00:00What is your day like? Tell us the things that are kind of coming front and center for you.
00:05Well, BMI is at the heart of financial services markets, 241-year-old companies.
00:10So I have my foot in many camps across technology.
00:14Of course, cyber and resiliency is really important to a bank like BMI.
00:19But obviously, AI is becoming increasingly part of everything that we do at BMI
00:23and really helping us shape financial services for the future.
00:26Sure. How much of it is internal in terms of the processes and the workflows that you guys are doing?
00:32You're a major custodian for so much.
00:35So is it also from what clients want in terms when it comes to AI?
00:38They're looking for functionality.
00:40So in terms of what we've built, we have built our own platform.
00:43And it's called Eliza, named after the wife of Alexander Hamilton,
00:46who built BMI and a philanthropist in her own right.
00:50And we've built that ourselves.
00:51And it is a platform.
00:53And we are a platform's company.
00:55And Eliza is a platform itself.
00:57So we have built that.
00:58It's model agnostic.
00:59We use all the LAM providers and multi-agent.
01:02And that is an internal platform.
01:04But it's also something that we use to interact with our clients
01:07and help our clients perform better.
01:09So give us a concrete example of what it can actually do.
01:11We have 170 concrete examples.
01:14We are way beyond the use case scenarios and very much into production.
01:18So 170 different versions of Eliza that touches everything that we do.
01:22So some great examples, KYC, know your customer, onboarding customers onto the platform.
01:27We have a lot of customers and $59 trillion of assets under custody.
01:32And so we have taken our KYC process that would have taken a number of weeks down to 20 minutes.
01:40So how does that work?
01:41You type the person's name into this model, and it tells you whether or not this person is who they
01:48actually say they are?
01:49Well, Eliza, in this case, for KYC, is a multi-agentic model.
01:54So you have, like, 20 different agents who talk to each other.
01:57So there's multi-steps in that process from knowing who the client is to, like, linking them to their different
02:03accounts across BMY.
02:04So it's not just one single agent.
02:06Let's say agentic AI, like, this is it in action.
02:10In action.
02:10And we have, like I said, 170 versions of that across the bank today.
02:14It sounds like you guys really kicked into high gear, like, over the last year in terms of AI.
02:18My understanding is 100% across your team, like, everybody's in it.
02:22Yes.
02:24How did you quickly ramp up?
02:26How difficult was it in terms of the process and the build and getting everybody on board and everybody functional
02:31in it?
02:32Well, three things I say was the silver bullet in terms of our adoption.
02:36First of all, we have a CEO who's very, very AI savvy, Robin Viz.
02:40Viz, yeah.
02:40He comes from a background in banking, but he, you know, is hands-on in the technology.
02:45He vibe codes.
02:46He uses it in his personal and professional life.
02:48So I think when you're CEO-led, I think that's a really foundational part of our platform.
02:53Then having our own platform.
02:54So our employees don't need to go…
02:56You kind of have to have your own platform.
02:58And they don't need to go into this different technology.
03:00Everything is an ELISA.
03:01It's our tech platform, but it's also our governance framework.
03:04It encompasses everything end-to-end.
03:07And then we really led with enablement.
03:09So instead of thinking about engineering, gatekeeping the technology, we said we are going to get all our employees trained
03:16in using ELISA.
03:17Almost everyone at the bank now uses ELISA.
03:20We did training last year.
03:21We had to reinvent the training program to get everyone to go deep.
03:25Do you need less workers because of it?
03:27And I'm not trying to be like…
03:29But we're trying to understand, like, what is…
03:32Is it replacing workers or is it freeing up workers to do other stuff which is helpful to an institution?
03:37So we actually have two different versions of ELISA when it comes to Agentix.
03:41So we had the solutions that we talked about, but we actually have digital employees.
03:45So employees who are not human, who have a human manager, who have their own login and work in the
03:50ecosystem.
03:51Now, that sounds like counter to what you just asked.
03:54But the fact is these help our employees become superhuman.
03:57And they're doing tasks that, quite frankly, are quite mundane and freeing up our employees to do much more interesting
04:03work.
04:03For example, vulnerability management.
04:06No software engineer likes to do vulnerability management.
04:08They want to code new skills and new tools into the environment.
04:12So our digital employees take that part of their work away and then that frees up our software engineers to
04:18work on the stuff that they, quite frankly, went to school to learn.
04:21At a large organization, and to Carol's point, like BNY Mellon, how do you do this in a way that
04:27provides an incentive for people who don't actually want to adopt these?
04:30They think to themselves, wait a second, I don't want, you know, maybe…
04:36I don't want this to take away my job.
04:39I don't want this to make me obsolete.
04:40But just to add on a conversation I had earlier to this whole idea that, like, people, almost more people
04:45become managers over kind of their agents, if you will, their AI agents.
04:50Is that part of what's going on or no?
04:52Well, I think the thing you have to do is, like, I totally understand why people feel like that.
04:56But when you train people to use the technology, I think it becomes people fear it less.
05:01So that's why we led on enablement, making sure that everyone had access to the best training.
05:05We did 100% by the middle of last year.
05:08We reinvented the whole training program.
05:10We gamified it.
05:11And when people can see how to use this technology as a superpower, they're much more inclined to use it
05:16to help them in their work than have the kind of reaction that you talked about.
05:19Was it really easy to train everybody, or how long did it take?
05:24So we had a goal of 65% and 25% was our goal.
05:28We made 99% by June.
05:30Wow.
05:30So I think the adoption, we talk about it a lot.
05:33Robin talks about it a lot.
05:34Our EC handshaked on having AI for everyone, for everything, everywhere, which was our mantra.
05:40And I would say it was just amazing how our employees really, really embraced Eliza as a technology that can
05:47help them.
05:47I love this terminology, though, of, like, digital employees, right?
05:50This is kind of our new world order or part of it.
Comments