Your Intelligent Risk Infrastructure

Lenders face a risk environment that grows more complex by the day. Fragmented credit profiles make it harder to understand a customer's liabilities. Digital fraud is accelerating as synthetic identities become even more difficult to detect. Regulatory expectations around transparency and compliance continue to rise. And, many institutions still rely on static underwriting models that slow decisioning, create operational friction, and leave costly blind spots across the risk lifecycle.

In this webinar, we explored how lenders can use data-rich artificial intelligence tools to reduce risk and make faster, smarter decisions without increasing friction for borrowers. You’ll learn:

• How AI can enable real-time risk signals, more accurate approvals, and clearer, more explainable decisions.

• Best practices to reduce fraud, improve approval rates, and build a more adaptive risk framework.

• How Oscilar and Spinwheel help lenders unify risk data, automate decisioning, and reduce fraud without adding friction to the borrower experience.

Read the Transcript.

I feel like I should give this shout out. So, Rob, I'm gonna look at you, Aarron, too. It's funny, you both have great beards going. I only, you know, have facial hair on my… on my mug.

You might be the only person to tell me I have good facial hair.

Thanks to the good folks at FinTech meetup.

Certainly better than mine, for sure. Uh, anyway, this is the perfect way to get started, folks. I'm excited because Aarron and Saurabh are probably both the smartest people I know when it comes to the discussion we're going to have today. But more importantly, they're probably some of the nicest people I know. And so I'm excited to have the conversation.

I'm excited to hear their insights and know that everybody, like I mentioned earlier, is going to walk away with tangible takeaways. And so we design. My name is Phil Goldfeder. I'm the CEO of the American FinTech Council. I'm very, very proud to have both Spinwheel

and Oscilar as members of the organization, but not just members, leaders within the organization, working with all of our membership in terms of driving the conversation about how we think about data, how we think about fraud, and how we think about so many important topics as it relates to the future of financial services, and not.

Where we once were in financial services, but where we need to go. And mostly because consumers are demanding it. And so, banks and fintech companies need the tools to get up to speed, but get up to speed and feel safe and confident about it. And so, we developed.

This series, uh, hopefully, you know, Saurabh, I always say at the end of every webinar that, you know, oh, this was great. I hope to do it again. You're one of the few that we've actually done this so many times together, and I'm excited to be here with you again. And Aarron, gosh, you know, you and I have been friends for a long time, and now having you as part of the series, as part of AFC is just been a really exciting for me.

Obviously, we love… we love Tomas and the entire team over at Spinwheel, but again, just getting to know you has been amazing for me and for the team at AFC. And so, today we're going to be talking about intelligent risk infrastructure.

Smarter decisioning. Less friction.

Equally sound when it comes to how we think about running, um, financial services and ensuring consumer, uh, consumer protections in all those ways. So, I'm gonna let you each introduce yourselves. Maybe Saurabh, starting with you, talk about what you do.

Specifically, maybe the work that you do as it relates to Oscilar and what you're bringing to the ecosystem.

Absolutely, Phil, thank you for the good words. About everyone in the ecosystem and feels very special to always collaborate with you once every quarter, for sure.

Um, so to give you a bit of a background, I'm sorry, I'm the chief product officer here at Oscilar. Oscilar is an agentic risk platform for decisioning across credit risk, onboarding risk.

fraud and compliance. And so what we have built is the compliance complete risk infrastructure structure stack, which allows you to kind of ingest any data, be able to kind of create decisioning engine and workflows for any use case and be able to kind of.

Have your ops investigation analytics team kind of review those, and so how do you bring that streamlined decisioning engine intelligent layer across all your customer lifecycle is at the core of what Oscilar does?

Amazing. Thank you so, so much, Saurabh. Aarron, you're the head of growth at Spinwheel. Like I said, and I told everyone, I've got the chance to get to know you probably even more personally than most, and I've really appreciated what you and the team at Spinwheel have done as it relates to data, and obviously your partnership with Oscilar. So why don't you tell everybody a bit about who you are and what Spinwheel does?

Thank you for the kind words, Phil. I'm Aarron. I lead partnerships and growth here at Spinwheel. We are a consumer permission platform operating within the consumer credit and infrastructure space. We help various institutions like marketplaces, PFMs, lenders.

Card issuers make smarter, more smarter, more intelligent decisions on the consumer. Right now, I'm focused on building strategic partnerships with, you know, Oscilar and the AFC that help lenders modernize their underwriting, improve risk, and have better outcomes for the consumers.

And one thing I love probably more than anything else is that when members come together, find ways to work together to collaborate. I think with trade associations, oftentimes there's this question about competition. But I think within the American Fintech Council, we find that that so many of our members complement the work that each other are doing, and when they can come together and offer an even better product and even better solutions, I think it's.

It's really truly what makes us happy in terms of the work that we do. And so, let's set the stage in terms of, like, what are the pain points that lenders are seeing today? How do we think about what those pain points are? More importantly, and.

Saurabh, we'll start with you. How do we solve them? And I think, especially, you know, sort of as technology continues to evolve.

As we come up with more and more solutions, fraudsters get more and more, you know, sort of.

more… they get better at the work that they're doing in terms of finding new ways to penetrate the system. So, let's talk about fraud. More importantly, I want to talk about the solution, but maybe talk about the problem first for a couple minutes.

I think if you take this in the context of the credit risk base, of course, as you are kind of looking at the credit risk journey, you're looking at 3 big components of evaluation. One is, of course, looking at the fraud risk. The second is compliance risk. And third, everything around credit risk. And as you look at the entire journey, there's falloffs in the credit lifecycle for.

Various reasons, right? Across these three sectors. In the old world, the entire credit risk.

tools, infrastructure, they were built on some assumptions that these applications are batch decisions. These models and policies that are being updated, they're probably quarterly refreshed.

best case. And there is some engineering gated policy changes. So if you have a strategy that you want to change, either as a fraud rule or a compliance rule or a credit rule, you have to kind of really go one product at a time, one rule at a time, get permission from engineering on terms of, like, you've built something, how do you test it, how do you deploy it?

And that is fundamentally quite slow, right? And there are four things that have kind of broken the model simultaneously, and the the first one is, of course, you see a lot of this embedded lending, making instead decisions across.

Not just consumer, but also, like, SMB credit opportunity, and so that scaled faster, and we are kind of seeing where people are asking for, how do we do more faster? I don't want to wait for days for decisions, even if it's a business, I want it to be done in ours, right? Uh, and so that's kind of one part of the problem. The second part of the problem is there's a lot of new data signals, right? And so if the.

fraud lifecycle, credit lifecycle, consumer lifecycle are becoming faster. How do you get more new data fast, right? Whether it's the real-time bank feed, student loan obligations, behavioral data. So, just the bureau-only view.

It looks like a bloody photograph of a moving target, and so that becomes a bit of a challenge, and that can pose any of the risks that we mentioned here. The third is around what the core aspects of people build tools for engineers, right? So they basically build developer productivity tools.

And so everything was like a code or a no-code, which still has code in it. And so, uh, the fundamental problem that exists right now is if you look at the developer side of the world with Claude and stuff, there's a revolution going on that has shaped how even developers are creating code, but the same hasn't come into the risk teams, and so they're fundamentally slower, right?

Um, and so that's kind of what is forcing and pushing everyone to think of, like, what is a modernization of the risk infrastructure mean in terms of the entire credit lifecycle of the risk, which includes fraud compliance, of course, credit. And so that's kind of where we are, and what we… what everyone is trying to look at is.

Yes, I do want to keep my delinquency less, I do want to keep my first-party fraud less, I do want to increase my approval rates. I do want sub-second consumer decisioning. I do want here real-time SMB or business lending kind of decisions, and so… How do I tie all of this together? Because now I don't want to do this for one product. I know I want to do it across my product, right? Portfolio. So that's kind of where we are. And I think that's kind of what is forcing everyone to say, like, things are fundamentally broken.

What is the new dimension for the next decade? And what does that risk infrastructure decisioning engine data orchestration, all of that look like in this new world?

Amazing. Aarron, he teed it up for you. So where does Spinwheel play in, right? Obviously, data plays a tremendous role in all of this, and so how do you work together, and how does that play in to sort of creating a solution?

Yeah, I think you hit a couple of the points, you know, directly on that like lending distribution has changed. Consumer experience is actually doubling as a risk tool now. And so I think where lenders used to have quite a bit of control, top of the funnel, think like direct mail prescreen. They don't have as much.

say or control in, you know, who's coming in the front door now, with everything being so embedded across the entire lending landscape. Partnerships like Spinwheel and Oscilar, I think we closed the gap. I think combining, you know, the better data between the two of us, the.

Instant risk decisionings, the compliance aspect from from Oscilar, real-time balances from the spin wheel side, and plus the ability to actually move money and disperse funds directly to liabilities, I think kind of changes the entire landscape as a whole.

Um, well, let me ask you this, and Saurabh, maybe the next obvious question is, you know, obviously, we always want to hear more efficient, we want to hear faster, but how do you sort of make people feel comfortable that we're moving faster, and arguably even more safe, right? We're moving more efficiently, and we're creating even more guardrails and creating a more.

sort of an ecosystem that is even safer for consumers and for those who are embracing the technology.

Oh, I think you're on mute, Saurabh. It happens to happens to everybody has to do it one time.

If you tend to… Just to make sure there's no noise from the kids in the background, right? I think the guardrails part of the question is very interesting, because the guardrails were the reason why we were slow. Right? If you look at the guardrails, everyone wants to say, like, are we doing the right things? Have we built the right policy? Have we tested the policy? How do we confirm that it's doing the right things?

And can we explain it? And of course, there's all the FCRA and all the compliance that you need to do from a perspective of credit models and decisioning. And so guardrails were built because there was less confidence around being able to go end-to-end around this, and all of the data was either stuck in some tables.

or required more specialized skills to do it, or needed some manual efforts to extract that and put it in explainable formats and confirm all of that.

All of these problems created more guardrails, and more guardrails meant more manual intervention, more people, more tasks, more types of tools. And so you kind of started having a death by not 100 tools, but at least maybe, let's say 10 tools.

And that is kind of the biggest problem. And if you look at these challenges of like today, can you think of automating a lot of these problems so that you don't need these guardrails? These guardrails are more of a… oversight that, hey, am I within this band of approvals? Am I making sure that every decision has an adverse action notice when it's declined? Am I making sure that credit memos that are generated have these types of things? Can I make sure that the drift of the models that are happening across the score bands, whether it's like consumer data, bureau data, I can understand those.

drift without having to do manual work? Can I kind of make sure that when I'm actually building a policy, I don't have to go to the engineers to figure out, can I test this? How do I test this? How do I backtest it? How do I AB test it? And that shouldn't require a lot of those dependencies either from the engineers or the vendors themselves. And so I think.

That's where you… the whole guardrails become more oversight guardrails versus like these manual interventions.

I think that again, I think that's fantastic. Aarron, let me come back to you and maybe shift to what everybody always loves to talk about, or at least the two letters that everybody likes to mention is sort of AI, right? Like, we talk about artificial intelligence and the growth of it. So how is AI enabled kind of real-time risk signals, more accurate approvals, and then.

maybe even tapping into your expertise and what you do every day is, how do you convince institutions that that it actually works, and it is sound, and it's the best path forward for creating a more durable and structurally sound infrastructure?

Yeah, I think you know there's a true balancing act with the institutions. They're looking for, you know, real time data, stronger decisioning. But then you also have to balance that with the consumer experience. Who wants transparency, instant decisioning. So I think, you know, with with the tools that we use and and how we're partnered with Oscilar for these institutions.

We're actually acting on behalf of the consumer, and so we get, you know, a specific set of permissions that we've gotten approved through the various, you know, partnerships that we have, where we actually act on behalf of the consumer to do the things that they would normally do, but then take it out of their hands.

So think, you know, go pull a real-time balance, so I can do a direct disbursement of funds if I need a consolidation loan, or if I want to get pre-qualified, or now that you know we're a CRA, lenders can actually use the information that we layer on top of the credit reports in order to make smarter, faster decisions.

I think that plus the, you know, the ways that we wrap up all of the consumer data and how we are compliant and how we're dispersing the information and, you know, tokenizing all of that, I think really puts at ease kind of what we're doing with the consumer data, but then also that the consumer is permissioning who they actually want to access their data.

Let me ask you, Saurabh, what have… what's evolved for you? What have you learned? What have you seen out in the marketplace that you've now taken back and have adjusted and then learned and gotten stronger because of? So what are the key kind of things that you've taken away over the last few years in this space?

Ooh, you're on mute.

Okay.

Couple of things. As you look at the what would drive a more seamless experience, faster iterations, better management of your risk. You have to start looking from ground up, like what are the challenges? And I think the first big challenge is.

Uh, the data orchestration, it's messy. Data is… you have internal data, right? Whether it's the application data, the data that they fill, the data that comes from all these different places, and how do you make a sense out of it, right? And then, of course.

When you have the permissions and you get the Spinwheel data, and now all of that data needs to be mapped with the data that you have. So data orchestration becomes a lot… very messy. So you have many data sources, you need those data sources to bring coherence, but how do you kind of structure it, orchestrate it, use it in the right way?

And how do you kind of really figure out the checks and balances of those? So data orchestration means not just having 100 data vendors on your marketplace. It means creating a knowledge artifact on.

For example, if we truly understand Spinwheel's data and their attributes and their modules, how do we help our customers drive intelligence off of that data? How does that data get used in different types of.

rules, credit policies, fraud policies, compliance policies, so… Like, making… connecting data is one, and creating knowledge on top of that is the second thing, and so I think the data orchestration and the knowledge base become like very important to activate. How do you extract the pieces of information? How do you create features on those types of data? And then how does those features get used in.

rules or models or within decision tables to kind of orchestrate things. So that's one part. The second part is the analytics layer is fundamentally broken. This one is very underappreciated, right? If you go look at some of the platforms out there, like, everyone will say, yeah, today we can do some kind of a backtest on a policy change. We can run an A/B test. So there are.

buttons in the UI where you can say, yep, backtest, AB test, etc. But… What's hard to understand is under the hood, there's a lot more problem. You need the data views to be there for someone to easily go and say, and when I do click that one button.

Are the data views already there for me to do the backtest, or will I have to go to my engineering team to say, hey, I need to prepare the last 30 months of last 3 months of data to be there, so that when I press this button, that that test can run, right? And so, uh… Either the platforms don't have this natively, or you need a data engineer to build the views before your credit analyst can actually run anything from the UI, and the analyst is always like one setup task away from being able to do this job, and that one-step task would take like.

a week, right? And so the analytics layer of what does it mean to.

natively, intuitively, without a lot of dependency, do backtest, shadow test, A-B test, confirm what you're doing, monitoring. I think that becomes a… those inline experiences are very hard, and you're still relying on data engineering or vendors to be able to kind of do professional services on those things.

And then this kind of opens up the broader engineering dependency, right? And this is kind of the multiplier. Every problem above gets worse because every fix requires like a ticket, a sprint, a review cycle, and.

You can't fix the data layer without engineering, and so you have to basically first fix the analytics part without… then you have to fix the policy part, you can't change a policy threshold without engineering, and so it's kind of becomes a product backlog, and people are managing a backlog, in my opinion, versus, like.

being adaptive on what it means to be managing a credit risk policy. And so, when we work with our customers, and if you see some of the case studies, people talk about, like, how do we get 2X faster, right? If it was taking us 3 weeks, 4 weeks to come up with, like, a policy, how do we do it in one week?

Right? And that's kind of the material advantage of thinking of this platform differently. And so that's kind of where I feel a lot of the learnings from how do you.

Get knowledge from the data. What does it natively mean to give insights and analytics so that they can move faster? And then what does it mean to provide the right tools so that there's less dependency on the engineering? Once you do all of that, yeah, of course it's like an AI experience. You just say something and it should happen, right? And that's kind of the model we're all trying to get to.

Well, thanks to your thanks to both of your work like it's easier like you you really again. You're building out the smart infrastructure. It was something that that Saurabh said, Aarron, that maybe I'll throw to you in terms of of data, right? Like.

A lot of data is not always helpful, right? Sometimes too much data delivered in the wrong way, you know, sort of calculated and managed in the wrong way, actually makes sort of the process in the back end harder. So how have you guys sort of, again, in your work together with Oscilar, sort of thought through ensuring that.

that you're… you're actually creating opportunities to get increased data, right? Additional data points, you know, to sort of help on the risk decisioning, um, and ultimately build, you know, strengthening the infrastructure.

Yeah, I think, you know, the partnership really boils down to, you know, Oscilar is providing the AI and the risk decisioning engine. So the intelligence behind this, and we help provide the real-time liabilities and the infrastructure together. Those 2 pieces kind of help create the entire picture for the financial institutions that we work with, so they're not actually relying on stale reports, disconnected systems.

You know, you don't have to do a huge rip and replace. We can actually work within some of the existing legacy infrastructure that they have already to then just help them provide and streamline any of these.

You know, decisions that, you know, a traditional credit risk analyst would need to do, as Rob mentioned, over two, three weeks, we can actually do that now in a week, and provide a little bit cleaner and better signal.

Um, I probably should have mentioned at the top, folks, as per always, and any and all webinars, feel free to drop questions in the chat. We're happy to sort of look at your questions and respond in real time, if possible.

And so, Saurabh, like, again, we… there's so much talk about innovation in financial services. Obviously, financial services heavily regulated space, right? And so ultimately, how do you balance that when you're sort of thinking about AI solutions, you're thinking about increasing sort of the.

The tech stack when it comes to infrastructure and risk.

in risk and compliance, how do you think about the balance between also staying in compliance with the regulatory, uh, regulatory demands that are required for financial institutions?

The same pattern follows. I think the the regulatory demands are definitely a lot more intense in the credit space than the payments fraud space. Payments fraud. It's kind of accepted that you're going to move fast, break things.

iterate fast. In credit, you're making decisions that materially affect how credit gets distributed to people. And so there's a lot more regulations around.

the decisions you make and how you're going to kind of have fair lending act, how you have bias, uh, determination on models, how you kind of look at adverse action notice, and BFC, are they compliant? And a lot of those types of data decisions that you make.

And that starts with kind of making sure that there is absolute transparency, right, for every data that comes in, how it gets used across the lineage of every rule, every model, every outcome, every decision can be backtest, right? That's number one.

And so that becomes the fundamental layer of making sure, like, are you… do you understand the data that you're using? Do you understand how you're using it? Do you understand the outcome of it? Right? So the data lineage becomes, like, the first problem to kind of.

As a foundation to make sure that there is everything built on top of it can use this. The second is, what tools do you have to confirm that, you know, you are making the right decisions? And I… we touched on this a bit before when it comes to things like backtesting, shadow testing, A-B testing.

Doing those tests across the distribution of your customer segments and control groups and protected, let's say, attributes. Like, how are those distributions happening across these different protected attributes to know that there is.

Uh, no bias. The errors are distributed equally across these segments, right? And so there is the fairness aspect of those things that are kind of looked at in the platform natively. And so that, I feel those tools are very important as you go through every cycle. What is the automation that is required for you to be able to do these backtests and be able to say.

I understand the decisions, I understand the decisions, not just at a level of Saurabh or Phil, but I understand this. How are these decisions being done at a peer cohort level, and that there is no bias in.

Yep.

I can understand I'm using the data that is only explainable, right? And so that goes to the third part, which is, how do you create explainability natively in the platform? Right? Because at the end you're kind of like.

There's an input, there's a lot of policies, and then there is output, which is a decision. But what does it mean to actually create explainability? It becomes very important? Is the explainability just a decision? Is it a decision with a reason code?

Is it a decision with a reason code, with a credit memo? And so how do you kind of really look at that? How do you build the right parameters of it?

If the tool doesn't natively allow you to construct all of this in a very automated way, then of course those will have to… those become challenging, right? And so I think when you think of all the aspects of governance, regulations, it becomes important that.

These fundamentals of traceability, testing, explainability become the foundations of the platform and how it's kind of embedded in the capabilities and features.

Aarron, as the head of growth and partnerships at Spinwheel, my guess is that's the first question, right? Like, you know, and again, I asked you a variation of this before, but I think maybe as part of as we close, and you think about next steps, is sort of what is your response, right? Everything Saurabh just said.

It sounds great. It's like, okay, we're thinking about it. We understand it, but then how do you actually convey that to potential partners as you think about their embracing your technology?

Yeah, I think you know the lending landscape tends to move, you know, quickly and slowly at the same time. There's a big dependency on, you know, the human element of review. And I think that there's.

slight aversion to kind of shifting everything into the AI landscape. So I think, you know, as time evolves, it's certainly getting there, and it's getting there very quickly. I think partnerships, like what we're doing with Oscilar, where we are creating the full transparency for the lender, and then ultimately the consumer at the end, you know.

They receive an adverse action, it's left intentionally vague, which doesn't really tell the consumer anything about what they did wrong or how they can get approved in the future. And so I think, you know.

Building on those blocks for the future and how we can actually be transparent to the consumers and to the institutions on how and why we're using those data. I think more and more of these lending institutions are actually getting more comfortable with it because they understand the landscape is shifting.

Consumers need transparency. They need to know how they can get access to capital, who's using their data? How can they delete their data with all of the, you know, the breaches that have happened through the various credit bureaus. I think institutions like, you know, Spinwheel and Oscilar are really kind of fighting against that.

To make sure that the data is back in the consumer's hands.

The reason I love talking to both of you and spending so much time with both Spinwheel and Oscilar is because you historically always been ahead of the curve. Both your companies have sort of seen ahead of the curve, sort of seen what's coming, and I don't have to tell you, when it comes to AI, when it comes to open banking and data privacy, those laws are ever-evolving.

Knowing that you have partners in Spinwheel and Oscilar who are seeing that, who are engaging with those regulatory structures long before they even hit the paper is critically important, I think, for potential partners to know. So with that, Sarah, but maybe how do people get in touch with you? What's the best way to get in touch with Oscilar?

Go to our website. You will definitely find a lot of resources, content to kind of look at, download.

Of course, you can contact us directly through the website as an inbound, but of course you can always try to reach out to one of our events. We do a lot of Riskcon conferences across different cities.

Please look forward to invites for those as well.

Aarron, same for you. Again, I know that you… I've been fortunate enough to sit in on a number of demos to see your work real time. What are the best ways for folks to get engaged and to reach out?

Yeah, very similarly, you go to our website, spinwheel.io. Feel free to reach out to me directly, Aarron at Spinwheel. Our PR and comms team is doing a rock star job at putting out a bunch of content about the new products and features that we have recently. And then, obviously, we'll be hosting some more events with you, Phil.

And, uh, and the rest of the Oscilar team.

Yeah, and I know they're both very active on LinkedIn, so don't be shy. I hit them both up very, very active. Again, I want to just thank Aarron Dumm over at Spinwheel and Saurabh, who's, again, a repeat offender with, uh, with us on these webinars, and just really always a pleasure. I feel like every time I get on a webinar with you, I'm learning something new myself. And so.

Really appreciate the work that Oscilar is doing and excited for more of these conversations with additional guests. So, like I said, we can continue to evolve in this ecosystem and be sure that people who are thinking about becoming more innovative, embracing a different infrastructure, a different viewpoint on infrastructure.

have the absolute best tools, like those at Oscilar and Spinwheel are offering. Thank you all very, very much for being here.