IBS Intelligence Global FinTech Interviews
Go one-on-one with the innovators, disruptors, leaders, and decision-makers driving change in FinTech and financial services. IBS Intelligence delivers exclusive global interviews that uncover strategies, challenges, and the ideas powering the next wave of financial technology.
IBS Intelligence Global FinTech Interviews
EP978: The Importance of Reliability and Resilience
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
This interview with Yethi Consulting CEO Srirang Srikantha explores the critical role of quality assurance in the modern banking sector. Srikantha highlights how digital transformation has heightened customer expectations, making application speed and reliability more vital than traditional financial metrics for maintaining loyalty. He cautions financial institutions against over-relying on software vendors for system integrity, suggesting that independent testing is necessary to ensure flawless performance. The discussion further examines how full-stack observability and artificial intelligence can help manage the growing complexity of banking technology. Ultimately, the text presents a case for proactive performance validation to prevent costly technical failures and protect a bank’s reputation.
Have you ever uh had that very specific moment of panic? You know the exact scenario I'm talking about.
SPEAKER_01Well, yeah, I think we all do.
SPEAKER_00You're just like standing in line at a grocery store, or maybe you're just sitting on your couch, right? And you open your banking app on your phone to quickly check your balance.
SPEAKER_01You're just expecting the money to be there instantly.
SPEAKER_00Right. Instantly. But instead the screen just goes white. Or uh there's just that tiny little spinning circle right in the middle of your screen.
SPEAKER_01Skimming and spinning, yeah.
SPEAKER_00One second goes by, then two, then three. And in that tiny, I mean, seemingly insignificant delay, your brain completely betrays you.
SPEAKER_01It really does.
SPEAKER_00You feel the sudden, completely irrational spike of adrenaline. Your mind just races to the worst-case scenario, like, was I hacked? Did someone steal my identity?
SPEAKER_01Is my money just completely gone?
SPEAKER_00Exactly. It is wild how, you know, three seconds of a blank screen can just completely unravel our entire sense of financial security.
SPEAKER_01Yeah. It's a very visceral reaction, honestly. And uh it captures perfectly how our relationship with financial institutions has fundamentally shifted. I mean, we no longer separate the digital interface from the institution itself.
SPEAKER_00Which brings us to today. Welcome to today's deep dive. We are exploring the invisible, highly fragile architecture of modern banking. It's a fascinating topic. It really is. Our source material today comes from a really insightful interview published in the July 2025 edition of the IBS Psy FinTech Journal.
SPEAKER_01Right. The sit down with Shurang Shirkata.
SPEAKER_00Yes, exactly. He's the CEO of Yeti Consulting. And they are this massive firm that specializes in quality assurance and uh application reliability for banks.
SPEAKER_01Right, doing a lot of the heavy lifting behind the scenes.
SPEAKER_00Right. So the mission of this deep dive is to understand why customer loyalty, which banks used to measure in you know decades, is now being measured in literal milliseconds.
SPEAKER_01Literally milliseconds. It's crazy.
SPEAKER_00It is. We're gonna look at why this massive push for digital transformation is exposing cracks in the financial system faster than ever before.
SPEAKER_01Aaron Powell To set the stage for this, we should really look at how Robin Amlott framed the interview. Uh Amlot is the managing editor of IBS Intelligence.
SPEAKER_00What did she say?
SPEAKER_01He stated, now more than ever, your bank could be just one bad transaction away from loss of business and loss of value.
SPEAKER_00Aaron Powell Wow. One bad transaction.
SPEAKER_01Aaron Powell Yeah, that gets right to the core of the issue. In our current environment, a bank's reputation is inextricably tied to its technology.
SPEAKER_00Right. They aren't separate things anymore.
SPEAKER_01Exactly. If the technology hangs or fails, the bank itself fails in the eyes of the consumer.
SPEAKER_00Aaron Powell Okay, let's unpack this because we really need to understand this entirely new metric of trust. Shrikantha points out a reality regarding customer patience that is just incredibly unforgiving.
SPEAKER_01It really is. People have no patience anymore.
SPEAKER_00None. If you open a banking app and it loads your accounts in, say, two seconds, you feel secure.
SPEAKER_01You're right.
SPEAKER_00It feels responsive and safe.
SPEAKER_01Sure.
SPEAKER_00But according to the data, if that exact same app asks you to wait for seven seconds, just five extra seconds, the user automatically assumes there's a systemic glitch.
SPEAKER_01Seven seconds.
SPEAKER_00Seven seconds. That is the threshold where a customer actively starts doubting the integrity of the actual bank.
SPEAKER_01Aaron Powell What's fascinating here is that creeping doubt is highly destructive to a bank's core business model. I mean, Shrikantha notes that a new generation of customers is making decisions about where to park their money based on entirely different criteria.
SPEAKER_00Aaron Powell Right. They aren't looking at the same things their parents did.
SPEAKER_01No, not at all. They aren't pulling up the bank's tier one capital ratio to see if the institution has enough core equity reserves to survive a financial shock.
SPEAKER_00Right. Nobody's doing that on their phone.
SPEAKER_01Exactly. They aren't reading quarterly financial statements to verify the strengths of the actual balance sheet. They are evaluating the app.
SPEAKER_00Just the app.
SPEAKER_01Yeah, they want to know if the app is always available. And frankly, they want to know how sleek the user interface is.
SPEAKER_00How cool it looks.
SPEAKER_01Exactly. The actual financial reliability of the business has become totally secondary to the digital experience.
SPEAKER_00I mean, that is a massive shift in human behavior. Historically, a bank proved it was safe by building, you know, a massive intimidating piece of architecture right in the middle of town.
SPEAKER_01Like those giant stone buildings.
SPEAKER_00Yeah. Giant marble columns, thick glass windows, and a heavy steel vault that you could physically see. That design was meant to communicate permanently.
SPEAKER_01Right. It said we aren't going anywhere.
SPEAKER_00Exactly. So it sounds like a banks app interface is basically the modern equivalent of that heavy steel vault. I mean, are are we really out here judging systemic financial stability by server load times?
SPEAKER_01We absolutely are. And the reason why is that in the digital age, perception dictates reality.
SPEAKER_00Perception is reality.
SPEAKER_01Right. To the user sitting on their couch, the interface isn't just a window into the bank. The interface is the bank. Wow. If it's clunky or slow, the user implicitly believes the institution itself is clunky and unreliable. Performance is synonymous with institutional reliability. But uh that creates a glaring paradox, doesn't it?
SPEAKER_00Well, if app speed and slick interfaces are the new gold standard, why are so many banks failing to deliver it?
SPEAKER_01Uh right.
SPEAKER_00With billions of dollars pouring into digital transformation across the financial sector, you'd assume these systems would be lightning fast. Are they just, you know, building bad technology?
SPEAKER_01Well, no. It isn't necessarily that the new technology is inherently bad. Shrikantha argues that digital transformation isn't creating new problems out of thin air.
SPEAKER_00Okay, then what is it doing?
SPEAKER_01Rather, it's removing the historical time buffers that banks use to hide behind.
SPEAKER_00Time buffers.
SPEAKER_01Yeah, think about the mechanics of banking 15 years ago. If you needed to transfer money, you accepted the friction.
SPEAKER_00Right. You had to actually go to the bank.
SPEAKER_01Exactly. You drove to a physical branch, you stood in line, and you waited 20 minutes for a teller. Furthermore, you only checked your financial standing once a month when a paper statement arrived in the mail.
SPEAKER_00Because the systems weren't real time back then. Right. Tellers batched transactions, and the bank ran these massive updates on their mainframes overnight.
SPEAKER_01Precisely. That overnight batch processing and those 30-day statement cycles meant the bank had hours or even weeks to quietly fix internal errors.
SPEAKER_00Oh, I see.
SPEAKER_01They could reconcile mismatched accounts or correct computation mistakes completely out of sight. You, the consumer, never knew there was a problem because the bank fixed it behind the scenes before you ever saw the ledger.
SPEAKER_00And today that buffer is completely gone.
SPEAKER_01Entirely gone.
SPEAKER_00We have real-time transaction-level transparency. When you hit transfer, an API like an application programming interface instantly calls the bank's database. If that database is locked up or struggling to compute, it doesn't just happen quietly in the background anymore.
SPEAKER_01No, it's front and center.
SPEAKER_00It's highlighted instantly on your screen as a spinning wheel or an error message. The customer sees the raw data processing in real time.
SPEAKER_01And they panic.
SPEAKER_00Right. And the source material gives a jarring anecdote about how badly this lack of a buffer can actually play out. Yeah. Shrikantha details the situation with a specific customer of Yethi. This bank decided to completely overhaul their system.
SPEAKER_01Right. They chose a new architecture that put their customer relationship management system, their CRM, at the absolute center of their operation.
SPEAKER_00It's just a huge change.
SPEAKER_01It is. But here's the kicker they rolled it out without actually validating how the architecture would perform under a heavy load.
SPEAKER_00And the consequences of skipping that validation were just severe.
SPEAKER_01I can imagine.
SPEAKER_00By putting the CRM at the heart of the system, it meant every single time a user tried to do anything like check a balance, send money, view a transaction, the system had to route through that customer profile database first.
SPEAKER_01Aaron Powell So it became a massive bottleneck.
SPEAKER_00Exactly. According to the interview, within 12 months of launching this architecture, the response time on their mobile app plummeted from a snappy 1.5 seconds down to an agonizing eight seconds.
SPEAKER_01Oh man, which firmly crosses that seven-second threshold into the the bank is broken territory.
SPEAKER_00But what happens next is the real tragedy of this anecdote. I mean, the bank spent heavily on this digital transformation to modernize, right? To push customers to the app. Right. But because the app took eight seconds to load, it eroded digital trust. Customers abandoned the app entirely.
SPEAKER_01They just gave up on it.
SPEAKER_00Yeah. And where do frustrated people go when they still need access to their money?
SPEAKER_01Right back to the branch.
SPEAKER_00They drove right back to the physical branches. They shifted the entire digital workload onto human tellers. The physical branches were suddenly overwhelmed by a massive influx of angry customers.
SPEAKER_01It backfired completely.
SPEAKER_00It literally destroyed the physical branch experience, too, which just completely defeats the purpose of the app.
SPEAKER_01It really illustrates how digital transformation is a double-edged sword. I mean, it offers incredible convenience, but it demands unrelenting responsiveness.
SPEAKER_00Yeah, people expect perfection.
SPEAKER_01It leaves zero room for error. Think about it. When a customer is standing in a branch, a human teller can smile, apologize for a slow computer, and de-escalate the tension.
SPEAKER_00Right. A little human empathy goes a long way.
SPEAKER_01Exactly. But an app cannot apologize. It just hangs. Making post-launch corrections to these digital platforms is incredibly expensive, not just in terms of actual IT infrastructure costs, but the severe damage to the brand.
SPEAKER_00So this brings up a major question about accountability then.
SPEAKER_01Definitely.
SPEAKER_00If rolling out an untested CRM can take down your app and flood your physical branches, why didn't the company that built the CRM test it first?
SPEAKER_01So Shrakantha points out a dangerous misconception about that, particularly among smaller community banks in the U.S.
SPEAKER_00What's the misconception?
SPEAKER_01These institutions often operate under the assumption that the vendor, you know, the company that's sold on the platform owns the testing.
SPEAKER_00They just assume it's handled.
SPEAKER_01Right. Their attitude is I trust the vendor. They expect them to deliver working software.
SPEAKER_00Aaron Powell, so what does this all mean? Shakantha views this reliance on the vendor as risky, really.
SPEAKER_01Oh, very risky. He's very direct about this dynamic. We have to look at the incentives here. The OEM, the original equipment manufacturer who built the software, their primary goal is to sell the platform, deploy it, and move on to the next client. Aaron Powell Right.
SPEAKER_00They want to close the deal.
SPEAKER_01Aaron Ross Powell Exactly. But the bank's primary goal is to protect its reputation and its depositors' funds. Those incentives do not align at all.
SPEAKER_00Aaron Powell I mean, it sounds like a bank buying a newly built house and just taking the builder's word that the plumbing is flawless.
SPEAKER_01Aaron Powell That's a great way to put it.
SPEAKER_00Like you wouldn't just skip the independent home inspection because the builder promises the pipes are fine, right? If you wouldn't blindly trust the builder of a house, why would an institution blindly trust a software vendor with billions of dollars in consumer deposit?
SPEAKER_01You need the independent inspector precisely because they didn't build the house.
SPEAKER_00Exactly.
SPEAKER_01They have no financial incentive to hide a leaky pipe or pretend the structural flaw doesn't exist. Farantha stresses that an independent quality assurance team is vital, but their role is often really misunderstood in the industry.
SPEAKER_00Misunderstood how?
SPEAKER_01Well, they aren't just brought in to hunt for isolated bugs or to cynically prove the vendor made mistakes. The real task of professional QA is a meticulous, exhaustive process to affirmatively prove that everything works.
SPEAKER_00Prove that everything works. I mean, that is a massive mandate. How do they actually do that?
SPEAKER_01It requires verifying the software across multiple dimensions simultaneously.
SPEAKER_00Okay, like what?
SPEAKER_01First is the functionality correct. If you hit send, does the money actually move from account A to account B without vanishing into the ether?
SPEAKER_00Aaron Powell That's pretty important. Yeah.
SPEAKER_01Extremely. Second is the performance acceptable under stress. They simulate millions of concurrent users logging in on payday to ensure the system still responds in 1.5 seconds instead of eight seconds.
SPEAKER_00Aaron Powell To avoid that CRM disaster.
SPEAKER_01Exactly. And third is the security airtight. It's a rigorous process to establish true accountability.
SPEAKER_00Aaron Powell But proving that everything works must be exponentially harder today than it was, say, 20 years ago.
SPEAKER_01Oh, without a doubt.
SPEAKER_00In the past, a bank might have just had one massive centralized mainframe doing all the math. But today, the underlying technology isn't just a single system anymore.
SPEAKER_01No, it's highly fragmented.
SPEAKER_00It's a tangled web of integrations. Srikantha actually notes that the number of technology components required to run a bank has been compounding rapidly for two decades.
SPEAKER_01The complexity is staggering. We are talking about layers upon layers of distinct technologies communicating with each other.
SPEAKER_00Right.
SPEAKER_01You have the user interface on the mobile phone. Beneath that, you have the application stack, the microservices handling specific tasks, the operating systems, the databases holding the actual ledger, and the physical network architecture transmitting the data.
SPEAKER_00Aaron Powell It's a lot of moving parts.
SPEAKER_01And it gets more complicated because banks want to offer innovative features, they constantly plug in specialized third-party platforms.
SPEAKER_00Aaron Powell Right, to stay competitive.
SPEAKER_01Yeah. They integrate one API for customer onboarding, a different API for loan origination, and another entirely separate system for fraud detection.
SPEAKER_00Aaron Powell And every new integration is a new risk.
SPEAKER_01Aaron Powell Exactly. With every new integration, the number of potential failure points increases.
SPEAKER_00Aaron Powell And when a system that complex breaks, it isn't obvious where the break actually happened, right? Not at all. If your transaction fails, did the app freeze? Did the fraud detection API time out? Did the legacy mainframe drop the connection?
SPEAKER_01It could be any of them.
SPEAKER_00Right. So to solve this, Rikanta introduces this concept of full stack observability. He describes taking all of these highly complex, disparate elements and bringing them into a single pane of glass.
SPEAKER_01Aaron Powell A single pane of glass, yes.
SPEAKER_00Aaron Powell But how does a single pane of glass actually track a transaction through all those different layers?
SPEAKER_01Aaron Powell Well, full stack observability essentially attaches a digital tracking tag to a user's action and watches it travel through the entire nervous system of the bank.
SPEAKER_00Okay, so it follows the data.
SPEAKER_01Yes.
SPEAKER_00Yeah.
SPEAKER_01From the moment you tap your screen, observability tools monitor that specific data packet as it hits the API gateway, travels to the credit check service, and finally reaches the core database. That makes sense. So if the transaction hangs, the IT team doesn't just get a generic error message. They can look at that single pane of glass and see exactly which layer in the stack caused the bottleneck.
SPEAKER_00Here's where it gets really interesting. Because it isn't just IT jargon, it translates the technical failure into actual business context.
SPEAKER_01Aaron Powell Yes. And if we connect this to the bigger picture, that is the crucial distinction. Observability answers the so what of a system crash.
SPEAKER_00The so what?
SPEAKER_01Right. If a traditional dashboard flashes a warning that, say, server four CPU is running at 100%, that is just a technical metric. Trevor Burrus, Jr.
SPEAKER_00Right. It doesn't mean anything to the CEO.
SPEAKER_01Aaron Powell Exactly. It doesn't tell the bank executives what is actually happening to the business. Full stack observability puts that metric into context.
SPEAKER_00Aaron Powell, so what does it tell them instead?
SPEAKER_01It tells the bank server four is overloaded. And because of that specific server, our retail payment processing is currently down, or our new account onboarding flow is completely stalled.
SPEAKER_00Oh wow, that's a huge difference.
SPEAKER_01Aaron Powell It really is. It allows the bank to triage a crisis based on the actual impact to the customer rather than just chasing blinking red lights and technical alarms.
SPEAKER_00But honestly, if the underlying architecture is layering on top of itself and becoming this incredibly complex, I mean, human QA teams can't possibly test every single pathway manually anymore.
SPEAKER_01Aaron Powell No, it's beyond human scale at this point.
SPEAKER_00Right. Nor can human eyes constantly monitor that single pane of glass for every potential anomaly across thousands of servers. So at a certain point, doesn't a bank just have to hand all of this testing and observability over to artificial intelligence.
SPEAKER_01That's the million-dollar question.
SPEAKER_00Aaron Powell Because there is constant hype right now that AI is going to write the code, test the code, and basically make human engineers totally redundant.
SPEAKER_01It is the natural question every industry is grappling with right now. But Shrikantha addresses this directly, and he provides a much needed reality check on what AI can actually do in this space.
SPEAKER_00Aaron Powell Oh, really? What does he say?
SPEAKER_01He acknowledges that the technology is highly capable, of course. I mean Yeti Consulting utilizes AI extensively. Roughly 25 to 30 percent of their audit design, the actual writing of the test scripts, is now generated through AI. That's a significant chunk. Aaron Powell It is. They've also taken the mundane, repetitive daily tasks that used to consume hundreds of human hours and automated them with bots.
SPEAKER_00Okay, so it is replacing some human effort.
SPEAKER_01Yes. But despite the global industry hype about AI autonomously navigating systems and figuring out what works, Strakantha firmly states that AI cannot operate in complete isolation.
SPEAKER_00Wait, why not? If an AI can scan millions of lines of code and spot a numerical anomaly in a millisecond, why does it still need a human looking over its shoulder?
SPEAKER_01Aaron Powell Because spotting a technical anomaly is entirely different from understanding the consequence of that anomaly in the real world.
SPEAKER_00Okay, that makes sense.
SPEAKER_01AI is a pattern matching engine. It might see a sudden spike in database timeouts and flag it as an error, but it takes a human being with domain expertise to look at that flag and understand the actual context.
SPEAKER_00Aaron Powell Like what kind of context?
SPEAKER_01Well, the human knows that it's the 15th of the month. A major local industry just processed his payroll, and the system is actually operating exactly as expected under a localized heavy load.
SPEAKER_00Ah, so it's a false alarm that only a human would recognize.
SPEAKER_01Exactly. AI lacks the live reality to understand how a technical glitch actually impacts a mother trying to pay for groceries or a small business waiting on a crucial loan approval.
SPEAKER_00Aaron Powell So you're basically clarifying the core difference between raw data and context. AI might be able to spot an anomaly in a millisecond, but it still takes a human to understand if that anomaly actually impacts the banking customer's experience.
SPEAKER_01Aaron Powell This raises an important question about the balance of tools versus expertise.
SPEAKER_00Aaron Powell Right, because Yeti operates at a massive scale, don't they? Trevor Burrus, Jr.
SPEAKER_01Huge scale. They work with 140 banks across 35 countries, and they have expertise covering over 300 distinct banking applications.
SPEAKER_00Wow. You can't just manage the cultural and regulatory nuances of 35 different countries with just an algorithm.
SPEAKER_01Not a chance. AI can write a test script to see if a password box functions, sure. But it doesn't know what an intuitive mortgage application feels like to a first-time home buyer.
SPEAKER_00Which really highlights the necessary balance here. AI is an incredibly powerful mechanism to help humans process data at scale and hold software vendors accountable.
SPEAKER_01Yes, accountability is key.
SPEAKER_00But it is the human understanding of the banking domains, the regulatory environment, and the ultimate business goals that makes the tool actually effective.
SPEAKER_01Exactly. The AI does the heavy lifting on the data processing, but the humans provide the compass to interpret what that data means for the end user.
SPEAKER_00Well, we have covered a significant amount of ground today.
SPEAKER_01We really have.
SPEAKER_00We have traced the evolution of financial trust from a physical reality, you know, standing in a marble lobby looking at a thick steel vault and reviewing paper statements once a month into a purely digital reality.
SPEAKER_01It's a completely different world now.
SPEAKER_00It is. We're now operating in an environment where a mere seven-second delay on a mobile app is enough to severely damage an institution's reputation.
SPEAKER_01And cause them customers.
SPEAKER_00Exactly. Digital transformation has completely removed the time buffers banks used to rely on, and it demands absolute real-time perfection. And we've seen how blindly trusting software vendors to test their own products is just a recipe for flooded physical branches and angry customers.
SPEAKER_01Banks must rely on independent quality assurance to prove everything works.
SPEAKER_00Right, everything.
SPEAKER_01They need full stack observability to untangle the massive complexity of their systems and a hybrid approach where AI tools empower human banking experts.
SPEAKER_00Absolutely. So the next time you open your banking app and it flawlessly loads your balances in a snappy two seconds, take a moment to consider what is actually happening behind the scenes.
SPEAKER_01There's a lot going on.
SPEAKER_00There is. Appreciate the intense, independent quality assurance, the real-time observability, tracing your data through a maze of legacy mainframes and modern microservices.
SPEAKER_01All of it working together.
SPEAKER_00And appreciate the hybrid AI human testing operating quietly behind that single pane of glass, just to keep you from feeling that sudden spike of panic.
SPEAKER_01You know, as we wrap up, I want to leave you with a final forward-looking thought built on the source material.
SPEAKER_00I'd love to hear it.
SPEAKER_01Right now, modern banks are judged almost entirely by the visual aesthetic and the two-second load times of their mobile apps.
SPEAKER_00Right. The screen is everything.
SPEAKER_01But technology is continually evolving. What happens when banking becomes completely invisible?
SPEAKER_00Invisible.
SPEAKER_01Yes. We are moving rapidly toward a future where smart devices or autonomous AI assistants manage our transactions entirely in the background without us ever looking at a screen at all.
SPEAKER_00Oh wow.
SPEAKER_01So in a world where the interface disappears, will millisecond speed still be the ultimate metric of loyalty? Or will we need to invent an entirely new, unseen way to measure financial trust?
SPEAKER_00That is wild to think about. I mean, if your digital assistant is doing the waiting for you, you might never even notice the seven second delay.
SPEAKER_01Exactly. The metrics of trust will have to change again.
SPEAKER_00It is a fascinating question to consider the next time you find yourself staring at that little spinning wheel. Thanks for joining us on this deep dive.