AI: Deconstructing AI and Digital Transformation in Financial Services | Ara Abrahamyan, Board Advisor at Cognaize & Ameriabank

Monica Millares: [: 00:00:00

Ara Abrahamyan: Hello Monica, I'm very well, thank you. How are

Monica Millares: you? I'm really good, thank you. Very looking forward to our conversation. Me too. Yeah. Thank you. Thank you for coming to the show because like just to add context for everyone, Ara and I met a few months ago in Amsterdam.

Like I was going to Money: 2020

And I did, but it doesn't look that good. So we never posted that video , because it's like impromptu not looking good. So instead we were like, let's jump into, proper podcast camera mode, and that's why we have [00:01:00] Ara today, so thank you, Ara.

Ara Abrahamyan: There was a signal from top so that we would happen to sit next to each other.

Exactly. Looking forward to it, Monica.

Monica Millares: Likewise. Okay, so to get started for context. This podcast is all about how can we create more purpose driven fintechs and have more impact as such. So that's a very important question. So in your opinion, how can we design and build fintechs that are more purposeful?

cts of their life in various [: 00:02:00

Of a FinTech? Yes. By disentangling the financial services per se.

Monica Millares: Can you expand on what do you mean with the DNA by disentangling financial services?

Ara Abrahamyan: I think, look financial services, industry, banks, insurance companies, and so on and so forth are very large, are very complicated, are not very customer centric, not very friendly people friendly and so on and so forth.

[: 00:03:00

Yes. Otherwise, if you glue them together, you will become the old, same, big, complex bank.

Monica Millares: Exactly. And that's what we don't want to do. Hence, I do agree that Fintechs have a DNA that's different to banks, and probably both of us we've worked in big banks and Fintechs, so there's a comparison point that you're like yeah, they are different.

They are not better than the other. They are just different DNAs.

Ara Abrahamyan: Absolutely.

Monica Millares: Absolutely. Yes. And then just building exactly on what I said about working in both fintechs and banks, like you have a very interesting career, like you've worked across the big banks now just sit in the advisory board, both big banks fintechs, you also teach.

visory work at this point in [: 00:05:00

Ara Abrahamyan: With pleasure. Yeah. So I'm an engineer. My background is an engineering background. I did whatever, 25 plus years ago, my PhD in, in computer science, apparently part of the artificial intelligence at that time, which was called decision support systems.

ginally from and then in year: 2000

Which was [00:06:00] very successful from my perspective because I had a pleasure to meet extremely talented and clever people who taught me a lot. I had no clue about financial services, no clue about banking. So I was genuine, pure information technology profession. But I've been building then systems ID for banks for 15 years at Deutsche.

And then, when, I was responsible in my last job, there was, I was responsible for risk technology here and I moved to Austria to Vienna. And five years later, or three and a half, four years later, I became a member of the board responsible for the digital transformation of Esther group, biggest financial services provider in central and Eastern Europe, headquartered in Vienna.

to Frankfurt and started to [: 00:07:00

And then I started to do my consulting business where I do consulting for startups. And that is the bit where I help, in the advisory function cognize and also another startup with profit but also do some consulting work to incumbent banks. Area bank is one of my clients.

is a part of me that kind of [: 00:08:00

I teach leadership, I do some parts of the digital transformation curriculum and so on and so forth.

at understands both which is [: 00:09:00

One is a big bank. The other one is an AI startup, both in financial services. Can you tell us what's the role of purpose in these two companies? One is big, one is small, and what's purpose for each

Ara Abrahamyan: of them? AmeriBank is One of the leading banks in some aspects, the leading bank in Armenia, and, my

to the people in my country, [: 00:10:00

From my career to support them on that journey, whereas cognize is another crazy engineering part of my of my heart and my maybe my brain where were the most advanced technologies, right? Often connected technology, which is that the leading edge of research is being implemented and put into the purpose of helping organizations to become more efficient.

etween my past and my future [: 00:11:00

Yes.

Monica Millares: And I guess like I want to dig a little bit deeper into that about building bridges, but the thing that is the common theme is digital transformation, whether it's a big bank doing a project of digital transformation, or if it's a fintech, because we ourselves as fintech, sometimes we do not see ourselves as doing digital transformation.

to becomes a little bit more [: 00:12:00

So there's an element of transformation as well in fintechs. So I want to expand on that. But first start with the big organizations. How do we breach this world between like big organization and innovative companies and deep tech while still being able to stay Okay. In the mind and with all the constraints and priorities and pressure from investors and regulators,

Ara Abrahamyan: and that's a very good question.

n the journey of the digital [: 00:13:00

Whereas if you think about the digital transformation of an incumbent company, that is much more. Transformation that looks in words, right? It looks into the organization itself. It looks at the way it does things and so called how and it looks at what it does, right? What do they do in terms of the implementation of?

Modern technology, whether they move their applications into cloud, whether they implement DevSecOps as their implementation methodology, whether they employ AI, whether they do RPA and so on and so forth, right? So they take their processes and procedures and products and kill everything they know.

About the way these [: 00:14:00

By splitting into two pillars, what you do in terms of the technology itself and how you do it in terms of ways of working, collaboration, innovation, agility, and so on and so forth.

Monica Millares: I love that. And I want to dig a little bit deeper on the what? That is taking the most modern technology and applying it to your company.

What are the challenges that big banks or the incumbents face when they are trying to implement these new technologies?

Abrahamyan: There are many. [: 00:15:00

And this process is usually consisting of many, small steps that in its aggregate contribute or end up being the big shift from the old world to the new world. Is. Not in a good match with the corporate governance. Usually people who are responsible for this kind of transformational programs are board members who have short, reasonably short, three years, four years contracts, [00:16:00] which is not always the term of such a transformational program, right?

And management often is simply Reluctant or afraid of starting this kind of transformation because it's bearing huge amounts of risks and their terms to not give them the opportunity to address them and actually see the results or harvest the results of this transformation. That's why I think it would be fair to say that most of the incumbents in a financial services industry still struggling.

at kind of big service pipe. [: 00:17:00

Monica Millares: but this is it's a challenge because exactly the fintechs are trying to get ahead.

While the banks have the customers and the banks have the processes and the banks have the. know how and the funds to do it. So how could banks overcome this challenge? Because probably we're not going to change the three year tenure of the board.

Ara Abrahamyan: I wish we could, but probably not here in this very podcast.

fit where you as a bank have [: 00:18:00

So For me, the key word here is a partnership. And, partnership is not procurement in a classical

Monica Millares: sense of a bank,

Ara Abrahamyan: because if you want to kill a startup, you can put the startup into a procurement process and they will go bankrupt. They will go bankrupt. Yeah. So, it's about actually, creating meaningful trust based partnership between the two worlds where the best of both can come into play.

But first I want to go back [: 00:19:00

And that's more challenging. So what are the biggest challenges that we face when it comes to culture in all this digital transformation process? If

Ara Abrahamyan: you.

they full of creativity and [: 00:20:00

So one of the biggest challenges in the transformation of financial services industry is In terms of the culture, it's the talent itself. You hardly get the people you need. But there are also some technical challenges as well. There is frequently outdated infrastructure, there is frequently outdated architecture, technical, technological architecture.

this is the regulation, the [: 00:21:00

bias and so on and so forth. Data protection. This is another huge component and block of impediments to the digital transformation that creates huge amount of ethics to the corporate officers.

Monica Millares: Yes. And I could add also to FinTechs, maybe not such a big headache, but it's something that as I've done all these episodes with people and talk to other people outside, when I ask, Hey, what is it that you would like to see change?

tter. So we need to have the [: 00:22:00

Ara Abrahamyan: Yeah. Maybe if I can add one sentence to that my corporate life has been full of interaction with regulators, right?

aboration with the regulator [: 00:23:00

And you want to make sure that the regulators understand the challenges and sometimes even help you to overcome the challenges to address their concerns and or to provide the best service to the market. So even there, a collaboration and communication is key.

Monica Millares: Yes. And that comes back to what you said at the beginning about trusted partnerships, including the regulator.

move more to the other side [: 00:24:00

Ara Abrahamyan: Sure, with pleasure. Cognize this is a very interesting company. It's an amazing company that does something which is reasonably simple on its surface, but hugely complex in its details. It extracts Information data from unstructured documents, right? It's a platform. It's a SAS company that provides tool to help people organizations, companies or clients and so on and so forth to [00:25:00] extract valuable information like reports, PNL report data.

Balance sheet data and so on and so forth from PDFs, pictures, photos and so on and so forth. Simple problem, very complex solution.

Monica Millares: Yes, that's what I was going to say because it seems that this is just, well to me, we're just at the beginning of the AI revolution, right? And this is just a very specific use case that just on that use case we can see the complexity.

Oh yeah, you extract data, it's not that, it's not that easy, no, it's not that easy. So if we go a little bit into high level and we say AI financial services, what do you think are going to be the biggest challenges in the next five years when it comes to adopting AI technologies?

Abrahamyan: First is skills, [: 00:26:00

Monica Millares: What can we do about it right now?

Ara Abrahamyan: Skills is very easy. Very easy in quotes. So I'm sorry. I'm being cheeky. You I think you have to acknowledge and understand that there are things that you can do as a as a company undergoing a digital transformation. There are things that you can do yourself and the things that you need to partner with.

If you look at the partnership landscape of the most advanced digital companies of these days, Google's, Amazon's and so on and so forth. They are partnering with hundreds and hundreds of the smaller companies to to get their services done. And so this you have to accept at a first place.

Second is the [: 00:27:00

And there is a lot, a huge amount of homework that organizations need to do in order to implement all the kind of data governance framework, data quality framework, and so on and so forth, so that's a big amount of, just a big amount of work that has to happen. Yes,

Monica Millares: because in my head I'm like, data quality just on its own.

ms of if you look at in this [: 00:28:00

It's in interest of all three, and it's Effectively to make sure that, the solution that is being built by one and implemented by the other is happily accepted by the third, by the independent observer.

So I see it as a

Monica Millares: collaboration. Think about yeah, I'm like, oh, that is a little bit more complex than just, like you as a fintech and the regulator, now it's you, the fintech the fintech, the regulator, and the third party.

and, try to make it work for [: 00:29:00

Making sure that the collaboration on this triangular collaboration is seen as such.

Monica Millares: Yes, so you just said something very true that it's like this is the new normal. That's it. AI is not going away. And then this time of the year Just happens to be that it's like Q4. We're all getting ready to do our strategic planning for next year

When it comes to AI, let's say that we have our strategy and roadmaps for the 95 percent of the business. How do we go about thinking our AI strategy? Whether that is for next year or for the next five years

ake in that context is maybe [: 00:30:00

It's what you just said is the fact of acceptance that AI is here, right? Our discussions with our clients two years ago or one and a half years ago were totally different. We were starting from explaining what is AI and why do we need it and so on and so forth. Now, I think this is done, right?

Everybody is, okay, good. AI is here. It's reality. That's the first step. And the second is

accept the fact that this Either with you or without you is going to transform every single industry in one or the other way, and it's the job of the leadership is to prioritize the capability of actually implementing different types of basic machine learning or generative AI or image processing or text, large language mode, whatever, right?

You name it [: 00:31:00

Is not always fitting into the logic of a large corporate where you have a project, where you have a benefit cost benefit case and you do it because frequently you will find out that some things just don't work the way you expect it do not have that kind of ROI on your kind of investments and so on and so forth, but not being engaged, not [00:32:00] being involved in that game is not a solution.

Yeah, so you better start playing it now,

Monica Millares: so if I summarize what you said that it's yeah It makes a lot of sense. It's number one ensure that you start building your AI capabilities Whatever that looks like for your business, but it's like getting to the game. Now it's about building capabilities because we don't have that knowledge, that expertise, the process.

We don't have that. Most fintechs or banks don't have it because it's new. So one, it's build the capabilities and two. The process that you explained, it's what I call, use the experimentation approach. Correct. It may not work, it will work, but then you go with a hypothesis, you have a very specific thing that you're going to do.

you do the next experiment. [: 00:33:00

Ara Abrahamyan: Listen, I will I think it may sound too simple, or even simplistic, but in reality it is the simple things that we struggle to accept and implement and we tend to overcomplicate things By making them dysfunctional.

And if we, just do this, okay, let's, if we translate it into real practical ways of doing it, what would I recommend my client? I say, okay, good. Hire 10 people, put one lead AI engineer on top of it and establish three partnership AI service providers in your industry and start it. Take 10 projects, prioritize top three of them, try them out.

Two will not work, one [: 00:34:00

Monica Millares: Practical, yes. Yes, because that's the way, right? To me. As we, let's say in the next one year, three years, five years, we start building all these capabilities.

And we start learning as an industry, if we fast forward 10 to 15 years from today, what do you think was the, in the future? What we, when we look back, what is the impact of AI in the industry when it comes to building more purpose driven fintechs?

en we talk about most of the [: 00:35:00

By removing some of the judgmental issues out of it by automating them or by digitally transforming them. Now, what does it mean in terms of the implementation of purpose driven technology? So by making processes efficient, you make them more accessible to a broader public. If something is much more cost efficient, products are cost efficient.

second thing, by eliminating [: 00:36:00

We will make them better for the, world, right? So you start, if you use tools that are assessing candidates, right? We know this classical case of a high bias based on the training information that is implied or applied by the to the existing batch of CVS will permanently choose a specific subset of people which is correlated to the historical best of them, but by knowing it, acknowledging it and actually explicitly eliminating that and many other biases out of it, [00:37:00] you can make a fair, transparent and useful, efficient process.

Monica Millares: So we need to become very good at finding those biases, such that we train the machine to take to, not be biased, basically, but it's, interesting, right? Because a bias is a bias.

Ara Abrahamyan: Yes, but in reality, I think the problem is not finding them. It's the problem is actually accepting, right? So we, as humans, right?

The catalog of our biases is known, right? If you list them there's confirmation bias this bias, it's on and so forth. So they, there is a reasonably comprehensive list of them. It's just, we are not good. We. Humans are not good at avoiding them. There are ways of handling them, right?

There are ways of [: 00:38:00

That your biases are being incorporated into your, or have been taken care of, and, then it just becomes the matter of routine, right? You validate your data set against whatever set of tests, if you will, is, and then you are good to go. You will not be perfect from the scratch, of course, that's normal, right?

fying and eliminating biases [: 00:39:00

Monica Millares: Yeah, and I think I'm seeing a theme here that it's like, of course, we have. The machine AI. But then we have the human side that it's exactly the biases as such.

And you use this word in the conference in Amsterdam. And it's also part of cognizes proposition. So cognize and you have talked about hybrid intelligence. And I think you were just getting into that. Can you explain what is hybrid intelligence?

Ara Abrahamyan: With great pleasure. We actually write the concept of hybrid intelligence on the big, kind of slogan of what and how we do.

ut a combination of strength [: 00:40:00

And you will normally hire a person into your team who is very good at that detail to each to hedge your weaknesses to offset your weakness. Thank you. So in reality, the A. I. And I. Q. Or each you are very good. How should I say partners? Yeah, and they are very complimentary night and the things that are done well by humans, are not so easy for technology.

re complex and difficult for [: 00:41:00

Because there's so much technology many, kinds of technology involved in it. So by accepting that fact, we think that the hybrid intelligence, so effectively a partnership between AI and a person, will make that person much more efficient and will help the person to get rid of the routine. Waste of time, waste of effort and make much more productive and efficient.

'm like, all we talked about [: 00:42:00

It's just just partnerships, but with different parts of the ecosystem as such. So as we start to wrap up the episode, just following on that theme that it's like partnerships, but also culture. And we have the incumbents, we have the established fintechs. And now we have the rise of the AI startups.

How,

can we all work best together? Because in some cases, like AI startup, big bank incumbent, somehow the definition of each of them, it's a little bit incompatible. So what's your take on how do we make? This partnership a successful one.

Ara Abrahamyan: My instinct [: 00:43:00

And I talk about the trust in the, in the relation of working environment in the relationship between leaders and stakeholders. And their team members in relation in the relationship between different parts of the organization's trust between, different departments, divisions and so on and so forth.

ate world and and innovative [: 00:44:00

It's, first the acknowledgement that they're different, that they're diverse, right? And acknowledgement as they are. Yeah, and, by accepting this, having the courage to put the trust as a key component of partnership between between both parties, right? I learned over my career in the most painful way that the contract is the worst possible place you need to look up for the solution.

t it is that we are going to [: 00:45:00

Some companies have the client service provider relationship. Some are investors. Some are just I don't know It was advisory relationship and so on and so forth. But the core idea is Are we in agreement about the common? Mutually beneficial goals of what we're trying to achieve and if you spent 90 percent of the time on alignment on that very specific everything else works out automatically.

ng of what it is that we are [: 00:46:00

Monica Millares: I love that. I have this saying that the future of Fintech is human, and that's exactly what it refers to.

It's it's not about AI, it's not about technology. It's about being more human across all the touchpoints in the ecosystem. Not just product, but like you said, partnerships. It needs to be the human element. It's trust, it comes back to human. It is. Yeah. Yeah, I'm not sure. I'm not sure.

Ara Abrahamyan: I'm not sure if that's if that was the conclusion that an AI topic, a podcast as an AI topic would come to.

But it is. Yes.

asking and asking and asking [: 00:47:00

Ara Abrahamyan: Oh, we are. Obviously our web. We're starting a series of podcasts actually ourselves. The announcement will come soon. Probably by the time this podcast, is published, we will be already live ourselves. We frequently go on to conferences, talk about ourselves, present our product. Our webpage and this is there with totally the shipped articles and so on and so forth.

So we're pushing all the channels to make sure that our ideas are heard and seen. Good

much. So as, thank you. As a [: 00:48:00

Ara Abrahamyan: That's a tricky question. I think the one thing that needs change or is necessary is to understand that they're different. It's the understanding of their difference or acknowledgement of their difference. Fintech is not a bank. Or an insurance company. A bank is not a fintech and an insurance company.

And by understanding, it's like with humans, by accepting who you are,

stand that they're different [: 00:49:00

Monica Millares: understanding that we're different. And then, like you said earlier, looking at our strengths, looking at our complementary weaknesses slash how we can work best together. Awesome. It's been an absolute pleasure having you in the show, Ara. Thank you so much.

Ara Abrahamyan: Thank you. The pleasure was mine. I really enjoyed it.

Thank you very much, Monica. Thanks for having me. Thank

Monica Millares: you. Thank you. Thank you, everyone. See you next week. Ciao.