Fintech Insider Podcast by 11:FS
00:06:57 · Laura Watkins, Pav Gill
[00:06:57] S0: Mhmm. [00:06:58] S1: And that's compounded in a few ways, right? You have departments that are kinda like built around the software that they use, and then you have people in different departments that are cross functionally working together, but not necessarily communicating, and then you have all these different solutions that they're using that also not speaking to one another. So I think that problem is just gonna get worse, and that's why we're trying to bring everything in house. The other bigger problem is the connecting the dots isn't there, it's very manual. Right? You still have very, very advanced or big companies relying on Jira and Slack, on Excel spreadsheets just to do investigations, for example. Right? This shouldn't be the way we are working in 2026 going on 2027. I think if you can have a place which can connect the dots, you can see what's going on between events, cases, people, subsidiaries, brands, everything. I think that's just a more intelligent way of doing it. You could literally speak for example, today to confide, and ask it like, how many bribery cases you had in the last ten months, how you dealt with them, or you know, what your audit universe looks like, and it just churns it out in seconds, right? But imagine if a regulator asked you that question, you would be scrambling through people, meeting notes, I don't know, like it's just a big mess. And so I think that's a very clear and obvious problem, and we are really trying to solve it.
00:08:25 · Laura Watkins
[00:08:25] S0: Wonderful. Yeah. I'm just sort of putting everything in one house and say that all the, you know, metaphorical scraps of paper aren't just lost in the And and just I know that's horribly simplifying what you've built, but essentially that is solving an enormous problem just in and of itself, the the the organisational piece. Maya, you were you were kind of nodding along as as Pablo was kind of explaining the sort of problems facing organisations. Obviously, in product yourself. Sort of what's your take on this and how sort of needed something like this can be? Yes.
00:08:56 · Maia Bittner, Laura Watkins
[00:08:56] S3: I think the opportunity is huge for compliance systems. I think the risk is also obviously really obvious. Right? We know that AI systems can hallucinate. That could be very dangerous in the governance context. We know those risks. But for a really purpose built tool, I think that it can mitigate those risks and bring so much value. So just... I know. I was nodding along because the investigations, every... There's so much manual work that AI is really well suited for. Right? It is not to make the big strategic risk trade off decisions, but automating all of that manual work, it's really, really good at things like that. And I think sometimes when people have designed old systems, they'll say things like, oh, well, we should have an audit trail of everything that everyone does. That sounds great. It'll be re... Right. Audit trail. Wonderful. But then when it's time to actually conduct an investigation and you were trying to dig through records of everything that everyone has ever done, it's really hard to make heads or tails of it. But AI is so good at turning it, like, through all of that data, finding what's important, finding those patterns, pulling out the dot piece. I think it could really speed up some investigations in particular. [00:10:11] S0: And, Vijay, you know, as a a sort of head of business innovation yourself with sort of, you know, how is the use of AI in businesses changing the way that you're sort of thinking about governance as you set up your business and your business expands and grows?
00:10:27 · Vijay Konduru
[00:10:27] S2: One of the most important things is, like, I believe that agentic AI sits at a very interesting knife's edge or at a really interesting paradoxical situation. Like Maya mentioned, the promise of what AI can deliver is enormous, and that's why we as a bank are investing so much in AI and agentic driven innovation. But by very nature, the models that these agents are built on are nondeterministic and probabilistic systems. Right? And that intrinsically introduces some kind of variability. The way we're thinking about this whole growing ecosystem is really threefold. The first one is to engineer whatever products we build for trust with the right robustness, with the right orchestration, guardrails, and evaluation frameworks and appropriate risk management built in. The the second is to make sure that not every agentic workflow or process, whether it's for a consumer or for another business or even for an enterprise, has to lead towards full autonomous execution. Right? The use case determines where and when autonomy plays a big role. And then the third big principle is we believe that given the state of AI and agentic AI, the human in the loop is absolutely critical. Right? And the human's gotta be more in the loop from validating the outcomes and outputs of these models and approving actions and so on as the consequences get more critical and as the workflows themselves get more complex. And that's what I believe we really have to do to manage agentic AI, at least for now.
00:12:25 · Maia Bittner, Vijay Konduru, Laura Watkins
[00:12:25] S3: And it's really a better use of humans to be in those really high stakes, validating that... All of that type of work rather than doing the real grunt work of information gathering. [00:12:36] S2: Absolutely. [00:12:37] S0: Pav, would you like to kind of comment on that and and and that sort of potentially the automation piece helping organizations identify misconduct almost in advance or, like, know, before it gets to sort of more dire situations down the road?
00:12:54 · Pav Gill
[00:12:54] S1: Yeah. So, you know, it... It's not always gonna be a reactionary kind of approach to things going wrong, but a proactive one as well. And I think that's also where AI gets able to just highlight and and kinda like give warning signals. Right? Before we relaunched into this whole agentic space, I think it was really a matter of just sitting down and getting fundamentals, like asking yourself the basic question, what happens in companies? So everything needs to start somewhere. Right? There needs to be a trigger, there needs to be a request, whether it's a whistleblowing case, or it's a marketing approval request, something needs to start, or vendor screening has been hit with something. So that could be in a software, it could be a person initiating, it could be an email, it could be whatever. And then the company, the relevant person needs to manage that. And then at the end, it needs to be resolved. Now if we could find a system, which is what we're trying to do at Confide, that just brings all of these in one place, it just makes things so much easier. Because if anything goes wrong, you have one place where you can go to see how something started, who dealt with it, and what happened at the end. So I think that's just gonna be the modern way of working. I think the human in the loop bit is a bit complicated, because I was just having this conversation a couple of hours ago where I was saying that, yes, there's this fear that jobs might decrease, because a lot of the manual basic, I would say, like junior type work can be automated. But then on the flip side, the risk will increase for those that remain. Right? Because now they have to be the accountable ones. And will be... Will they be the ones that then, in a very tempting way down the line or predictable way, say it wasn't me, it was the AI. Right? And that also led me to question whether will insurance companies now have like an AI risk policy or premium, you know, the same way professional indemnity is. Right? So I think that's that's really like the bigger risks, Like people that are actually in jobs who are using these tools would have to think and be a whole lot more accountable than in the traditional sense.