We welcome Ioan Iacob, the Founder & CEO of FLOWX.ai to Unbounded Talks to share his knowledge on unlocking technology in an era of spiraling software costs and complexity.
Software has become an ingrained part of not just business - but the very fabric of human life on earth today, and has been fueling a radical transformation of our society. Supporting the exponentially growing scale and complexity of today's businesses.
Between legacy technology challenges of scalability and extensibility, inflexible closed solutions, security threats, and the scarcity of talent - sustainability of software development has become a topic of growing concern.
Enabling better software is a vital topic for our civilization, as it has a profound impact beyond just technology.
--------
Today, we are talking to the man himself. We are talking to Ioan Iacob. That's right. He's an engineer, an entrepreneur, and on a mission to make software better. That's right.
He is the CEO of none other than FlowX.ai. So get ready to dig in. We're going to talk big picture about what's going on in the software, and we're going to give you a ton of clues on how you might make the most of those opportunities in this crazy fast-changing world. So let's get ready to jump into the growth equation.
Ioan. Welcome to the show.
Mike, thanks for having me. It's so great to speak again.
It is great to have you back because since we last talked, a lot has happened; I feel like we have to try and process. If you think about the massive change in tech, particularly big tech, many things have happened.
Then when you flip over to AI, something that you and I discussed in our last show, oh my gosh, here we are in February of 2023, and everybody's talking about it. Who would've thought that people are pre-registering for Bing? Let me remind you, Bing, things have changed very quickly. So what do you think about this change over the last year?
I think we're looking at this exponential rate, the pace of change. And I think that what's interesting about exponential trends is that the human mind isn't trained to understand exponential trends. So you can understand them rationally, but you don't feel it.
It lingers at the bottom of the chart and then takes off. And this is what's been happening. It's no longer than the summer of last year when we talked about AI-assisted development and every single executive you know and technologists, right?
Like CTOs, CIOs oh, come on. That's not a thing. That's just some marketing stuff. It's not happening. And now everybody's faced with the brutal reality of ChatGPT being able to write pretty good code actually. Pretty good functional code. It's in your face; you can't ignore it anymore.
And the exciting thing, Ioan, is some of the models they're using at ChatGPT. We're based on papers written by guys at Google, yet it's got Google in all of a fluster. It just shows you that nobody can afford to rest on their laurels. So even Google has to be ready for what's next.
I think we are looking at a couple of years that will be insanely transformational for the entire industry. And we better be ready for that.
So you've spent the last several years really getting after it. You're right to call it out.
A year ago, we talked about the role of AI, and what struck me when we were preparing for this show was how crazy things had got in the world of software, not just with the AI, but how people approach tech today and what they're working with in terms of the current environment of trying to find good developers and trying to deal with tech, debt and legacy systems. What is the greatest battle for people trying to build software and products now that we've got another vector with AI-powered chatbots? That's just another challenge on the CTO's desk.
So what are you seeing when you travel the world talking to folks? What are they stuck with? What's causing them sleeplessness?
I think it's the same symptoms, much more accentuated, that we're seeing, and it's the costs of software development and cost of IT just spiraling out of control. And again it's a function of the same exponential trajectory.
And we've been building this pyramid scheme of software engineering for quite a while now. If you look back, historically, we used to have a bunch of really good software engineers as a civilization. They went through really hardcore schools, and then we started addressing the problem of delivering software with volumes. Bringing on more headcount and not necessarily looking at people who can be engineers. And the problem with that is, beyond the philosophical, let's say, an oxymoron happening there because software is fundamentally a high-leverage area. So we're trying to build software using more brute force.
But what, what's been happening is people. That we're not properly prepared to be engineers have created more problems in software. So if you bring a hundred engineers, out of which maybe you have ten real engineers who are good at building enterprise systems.
And then 90 people are good at writing code but they are not really good at thinking in terms of architectures and enterprise complexity and scalability and foreseeing things. So they create problems in the code and then you end up with a piece of software and bring another thousand engineers, out of which probably 20 are really good engineers.
So it's this pyramid scheme that's been happening in particular in enterprise. And I mentioned enterprise because I think that's where scalability and security are mostly required. And the level at which companies have been growing, it's something that technology hasn't foreseen back in the nineties or 2000’s. So the complexity and the scale at which software is being used right now is something we haven't been preparing for as a civilization.
Do you think large enterprises have been throwing so many engineers at the problem? It's that famous mythical man month. Throw more people at it. This brute force just means that if you take the law of averages, that really more and more bad code is going into the system. So invariably, the enterprise is creating sort of a house of cards.
It's all band-aids here and there. Is that a way to interpret what you're saying?
That's exactly the image. And then the more bad code goes in, the more engineers we hire, and quote-unquote engineers.
And what we're learning is that very large organizations, and not just very large organizations but humankind generally doesn't have a good way of measuring engineering quality.
Software is not measured by quality. It's measured by quantity. It's about headcount. It's about the number of people, which, in all honesty, doesn't mean anything.
How many lines of code, these kinds of concepts.
Rather, it's pure headcount, that's the main metric in software.
It's just a headcount. And I think Steve Jobs was the first one that was screaming at the top of his lungs about exactly this. That one good, one great engineer is not worth two or three. It's worth a hundred x. By definition, there are many more average engineers than great engineers.
Exactly.
If it was unanimously accepted, in2000s, 2010, that 80%, 85% of the code being written is reworked code. I would say that right now 95 to 98% of the code being written is reworked. So there's very little code being written that's adding value.
Most of the code being written is patchwork and not great patchwork at that, too, to be fairly honest.
Your brain fails to comprehend when you have bad fixes. On top of bad fixes or bag code upon bag code to fix the bad code. And we all know that when you sneak in a quick bandaid to fix this thing, often something else on the other side breaks the unforeseen.
Exactly. And then what executives in banking are seeing, and not just in banking, but also enterprise in general. What they're seeing, and they're incredibly frustrated, is that they are spending all this money on software, and not getting exceptional results.
I was talking to the CEO of a large financial services group in Central Europe, and they were saying we're present in about ten countries, and we have 4,000 people in IT. And we're barely keeping up with doing regulatory-mandated updates to the systems. So there's very little chance for us, and by the way, every year, we need to increase that number to keep up with just doing updates and maintenance on the systems. So there's something directly broken in that system.
The business implication of what you're talking about means that, the idea that an IT department or a tech team or engineers might think about or work on the products of the future is getting further away because people are becoming more devoted to bug-fixing compliance; essentially, keeping the trains running is such a full-time job.
No one's thinking of what's the next business to be in? What's the next product to launch? Would that be fair?
Right, and the big part of the problem is legacy systems. And again, we just have not been designing systems for the past 20 years for the scale at which software is required today.
It's been driven by all these digital growth trends. So today, 80- 90% of the cost in a large organization goes towards maintaining and updating legacy systems. And it's barely keeping up, and costs are increasing, but as you say, it's barely keeping the trains running.
So, that's insanely frustrating for executives in the enterprise. We're talking to another global bank this time. They have 80,000 people in it, 80,000 people in IT. They have, by the way, almost 1000 people just doing iOS development.
Those numbers are mind-boggling because they don't have hundreds of iOS applications. They have 1000, almost 1000 people just doing iOS development . And it's this inflation of headcount that's been happening. Throwing more and more people at the problem but not solving it.
I think what's interesting is what we see at the enterprise is reflected in different ways throughout small and medium businesses and the developer community. I wanna read something you posted recently which caused an enormous amount of conversation on LinkedIn. It was a user posting to Y Combinator Hacker News, and I will just read it.
And I think this also shows the other side of what's happening with software. So this is a post done on Hacker News. An engineer, and I'll read it out, and then we can discuss it. So Dreyfan says, I currently have ten fully remote engineering jobs.
This individual says he has ten gigs, and the bar is so low. He says oversight is non-existent, and everyone is so forgiving for underperforming. I can coast for about four to eight weeks before a job fires me, currently on a 1.5 million run rate for compensation this year, and he goes on a little further.
But this happens when you have too many people. Ioan, this is what you're saying: this guy is moonlighting in 10 jobs. This is the definition of an industry that's at breaking point.
Oh absolutely. And you can imagine this guy isn't even trying, right?
He does nothing literally. And then it takes two months for an organization to figure out that this guy's doing nothing and let him go. But then some people travel along, go with the flow, do a bit of code, and they coast along. Those people stay in these jobs forever. At least this guy; I love him, and he is honest.
But he is on the other side of that equation of large organizations. He's probably moonlighting at ten startups, all of whom are just throwing people at the job trying to fix with this brute-force mentality.
And I think this is so powerful, and honestly, it's very frank; you're almost holding the industry you've grown up in to account. Does it feel at all awkward or uncomfortable? For someone who was in math Olympias, an engineer who's founded several companies, how does it feel to be calling your industry?
Again, I think for us, this has been a mission for almost two decades, even when we started that first company. And we started it because we wanted to fix how software is being developed. And then we realized that the problem and the rabbit hole goes much deeper.
We realized that there's a need for much more systemic change. Not just for a company that helps other companies create software better. And this is why we started FlowX and we re-embedded all that thinking about creating proper software and providing a systemic solution to this challenge as a civilization.
Now we've painted a picture of the problem, and you are starting to take us towards the solution. I want to remind all of our listeners and viewers that if they want to get the show notes, the transcript, and all the links from this show, head over to Unbounded.flowx.ai.
Okay, Ioan, here's the big question. How do we start to think about what you are saying effectively - let's rethink software? Before we get to all the answers, did you realize when creating FlowX that you were embarking on such a radical mission? Did you realize the can of worms you were opening, or were you still going into the rabbit hole, as we say?
There was some understanding of what we're doing. This was our mission, and this is why we started the company and we had some understanding of how inefficient it is. But then even when you see the numbers, it doesn't hit you as hard as when you start working with organizations.
You look at reports that say that the cost of bad software in the US is 2.5 trillion. And that's such a de-dimensionalized way of looking at it. That doesn't even mean anything. So what does 2.5 trillion mean? But one of the things that we are really, let's say, passionate about is actually that 2.5 trillion, that means people are wasting their lives, building things that are not useful. And I think that's Incredibly bad for them, that's incredibly bad for the users of that software and for companies.
It's a very different picture when you think about it as a waste of human intelligence and waste of the human brain. The sheer amount of waste that goes into that 2.5 trillion is unfathomable.
And that's just the cost of bad software to companies.
Interestingly, you touch upon the idea that it's almost so unfulfilling for engineers, and I would imagine it's also for the product owners and the business people. To all be sucked into this system of bad software, bad code, this brute force approach, nobody can be happy deploying 80% of their tech budget just to keep the legacy system lights on.
That cannot be a good feeling right at the top because you said, what impact do we have daily for an engineer? You are going through cleaning up mistakes, just to keep things going at all ends of the equation. Nobody's happy.
And that cost keeps increasing year after year.
So it's almost like this compounding problem that demands a radical solution. It sounds like you are quite ready to be radical in how you go about solving this problem.
I think the people in the industry have realized for a long time something is broken, and they're trying to fix it, but we don't yet have something widespread, like a proper systemic fix.
We're now arriving at a new generation of tools to potentially fix that problem. Because if you look at large companies, they're stuck between three main paradigms. One is they're stuck and completely captive to solution vendors.
To inflexible solutions that do not represent their business and that you cannot tweak, or you don't have the resources to tweak, change, tailor, or adapt to your business and have those solutions reflect your business. So you're stuck with this kind of one size fits all blueprint or template.
And that's one category of organizations; another category of organizations is stuck with insanely high costs of internal IT departments, right? And homegrown technologies and solutions that have some flexibility but are incredibly slow and inefficient to update and tailor.
And we're all seeing the crisis of these very narrowly niche languages. And low code contributed to that and then we tried as a civilization to bring in this generation of no-code tools. This is not a bad idea per se, but the first generation of no-code tools we've seen until now actually doesn't work in the enterprise.
So it works for very simple use cases. To build fairly simple things, but in an enterprise, they deepen the problem because you cannot get away without adjusting that. And using full code. To tweak and create all these workarounds to make it work in that specific context.
Every time you update the platform, you must change all the workarounds and end up in a much bigger mess. And very fast they get to a spaghetti approach because they don't have an engineering structure at the core. You design screen by screen, or you have this simplistic way of representing processes, so you end up with a much bigger mess than initially when you started.
So you see, all these projects started and then abandoned. They have a first release, and people are then like, I'm done with this. So I think this is where our opportunity as not just FlowX, but really as a civilization is, bringing in this new generation of tools that is a hybrid no code, full code approach, because, we cannot get rid of engineers no matter what the no-code tools are saying you cannot get rid of engineers. The structure and the complexity of the enterprise are just way too high.
Just to check in with you there, what you are saying, this is your first big step into almost your first principles of the solution.
Start with combining no code and full code.
Yes. I think that's one essential, quintessential if you will, characteristic of this new generation of tools that can solve the problem. And it's not about replacing engineering; it's about augmenting engineering and taking off the plate of engineers, things that are easy.
And again, philosophically speaking, it's about how you encapsulate the most time-consuming, resource-consuming and complex things in enterprise software, which are scalability and security, and how do you let people focus on functionality? How can you let people build things, functional things?
Because the challenge is not building a functional thing. The challenge is scaling it and how you wrap it in a scalable way and turn it functional into something scalable. I would say that's a second principle.
Again, 99% of the complexity in the enterprise is not the functional or design aspect. It's actually security and scalability. The non-functional aspect of a solution and the robustness of that.
So it sounds like number one; it's about bringing together the two worlds of full and no code.
The second thing, it sounds like the elimination of repetitive standardized tasks that are about managing the legacy to free up that time to go and work on real features, new products, new functionalities.
Absolutely. Providing a layer that encapsulates security and scalability, so you don't have to worry about that and you can let people focus on the functional things they need to build. That simplifies the task by orders of magnitude, not just one magnitude.
And then since we're there, actually we're seeing now AI being able to create decent functional code. So that's another way of augmenting the capability of engineering. And I think that allows us to reduce engineering workforces to true hardcore engineers who can contribute in an engineering-ish way to solving the problem and to the business. Rather than having to throw bodies at the problem.
So helping, organizations focus much more on quality than quantity. And other than that, we're doomed into this self-colliding, imploding Ponzi scheme.
You did paint a house of cards with diminishing returns over time, calling upon this radical solution such as integrating low code and full code, such as automation, or at least reducing redundant tasks so that engineers can focus on high-value activities.
I think moving away from the brute forces, as you said, and then using things such as AI. So what's next for FlowX? What's next for the platform? What can we expect in 2023? What are some of the things you are most excited about and how can we start addressing this problem you've talked about?
We already, at the platform's core, had these concepts. The hybrid, no-code, full code, AI, the generative AI for user interfaces and encapsulating the non-functional characteristics and letting people build functional things provide a layer of scalability and robustness for what people build.
2023 for us is super exciting because we're starting to bring out actual machine learning and AI-assisted development, and AI-assisted operations. We have been running for the past two years with some very large banks, and now we have gathered enough data to start producing, not just good, but really great results in terms of assisting people building software. But also assistance to people running processes and optimizing processes. Such things as AI-driven process optimization and AI-assisted development of the interface.
Assisted development of business rules. These are all coming, and we're going to announce a couple of releases; I think in Q2, Q3, that is going to be very exciting I would say, the entire community.
That's great. So it sounds like understanding the problem, we've got this house of cards in the way software has been built, and you are taking a few key concepts, this hybrid approach, this streamlining automation of basic tasks so people can move up to higher order.
Functional, feature-driven, product-driven new releases, and being as smart as possible to deploy AI to assist in that. So we get more leverage in this game. We can move away from brute force and become smarter, not working harder, and creating more value for all those constituents because it sounds like if the CTOs are happier, and his engineers are happier. And you know what, if they're all doing well, it sounds like the customer wins too.
Absolutely. And again I think, what we're benefiting from here is leveraging technology and software, which has this very interesting characteristic of having incredibly disproportionate results.
Like orders of magnitude better results than approaching things in the brute force way. And I think that's the opportunity for our civilization to do things in a way that's not just a bit better, but that is orders of magnitude better going forward.
I can't wait to have you back again. It would be great if we could put some time in the calendar around Q2 and Q3 so we can talk about how the platform is delivering on solving this problem and unfolding this house of cards so that all of those stakeholders we talked about so they can actually do better in their work or if they're actually on the other end. If they're a customer of these financial institutions, they can be happy, loyal customers who see value in that service provision.
Absolutely. And imagine if, as a civilization, we were able to build software and technology ten times faster, what that would mean..
There are many people who are excited about curing cancer. Imagine we're 20 years away now; imagine we would be 10 times more efficient at that and that cure could be two years away instead of 20 or just five times even, meaning It's four years away!
All of that is powered by software. Imagine we get ten times or five times better than we are today at building software. So what's the impact we have on civilization when instead of something that's maybe 10, or 20 years away, we could have something one or two years away?
And so there's this interesting transformational opportunity that we have for humankind as we launch into a new era with better software, with better technology that's better, and more robust and reliable instead of just the current house of cards if you, as you call it.
Ioan, listen, thank you so much for really being so frank and open about an industry you've been in your entire career, and calling it to account, challenging the status quo, and saying, we can do better. It's great to be part of it. So it's great to hear what you and the team are thinking about it.
I certainly can't wait to have you back to hear about some of these future releases. So I want to say a big thanks to you and a big thanks to all of our listeners and viewers. If you want to catch up on the transcript, show notes, and links to what anyone was talking about, head over to Unbounded.flowx.ai.
Okay, that's a wrap.
--------
FLOWX.AI is part of IBM Cloud for Financial Services. See listing and more details here.
Author: Monica Dumitriu
SHARE