my enterprise agents are. And so whether it is top down from an innovation team or bottom up from you know just users that have a great idea and are starting to build, right? You don't have to manage or govern them separately. They can be they are part of the same governed platform. And so
it speeds up the rate of adoption. We've we've seen some of our our customers already that you know the number of agents they're building now ⁓ is is just off the charts. 10x or more development ⁓ versus the ones that are are sort of in central IT. Because IT, again, has become sort of a bottleneck, not because it takes you that long to to build anymore, AI is accelerating that, but but the skill sets and knowledge are still immature, right? And and so we're seeing that ⁓ to be important.
The the second sort of shift I would say is ⁓ it's related to that sort of bottleneck concept, which is ⁓ as as building an R and D, you know, is is becoming less and less of the bottleneck, but it it sh becomes more important now to know ⁓ what to build. Right. And you can you can build the wrong thing much more quickly now than you used to. You you and so ⁓ just from a like a a a prioritization and a sort of a a lens on you know business case and
project management, product management, that becomes really important to identify what to build. ⁓ and so we're we're investing a lot ⁓ in in sort of those early upfront stages because I think ⁓ you you're seeing in this race that to to just sort of keep up with the pace of innovation, a lot of organizations are sort of short shortening and and maybe taking ⁓ shortcuts on on sort of some of that upfront planning. And I I still believe that planning is essential if you're going to build the right thing, not simply the wrong thing ten times as fast.
John Rymer (23:57)
Cool. Number three.
Burley Kawasaki (23:57)
⁓ I think number three is around just this ⁓ ongoing iteration. You mentioned this around how the you know the importance of a model driven AI approach is that you can change more rapidly. ⁓ I think everyone is sort of still experimenting and and so you have to embrace that you're you you don't know the right ⁓ way to build things in most cases, but it's still important to learn. And so it it's back to sort of this, you know, get it out there, use feedback.
iterate, adapt. And I think that's where the model driven approach to to development still is essential, right? Because you c you can change and adjust it much more quickly.
John Rymer (24:36)
Awesome. Rob, we should probably saw it off, huh?
Rob Koplowitz (24:39)
⁓ well let's do one more, John. I think let's can audience can you stick with us for another ninety seconds? I want to do my best Larry King impression. Broy! Platforms cost too much and they're and the licensing is too rigid. Respond
Burley Kawasaki (24:45)
Mm-hmm. I'll I'll be brief, yeah.
Mm.
Yeah, yeah. Well, you know, the people have talked about the SAS pocalypse and, you know, I I I don't believe that SAS is going away completely tomorrow, but I but I think what we're seeing is that the user based licensing model is you know, time is is running out, right? Because of I think that in particular doesn't really ⁓ support this new model of agents everywhere that, you know, y you may have all sorts of parts of your solution that are not really user ⁓ sort of based. And and so we've
we've actually moved to a new unlimited platform licensing model. And so we w we will still offer pre user licensing for ⁓ customers that are on it today, but we're we're trying to encourage moving to more of a consumption based model. And so with an unlimited platform, there is no user license. You don't have to pay per agent per workflow per API call. There's none of that ⁓ sort of and and so it really opens up the the discussion around, okay, if everyone in the organization can take advantage of the platform and everyone can simply ⁓
you know, innovate with whatever function or department, that that just creates a a broader adoption than than I think traditional SaaS price. ⁓ we've already seen that expanding our deployments much more rapidly. ⁓ so that that is the future. ⁓ it's consumption based now, so you'll once you have access to the license, there's still a cost for tokens and things like that. But but in terms of the accessing it per seed or per agent or per API, we we have no limits ⁓ to to what you use.
Rob Koplowitz (26:23)
L a lot of thoughtful things have been going on at Creatio. And you guys, you guys tend to think about these things a little ahead of time. So thank you everybody for joining us. ⁓ listen, go find out more about Creatio at Creatio.com, C-R-E-A-T-I-O.com. We're always happy to have visitors to us at analysis.tech. Thank you, John, for joining today. And Burley, thank you for coming back and spending more time with us. I would have thought you would have after your first trip, you would have said, You've had enough of us, but you're back.
Burley Kawasaki (26:32)
Thank you.
Absolutely.
No, always happy to be back. Thank you, John. Thank you, Rob. Mm-hmm. Thank you.
John Rymer (26:52)
Thank you for the
Rob Koplowitz (26:54)
Alrighty. Alright. Thanks everybody.