EP976: Modular, cloud-native, and digital-first

This interview features Udeet Bhagat from PureSoftware discussing the success of their flagship banking platform, Arttha. The source highlights how this cloud-native and modular technology enables financial institutions to automate digital onboarding and loan origination processes. A major focus is placed on the platform's ability to remain scalable while adapting to local regulatory requirements in diverse markets like Africa and Southeast Asia. The text also outlines future plans to integrate AI-driven credit scoring and alternative data, such as mobile money trails, to improve financial inclusion for underserved populations. Ultimately, the company emphasizes its commitment to innovation and agile design to keep pace with the rapidly evolving fintech landscape.

Transcript

SPEAKER_00 0:00

Every time you tap your phone to pay for a coffee or you know use an app to split a dinner bill.

SPEAKER_01 0:05

Right. Or even get instantly approved for one of those small loans at checkout.

SPEAKER_00 0:08

Yeah, exactly. Whenever you do that, you are interacting with this massive layer of invisible digital infrastructure. And I mean, most of us never really think about this underlying plumbing of the financial world.

SPEAKER_01 0:20

Aaron Powell Oh, definitely. It's just expected to work. But when you look closely at how that architecture is being rewritten right now, and it is being completely rewritten, you realize it's fundamentally changing things.

SPEAKER_00 0:31

Aaron Powell Changing who gets to participate in the global economy, right? And unfortunately, who gets left behind.

SPEAKER_01 0:37

Exactly. The shift happening in back-end financial technology as well. It's rewriting the rules of access. The tools that banks use to decide who is trustworthy are evolving way faster than the public realizes.

SPEAKER_00 0:49

Aaron Powell, which brings us to our source material for this deep dive. We are looking at a highly detailed interview from the September 2025 edition of the IBSI FinTech Journal.

SPEAKER_01 0:58

Aaron Powell Yeah, it's a really fascinating read.

SPEAKER_00 1:00

It really is. It's a conversation between Robin Amlot, the managing editor of IBS Intelligence, and Udit Bagat. He's the senior vice president for a company called Pure Software.

SPEAKER_01 1:11

Aaron Powell Right. We're going to focus heavily on their main banking platform, which they call ARTHA.

SPEAKER_00 1:15

Yeah, ARTA. So the mission for today's deep dive is to understand how this specific platform is completely rewiring the global banking system.

SPEAKER_01 1:24

Aaron Ross Powell Because it's not just about updating software, it's about how they balance cutting-edge artificial intelligence with incredibly strict hyperlocal regulations.

SPEAKER_00 1:35

Trevor Burrus, well, to bring uh millions of underbanked people into the financial fold. Okay, let's unpack this to truly grasp the sheer scale of what pure software has achieved with ARTA. We really have to start with the foundational numbers.

SPEAKER_01 1:47

Aaron Powell Yeah, the footprint is just massive. I mean, when we talk about scale in the consumer tech space, a successful app might boast, what, a few million active users?

SPEAKER_00 1:56

Aaron Powell Right, which sounds like a lot.

SPEAKER_01 1:57

Aaron Powell It does. But ARTA is operating as the critical back-end infrastructure for financial institutions across more than twenty global markets. Aaron Powell 20 markets. That's huge. And because of that reach, the platform actually facilitates over three billion transactions annually.

SPEAKER_00 2:12

Aaron Powell Wait, three billion? With a B.

SPEAKER_01 2:15

With a B. Over three billion a year.

SPEAKER_00 2:18

Wow. I mean, three billion transactions a year requires an infrastructure that most regional banks simply do not have the capacity to build or maintain on their own.

SPEAKER_01 2:27

Aaron Powell Which is exactly why those institutions adopt Artha instead of trying to build it. The platform currently supports over 150 million users and more than 10 million merchants. That is wild. Yeah. And this goes far beyond just like basic ledger management. We aren't just talking about tracking simple deposits and withdrawals here. Artha handles complex financial orchestration.

SPEAKER_00 2:48

Right, because didn't Pure Software recently win some big industry awards?

SPEAKER_01 2:52

They did, yeah. They won dual honors at the IBSI Digital Banking Awards for 2025. And that was specifically for their ability to deliver fully digital client onboarding and loan origination.

SPEAKER_00 3:03

So in the interview, Bagat uses a few specific terms to describe the architecture making all this possible. He calls the platform a cloud native, modular, and API first.

SPEAKER_01 3:13

Yeah, those are definitely buzzwords you hear a lot in tech.

SPEAKER_00 3:15

Oh, for sure. But to understand why they are so disruptive to traditional banking, we kind of need to break down what they actually mean in practice. Because traditionally, if a bank wanted to offer a new loan product, they had to build this massive monolithic software system from the ground up.

SPEAKER_01 3:32

Aaron Powell Right. And store all that data on physical servers in a basement somewhere.

SPEAKER_00 3:37

Aaron Ross Powell Exactly. And that legacy approach is incredibly rigid.

SPEAKER_01 3:41

Aaron Powell Oh, it's so slow. When a system is housed on physical on-premise servers, scaling up is literally a physical process. If you suddenly get a massive influx of users, you have to go buy and install more hardware.

SPEAKER_00 3:53

Aaron Powell Which takes time and money.

SPEAKER_01 3:55

Aaron Powell Exactly. So cloud native just means the entire architecture was born in and lives entirely in decentralized cloud environments. It can scale its computing power up or down instantly based on demand.

SPEAKER_00 4:06

Aaron Powell Without the bank ever having to touch a physical server.

SPEAKER_01 4:09

Aaron Powell Yep. And then there's the API first aspect.

SPEAKER_00 4:12

Trevor Burrus Right. An API or application programming interface, it's essentially a universal translator, right? It allows two completely different pieces of software to talk to each other securely.

SPEAKER_01 4:22

Aaron Powell That's a great way to put it. So if a banking app needs to verify your identity with a secure government database, it doesn't need to absorb that entire database. Trevor Burrus, Jr.

SPEAKER_00 4:31

It just uses an API to send a quick request and get a verified answer back.

SPEAKER_01 4:35

Exactly. And that leads directly to the third term, which is modularity. Because Artha relies on these APIs, it doesn't have to be one giant tangled web of code. It's broken down into independent modules.

SPEAKER_00 4:48

Aaron Powell So there's a module for like intelligent document processing.

SPEAKER_01 4:52

Right, which uses AI to read and understand uploaded paperwork. Yeah. There's a totally different module for biometric authentication and another one for credit evaluation.

SPEAKER_00 5:01

It honestly sounds remarkably like a set of highly advanced Lego blocks for banks.

SPEAKER_01 5:05

Ah, yeah. Lego blocks is the perfect analogy.

SPEAKER_00 5:08

I mean, instead of a financial institution spending three years and millions of dollars to build a cumbersome onboarding process from scratch, they can just look at this platform and snap these optimized digital first modules together.

SPEAKER_01 5:22

Exactly. If they want to launch a new loan service in a specific region, they just grab the EKYC, block the electronic new customer protocol, snap it into the credit evaluation block, and they're basically ready to go.

SPEAKER_00 5:34

They can roll out a fully digital product in a fraction of the usual time.

SPEAKER_01 5:37

What's fascinating here is that this modularity isn't simply a convenience for software developers. The ability to snap these components together translates directly into plummeted operational overhead.

SPEAKER_00 5:49

Because it's cutting out all the manual steps, right?

SPEAKER_01 5:51

Right. Just think about the physical manual labor involved in traditional banking. You've got a human clerk reviewing photocopied documents, running manual background checks, physically entering data.

SPEAKER_00 6:02

It's so tedious.

SPEAKER_01 6:03

It really is. By replacing that archaic paperwork with seamless mobile first interfaces and intelligent document processing, the cost to acquire and service a single customer just drops dramatically.

SPEAKER_00 6:15

And the economics of that drop in cost are what actually drive financial inclusion.

SPEAKER_01 6:20

How do you mean?

SPEAKER_00 6:21

Well, if a traditional bank spends $50 in operational overhead just to open and maintain an account, they are financially incentivized to only target wealthy customers who will deposit thousands of dollars.

SPEAKER_01 6:33

Oh, absolutely.

SPEAKER_00 6:34

They literally cannot afford to serve someone who only wants to deposit $10. The profit margins entirely dictate the audience.

SPEAKER_01 6:42

That is a really crucial point. But when you deploy a cloud native modular system that automates that entire lifecycle from the application intake to risk evaluation to the actual disboosement of funds that costs to serve a customer shrinks to pennies.

SPEAKER_00 6:57

Pennies. That's incredible.

SPEAKER_01 6:58

Right. So suddenly offering an account or a microloan to a rural farmer or a small market vendor becomes not just viable, but highly profitable at scale.

SPEAKER_00 7:07

And that deep understanding of regional economic dynamics, particularly in places like Africa and Southeast Asia, is precisely why Pure Software secured those IBSI awards.

SPEAKER_01 7:17

Definitely. But that regional focus actually introduces a pretty significant tension in the source material.

SPEAKER_00 7:23

Oh, the compliance issue. Yeah. Bagat talks extensively about operating in over 20 different markets. They are providing this globally scalable, automated Lego set across Africa, the Middle East, Southeast Asia, and the Americas.

SPEAKER_01 7:38

But banking is a hyper-regulated sector. Every single one of those countries has wildly different legal frameworks.

SPEAKER_00 7:45

Right. Like what is perfectly permissible regarding customer data in the UAE might require massive structural changes to be legal in Indonesia.

SPEAKER_01 7:53

Exactly. So the big question is: how can a single, unified platform scale globally while simultaneously molding perfectly to the strict hyperlocal regulations of 20 different governments?

SPEAKER_00 8:04

It is arguably the most difficult hurdle in global financial technology.

SPEAKER_01 8:08

Ah, without a doubt. The industry refers to this specifically regarding things like data residency laws. For example, a country might legally mandate that any financial data generated by its citizens must physically remain on servers located within that country's borders.

SPEAKER_00 8:21

So it literally cannot be exported to a global database.

SPEAKER_01 8:24

Right. Not legally.

SPEAKER_00 8:26

Wait, if you're using a massive, globally scalable AI to do real-time KYC or risk scoring, doesn't that inherently risk bulldozing over strict local data residency laws or biometric regulations?

SPEAKER_01 8:38

That is the exact paradox.

SPEAKER_00 8:40

Aaron Powell Because an AI model, by its very nature, wants to ingest massive pools of data to train itself and become more accurate, right? If the AI is centralized, but the laws demand the data stays localized, those two things seem entirely incompatible.

SPEAKER_01 8:56

They really do. But pure software engineered a highly specific governance framework to solve this. The strategy relies on fully decoupling the AI decisioning from the compliance logic.

SPEAKER_00 9:06

Okay, decoupling them. How does that work in practice?

SPEAKER_01 9:09

To visualize it, think about separating the brain from the rulebook.

SPEAKER_00 9:12

Oh, I like that. So the brain being the artificial intelligence models determining the risk.

SPEAKER_01 9:16

Correct. The core intelligence that handles real-time risk scoring, fraud pattern recognition, and document analysis that is the global brain. It is powerful and unified. However, the platform architecture ensures that before this brain is allowed to interact with any local data or finalize any decision at all, it must process its actions through a region-specific policy engine. That is the rulebook.

SPEAKER_00 9:42

So when a bank in Kenya deploys the ARTA platform, they configure the rulebook module to strict Kenyan compliance standards.

SPEAKER_01 9:49

Exactly. They set the specific KYC thresholds, the local biometric consent rules, the data residency boundaries.

SPEAKER_00 9:56

And the core AI platform doesn't need to be rebuilt from scratch, it just operates within the constraints of that local engine.

SPEAKER_01 10:02

And that is how they ensure banks don't have to choose between innovation and regulation. The AI still delivers its lightning fast decisioning, but it's entirely governed by local controls.

SPEAKER_00 10:13

Okay, but if the AI brain is constrained by this local rule book and it is legally barred from extracting or moving the raw customer data out of that specific country to train itself, how does the global platform ever learn or get smarter?

SPEAKER_01 10:26

Aaron Powell It's a great question.

SPEAKER_00 10:27

I mean, if it can't see the raw data, doesn't the AI just stagnate?

SPEAKER_01 10:30

Aaron Powell It would, if it actually needed the raw data. But modern AI architecture relies on tokenization and localized parameter updates to get around this.

SPEAKER_00 10:39

Aaron Powell Tokenization.

SPEAKER_01 10:41

Yeah. The global AI brain doesn't actually need to see a raw scan of a citizen's passport or their specific name to learn about fraud. The local system processes the raw data, and the AI extracts an anonymized mathematical representation of the pattern.

SPEAKER_00 10:56

Aaron Powell Oh, I see. So it learns from the math, not the personal information.

SPEAKER_01 10:59

Aaron Powell Exactly. It uses APIs to securely transmit these mathematical updates, you know, the learned patterns of new fraud tactics or risk indicators back to the global model.

SPEAKER_00 11:09

Aaron Powell Without ever moving the underlying regulated private data across a border.

SPEAKER_01 11:13

Aaron Powell Exactly. The brain gets smarter globally while the data remains totally secure locally.

SPEAKER_00 11:18

Aaron Powell That is brilliant. That architecture perfectly solves the compliance side of the equation. But what really stands out in the interview is what happens once that AI is safely deployed in these emerging markets.

SPEAKER_01 11:28

Aaron Powell Right, because of what they use it for.

SPEAKER_00 11:30

Yeah. Because it is securely managed, pure software is using it to process entirely unconventional types of data to evaluate users. And this is vital because when we think about applying for a loan, we think about credit bureaus, W-2 tax forms, long histories of mortgage payments.

SPEAKER_01 11:47

Yeah. In established Western markets, that traditional infrastructure exists. But in the emerging markets where ARTA is expanding, traditional credit histories simply do not exist for the vast majority of the population.

SPEAKER_00 12:00

Aaron Powell Right. So if a financial institution relies solely on a traditional credit score, they are automatically excluding millions of individuals who have never possessed a formal credit card or a bank account.

SPEAKER_01 12:10

Aaron Powell It's a massive blind spot for traditional banks.

SPEAKER_00 12:13

Aaron Powell Here's where it gets really interesting to me. Traditional credit scoring is functionally identical to a rigid standardized test in college admissions.

SPEAKER_01 12:20

Oh, that's a good comparison.

SPEAKER_00 12:22

Yeah. Think about it. If a student didn't take that specific standardized test, or if their high school didn't offer the prep courses for it, their application is rejected by default. The admissions office basically says, well, we have no standardized data on you, so we cannot verify your academic capability.

SPEAKER_01 12:38

Aaron Powell And the traditional banking system does the exact same thing. They look at you and say, you have no formal credit history, therefore you are an unquantifiable risk, loan denied.

SPEAKER_00 12:50

Exactly. It creates this awful, self-perpetuating cycle of exclusion. You cannot get credit because you don't have a credit history, and you cannot build a credit history because no one will give you credit. It's a catch-22. But what Aarth is doing is shifting the evaluation model. Instead of relying solely on that single standardized test, the AI looks at a massive portfolio of alternative consent-based data.

SPEAKER_01 13:14

Right. So to use the college analogy, it's like an admissions officer looking closely at a student's extracurricular activities, their daily attendance record, their part-time job to judge their true reliability. Yes. In the financial context, Pagat explains that they integrate unconventional sources like telco data and mobile money trails.

SPEAKER_00 13:33

Telco data, so phone records.

SPEAKER_01 13:35

Or to understand how powerful that is, we have to look at the mechanics of what telco data actually represents. It isn't just knowing that someone has a phone, the AI analyzes the behavioral patterns associated with mobile usage.

SPEAKER_00 13:47

Like what kind of patterns?

SPEAKER_01 13:48

Like does this individual top up their prepaid data plan consistently on the exact same day every week? Do they have a stable pattern of incoming versus outgoing calls, suggesting a reliable network of business or personal contacts?

SPEAKER_00 14:02

Oh wow. And the mobile money trails are even more revealing, right?

SPEAKER_01 14:05

Absolutely. In many emerging markets, informal economies run entirely on mobile money transfers people just texting small amounts of cash to each other to buy groceries or pay for services.

SPEAKER_00 14:15

So to a traditional bank, a street vendor running a cash-only stall is financially invisible.

SPEAKER_01 14:20

Completely invisible. But to Artha's AI, their daily mobile money receipts show a clear, consistent cash flow. It builds this mathematical picture of financial stability that a traditional credit bureau is completely blind to.

SPEAKER_00 14:34

That's incredible. And Vigat also brings up the integration of psychometrics, which takes this alternative risk profiling into incredibly nuanced territory.

SPEAKER_01 14:42

Yeah, psychometrics is where it gets really cutting edge.

SPEAKER_00 14:45

It sounds slightly like science fiction, honestly, using psychological data to determine if a person will pay back a microloan. How does a digital banking platform even measure a psychological trait?

SPEAKER_01 14:57

It does it by analyzing the microinteractions a user has with their device while they're applying for the service. Psychometrics in this context refers to the behavioral data generated during the digital onboarding process itself.

SPEAKER_00 15:09

We like how they hold the phone.

SPEAKER_01 15:13

With the user's consent, the AI can measure incredibly granular data points, either typing speed, the frequency of corrected mistakes, or even how long a person spends hovering over a specific question before answering it.

SPEAKER_00 15:25

Oh, I see. So if a user aggressively scrolls to the bottom of a complex terms of service document and hits accept in less than a second, the AI logs that interaction.

SPEAKER_01 15:36

Yes. And behavioral scientists and data analysts have actually mapped these digital micro patterns to broader psychological traits.

SPEAKER_00 15:44

Aaron Powell Really? Like what?

SPEAKER_01 15:45

For instance, if a user takes their time, reads the prompts, and carefully corrects typographical errors in their application. Statistically, that correlates with high conscientiousness.

SPEAKER_00 15:56

Aaron Powell And I'm guessing high conscientiousness is good for banking.

SPEAKER_01 15:59

Very good. In risk modeling, high conscientiousness strongly correlates with a much lower likelihood of defaulting on a loan.

SPEAKER_00 16:06

So what about the person who just scrolls to the bottom and instantly hits accept?

SPEAKER_01 16:10

Right. Conversely, erratic swiping and instantly clicking through critical financial disclosures can be mathematically correlated with high impulsivity, which might raise their risk profile.

SPEAKER_00 16:20

That level of granular analysis is staggering. It's so detailed.

SPEAKER_01 16:24

It is, but Pagat is careful in the interview to clarify that these alternative data points are not meant to entirely replace traditional underwriting.

SPEAKER_00 16:32

Oh, they're not?

SPEAKER_01 16:33

No, if we connect this to the bigger picture, the objective is to build holistic, explainable risk profiles in low data environments. The platform is designed to seamlessly integrate traditional credit data whenever it actually is available.

SPEAKER_00 16:48

So the alternative data, the telco patterns, the psychometrics, they're just used to illuminate the dark spots.

SPEAKER_01 16:54

Exactly. And the requirement that the AI be explainable is critical here. Regulators and the banks themselves, they cannot rely on a black box AI that simply spits out a yes or no.

SPEAKER_00 17:06

Right. They need to know why the loan was approved or denied.

SPEAKER_01 17:09

Exactly. The system must be able to demonstrate the specific combination of localized data points that actually led to a decision.

SPEAKER_00 17:15

Aaron Powell, which brings everything back to the core mission, I think. By utilizing this cloud native architecture, decoupling the AI from local compliance rules, and engineering the system to analyze these incredibly localized behavioral patterns, pure software is moving far beyond just providing software upgrades.

SPEAKER_01 17:31

Aaron Powell Oh, way beyond. They are structurally engineering financial inclusion at an unprecedented scale.

SPEAKER_00 17:37

Aaron Powell And they're actively anticipating the shift in the global market, right?

SPEAKER_01 17:41

Yes. But Gott really emphasizes their heavy investment in research and development and interoperability. They understand that in emerging markets, the transition from being entirely unbanked to demanding fully integrated digital financial services doesn't take decades anymore. Trevor Burrus, Jr.

SPEAKER_00 17:58

It happens fast.

SPEAKER_01 17:59

It happens in a matter of months. The platform has to evolve faster than the customers' expectations, constantly integrating new APIs and new data models just to remain relevant.

SPEAKER_00 18:08

Aaron Powell So what does this all mean? Why does a deep dive into an enterprise banking platform and the intricacies of emerging market credit evaluation matter to you listening to this right now?

SPEAKER_01 18:19

Aaron Powell It's a fair question.

SPEAKER_00 18:21

Right. Even if you live in a region where traditional credit scores are the unquestioned standard, this matters because the fundamental definition of financial trust is being rewritten globally.

SPEAKER_01 18:30

Yes. Trust used to be such a static concept.

SPEAKER_00 18:33

Aaron Powell Exactly. A paper trail locked in a filing cabinet or a three-digit number managed by a centralized credit bureau. But what platforms like Artha demonstrate is that the future of trust is dynamic. It's constantly updating.

SPEAKER_01 18:46

Yeah. It is built on modular cloud native architecture that is capable of securely processing our daily microbehaviors to find reliability and potential in places the traditional system was just never designed to look.

SPEAKER_00 19:00

And that transition toward AI-driven alternative risk profiling, it's basically inevitable at this point because efficiency and inclusion ultimately drive the market forward.

SPEAKER_01 19:10

But it does leave us with a complex reality to navigate as these systems become the global standard.

SPEAKER_00 19:16

It really does. I mean, think about it. If an artificial intelligence can now accurately assess our financial reliability based on our typing speed, the consistency of our phone habits and our daily digital interactions, it undeniably opens a massive door for global economic inclusion.

SPEAKER_01 19:31

Aaron Powell Which is amazing.

SPEAKER_00 19:32

It is amazing. Yet it also forces us to consider where the boundary lies between a helpful, holistic financial profile and comprehensive behavioral surveillance.

SPEAKER_01 19:41

Wow, yeah, that's a fine line. It is. As we step into a financial ecosystem where every digital microaction could potentially factor into our trustworthiness, it is something to weigh carefully the next time you tap I agree to share your data.

SPEAKER_00 19:56

A profound thought to end on as the architecture of our financial lives continues to evolve in the background. Thank you for joining us for this deep dive. We look forward to exploring the next topic with you soon.