Andreas Welsch: 0:31

Welcome to 2026, our new year of What's the BUZZ? in its fifth season. I'm so excited you're here. Everybody's been talking about agen AI last year, but I've seen very few people go beyond the what if, the visionary scenarios, the state of state of the art of the possible, whatever you want to call it. Today we'll talk about the top lessons from deploying Agentic AI in banking, and who better to talk about it than someone who's actually done that. Mo James. Mo, thank you so much for joining

Mo Jamous: 1:02

Thank thanks for having Andreas.

Andreas Welsch: 1:04

Mo, maybe can you share a little bit about yourself, who you are and what you do for our audience who might not be familiar with you yet?

Mo Jamous: 1:11

Yeah. Yeah, so I'm Mo Jamous. I've been in the financial industry for the last 15 years. Majority of my career. I spent the majority, also my career around data, technology and digital.

Andreas Welsch: 1:23

Really looking forward to the discussion today. So thank you for having me. Hey we met in Austin, at Generative AI Week in November, and said, we definitely need to have a conversation about this. What's working? How are you bringing it to the enterprise? So I'm so excited that you made yourself available. I know you're incredibly busy. Hope folks for you in the audience. You appreciate our conversation as well. By the way, if you're joining us live, please put in the chat where you're joining us from. I'm always curious to see how global our audience is, and if you want to learn about how you can turn technology hype into business outcomes, consider picking up a copy of my book, the AI Leadership Handbook where you'll learn all about that. Also, by the way, I have a new book coming out. Lemme show that. Very briefly too. It's called The Human Agentic, AI Edge, and it helps you shape your next generation AI ready teams. There's so much talk about AI and we need to use it. We need to roll it out, but not so much guidance on how do you do that, how do you do it well, and what do you need to do as a leader? Stay tuned. It's coming out at the end of February so you can learn more. Alright, now more in good old fashioned. Should we play a little game to kick things off?

Mo Jamous: 2:28

Why

Andreas Welsch: 2:28

not?

Mo Jamous: 2:29

Go for it.

Andreas Welsch: 2:29

Okay, so here we go. When I hit the buzzer, the wheels will starts spinning and when they'll stop, you see a sentence and I'd like for you to answer with the first thing that comes to mind and why, in your own words. Are you ready for, What's the BUZZ?

Mo Jamous: 2:42

Yep.

Andreas Welsch: 2:43

Okay, here we go. If AI were a. Sports car, what would it be? 60 seconds on the clock. Go.

Mo Jamous: 2:57

I would say Mercedes Formula One Car.

Andreas Welsch: 3:01

Oh, okay. Why? Why Mercedes And why Formula One Car?

Mo Jamous: 3:06

I've always loved Michael Schumacher and Saul. He was always one of my favorite drivers. I would always want them to win. And knowing how powerful AI is I can always use that help. So

Andreas Welsch: 3:17

I love it. It is very powerful. And sounds like you, you're in the poll position as well seeing this from very exciting angle at an exciting time. I'm curious we talked a little bit about this at Generative AI Week. You said you've deployed agent AI in your organization. But I'm curious, what has your journey been like? Where are you at right now in the organization and what are some of the key learnings?

Mo Jamous: 3:40

Yeah. Just to set the context before we kind of dive deep into banking and where are banks with agenda AI? I think just, it's always fair to say that the agenda, AI technology itself. It's in the nascent stages as a technology and I think it's fair to set that context. So everyone, a lot of companies feel that, oh my gosh, we are so behind in implementing agenda AI. I think everybody feels that way, which really tells me. It's not really everyone is behind, it's just the technology is still taking hold and maturing. And I think, we are seeing it something that we're it's an amazing and powerful technology that we're seeing maturing in front of our eyes just to see. Kinda where we were a year ago and when things came up with MCP and agent to agent communication and a lot of protocols to how make agents talk to tools and be able to do more. I think that's been very powerful in 2025. Saying that I think, for, in the banking industry and or any regulated industry, you're always gonna see things move a little bit slower than a non-regulated. Industry, which is always fair to say. I think our journey for us with agenda AI in the banking world has been around. Okay starting with the things that are not customer facing. Things like productivity, for example. Like how do we use agents and build agents that will allow us to make our employees more productive and be able to. To better help and allow, give them the capabil new capabilities and added intelligence to help our customers. So the way we been thinking about productivity in particular is what I say. Persona based. So if we look at if we take pdlc the product development lifecycle, within the product model we have four personas. A software engineer, a product manager, a designer an agileists. And then what we've really been looking at it, okay how do we make each of these personas more productive and build agents? That will really kind help enhance the productivity and accelerate the speed to market and getting the products in the hands of our customers, but also outside of PDLC. I wanna show we'll talk a little bit more about more concrete examples later, but that also outside of the PDLC, we'll look at. Bankers, branch bankers, how do we make'em more productive? We look at call center agents, how do we make'em more productive, et cetera, et cetera. So there are, so we're looking at productivity as persona based and identify the set of agents for each of the personas and how do we use agent AI to help'em make'em more productive. So I think, and this is really what I consider the productivity bucket, but then also we have two similar buckets that I would probably say. Outside of productivity, things like how do we grow the business and identify business opportunities you with agents and building agents to really boost whether sales or or improve customer experiences. And then the third one I would say operational excellence or operational efficiency. Things like fraud, cybersecurity things around DevOps. And then data center management and high availability and resilience fail over things that we could do with ages. So in my mind, these are the three buckets where kinda identified where we can build and deploy agents in.

Andreas Welsch: 7:07

So to me that's especially interesting because in a lot of the workshops I gave in my interactions with leaders across different industries. They always ask, show me the use cases. Show me the use cases for Copilot, for ChatGPT for some AI tool. And in, in my perception, it's indeed a lot about productivity and personal productivity. And I know before the the emergence of these tools, we were actually more talking about operational efficiency, operational excellence. Where can we use AI in HR? Where can we use AI in Finance? What are the the business drivers, what are the KPIs? Now? I see a lot of the conversation is about how can I write a better email? How can I summarize my meeting minutes, and how can I go beyond that? Where do you feel this conversation is from your vantage point? And where and how does personal productivity help Maybe unlock some conversations, open some doors now that it's in anybody's hands.

Mo Jamous: 8:03

Yeah, I think, I'll give a couple of examples. So if you look at the developer persona that I mentioned earlier, we're giving them, we're giving developers very powerful tools to generate code. Lots and lots of massive volumes of code that are AI generated and that is amazing and that's very powerful. But pretty quickly you'll notice that, you're generating all of that code, but then you're creating a bottleneck. When it comes to code reviews. So you have all of these commits that you've made and but there's not enough. People keep it, that are able to keep up with that level of frequency of ESA that are coming in. So one of the things that we've done, we've built an agent to automate code reviews, which we can assess. We've evaluated about, we're doing hundreds of thousands at us before. We're doing hundreds of thousands of code reviews or merger requests a year. Across the entire technology organization and then, and it normally takes, about an hour to do a merger request. If you look to any developer and then. Being able, we built agent agenda capabilities that took it, from almost an hour to less than 10 minutes to be able to review a MER request end to end. And that is super powerful, not just only saves. Money or actually produce efficiencies, but it also accelerate the whole development lifecycle. And it improves the quality of the product at the end of the day.'cause if you have so many major requests that are coming in, but you're not able to keep up with the reviews, then eventually you're gonna have quality issues. If you're putting things in production, would that being reviewed properly? So a little kinda agent like this has really took about. Two to three weeks to build. It's not very difficult to build these agents. But it's super powerful in terms of, the impact that it can have on the entire development lifecycle. Another, capabilities. We, we're looking at, for example, product managers. They do. A lot of, they have a lot of documents, a lot of research, a lot of ideas, a lot of big ticket items on the roadmap. Looking at building agent capabilities to, to be able to help automate the creation of those user stories. Out of the product documents prioritizing some of the those stories and refining them as we put them into the backlog. So that is really, makes a huge impact on the productivity of the teams and the acceleration of of the development lifecycle.

Andreas Welsch: 10:35

That's awesome. Great to hear. Sounds like some really tangible benefits and also with manageable effort. Is it two, three weeks to build the agent? Given that there are thousands or hundreds of thousands of commits to the code base, sounds like it's really saving a lot of time. What are you seeing in other parts of the business? How are they viewing this? You mentioned Agentic AI is still rather nascent. I would agree. What's the risk appetite? What's the opportunity that other business units see?

Mo Jamous: 11:02

Yeah, look, I think everybody understands the value and the power of Agen AI and really know that, in, in the next couple of years that our world will be completely different than how we build software and how we deliver value to the customer. But I think, everyone at least understand for the foreseeable future. This technology is not magic and it's not really going to do things, on its own. I think having a human in the loop is always going to be the case, at least for the fee foreseeable future. We see a lot of opportunities, with how bankers interact with customers. How do they present opportunities to them? How do they talk about opportunities with the customers seeing things that, branches. Banker, bankers and the branches do spend, half an hour to an hour or more on a manual task and building agents to help automate or semi-automate some of this capability with the banker pushing the BA and. The go button before, the task is executed. So there's tons of opportunities there as well. But also in the call center, which, this is probably every company is looking at opportunities, with knowledge based documents when customers call, being able to. The, the call center agent with making sure that they get and retrieve the right information as quickly as possible to help the customers. Which actually not just only improve speed, but also the quality as well. Because when you add more intelligence to to the process, you're always making sure that you're getting the right documents and that in a curated way. And and most of the times, with way better accuracies than you would get with most humans. Correct.'cause we are prone to make errors as well as humans. And I think adding that level of intelligence goes a long way. But that also things, for example, like trying to understand customers intent and deflect the calls based on. What is the customer is calling about? Is this something that we can now do an automated queue or things like that. So these are ideas that every bank is looking at. Every financial industry is is evaluating but really goes a long way. And I think, it's not just only about the execution and building of these capabilities, but also the change management. Of a process of how do, once you build these agents how do you get, you need your team to adopt those agent. How do you make sure that you go through the rigorous review process from a risk and governance perspective before you're able to deploy these capabilities into product?

Andreas Welsch: 13:43

So one of my previous guests said, everybody just shows you the demos, but nobody talks to you about how do you do the governance and what happens when you've actually built your first agent. What does the lifecycle look like? So it sounds a lot like that. Yeah. What were some of the hurdles that you ran into that you and your team maybe didn't expect when you started on this journey or when you said, okay, now let's look at Agentic AI and how they can augment. What we do or change entirely how we do things. What were some of the hurdles?

Mo Jamous: 14:11

Yeah, so I think there is a couple of things. Some of'em are technical in nature. I think the, if you look in the industry, you don't really have home ready agent AI experts or resident experts. And so a lot of our engineers were. Smart, extremely smart engineers and amazing engineers or AI experts. They're still learning a lot about these technologies. So figuring out what does it take to really build an agent? What does it really take to put an agent into production in terms of making sure you have the proper guardrails, making sure that you have the gateways, you making sure that you have. The evals and the monitoring capabilities. I think there's a little bit of a kind of struggle still that I see in the industry overall is pe companies not knowing. What does it really take to productionize and agent AI capability from what is all of the components that you need? And the ecosystem from observability to evaluation, to building to security, to making sure you have the guardrails. So I think there was a lot of kinda learning process and we had to go through that learning to be able to get there. And I think, the second aspect is definitely the change management aspect of it. Security and making sure what are the things that risk and compliance and security need to have before some of these CS cases are approved. But then also once approved how do you get people to use them? Also adoption is a big thing because building. An agent is one thing, but getting people to use it especially when you have thousands of employees that need to be familiar with it and get comfortable with using some of those capabilities, it takes a long time.

Andreas Welsch: 16:02

How do you facilitate that, that change management to get them to accept it or to warm up to the idea or to be even promoters of it?

Mo Jamous: 16:10

Yeah, I think, as part of, if you look at it, when we looked at our agent AI enablement strategically the way, thought about the strategy as a three three pronged strategy or one, I would say people process and platforms and starting with the people, just making sure that. We are upskilling our teams. We're making sure that we are getting our teams excited and have the executive sponsorship, around agenda AI and gen AI overall. And then also just galvanizing the whole organization around agenda capabilities to hackathons events and really kinda. Brown bag sessions just to be able to get people to learn what's out there, what's being built, and learn from each other as well. And the second aspect I say from a process perspective is making sure that we have the governance council, between technology risk, compliance, and security that is reviewing and approving all of the use cases that are going into production. And then last but not least is really building out the platform, which are the things that I mentioned from the the gateways to the observability, to the security guardrails and then the frontier large language models that we were going to use. But yeah, it just really, the change management aspect, back to your question is really just hackathons and upskilling and communications and events. As well and gamifying that process as well. I think, and getting the engineers and the teams excited about using some of those capabilities.

Andreas Welsch: 18:01

So I can tell you my mind is racing. I have so many questions that I want to ask. We only have about 10 minutes time or so. But when you so rightfully shared, it's one thing to build an agent. It's another thing or a different thing to get it adopted and to get it used well. How do you measure success in that situation? What are some of the metrics that you look at, some of the metrics that you maybe report out to your stakeholders and what can others? Where early on, on their journey take away from this.

Mo Jamous: 18:28

Yeah, so that is super critical. Ands and actually throughout my career I believed in philosophy like fact-based engineering or dragon driven development where part of that instrumentation and measurement is built. Into the capability we're building. So I think to me, instrumenting your agents as you're building them is super critical to be able to measure usage and consumption. Along with ROI, so one of the things we've done was actually as we are creating those agents, we actually added an extra components or metadata components that allow us to define what type of agent it is. So let's say. If it's an agent that's supposed to boost productivity, so this is a productivity agent, and then, and let's say this agent is supposed to save you an X number of dollars. Let's say it's a, it's gonna save you a hundred dollars every time it runs because if a human was to do it, it was gonna be a hundred dollars an hour or whatever, or$50 an hour, whatever the right rate you have. And then what we actually have done now that you have the instrumented the agent with that type of data, or what we had to do is to just start counting how many times the agent actually successfully executes. And all of a sudden, first of all, you're able to tell how many people are using it, and then second. How much value it's creating for the business based on the amount of time that it runs. And I think this is, it sounds, basic, but it's actually super powerful in terms of being able to tell the story of really what value the agentic AI is creating for the organization, especially when you start to do it for every agent you create. And then define the goal and the objective of the agent. And the value is supposed to create upfront. And then you can start to aggregate that in a dashboard across different business units across. Different areas, whether the three buckets we mentioned, productivity, sales, or efficiency. And then you're able to aggregate that at the enterprise level and demonstrate to leadership and and the lives of businesses, how much value you're getting out of Agen AI. Yeah.

Andreas Welsch: 20:40

How do you do that when it's not necessarily a one for one comparison, meaning a person sits there, does this task, it would cost us a hundred dollars. Now an agent does it and it costs a couple cents. But if it's because of a person usually completing this task or this transaction, it creates a much larger amount of value than the a hundred dollars that, that it would cost in time, basically. How do you think about business metrics with agents where, yes, it's a transactional comparison or it can be hours to seconds, dollars to fractions of percent. What does it look like for business metrics or KPIs or process performance indicators?

Mo Jamous: 21:20

Yeah. So from a business perspective, so if you are using for example an agent that helps you sell better,'cause now you are giving, for example, a banker or within the digital flow more insights about about a product that you're trying to sell. So you know what your baseline, what you're selling in terms of today and each product you sell is an X number of dollars that brings to the organization. As you are building those capabilities and moving the needle and the and how many products you're able to sell then and counting how many times this agent got executed and then each successful execution or conversion is associated with an X number of dollars, then it becomes also much easier. To really understand the value that this agent is creating. So it's not just only about saving dollars in terms of productivity, in the bucket of productivity, maybe we think about it and saving doesn't mean like it's really less people, it's just really producing more efficiencies where these dollars could be repurposed for value creation for the customer or really growing the business in other areas.

Andreas Welsch: 22:27

I love that. That's music to my ears. Yeah. I feel a lot of organizations are somewhere stuck in this transactional comparison, apples to apples. Today I have a team of people or a person doing this tomorrow I have agents or a team of agents doing that. How long does it take? How much does it cost? But it, it's so good to hear that there additional metrics, especially if it's about the operational efficiency and excellence, if it's about creating new value that you should look at. Now, Mo you also mentioned you've, obviously have agents in production. There are some things that the team is working on. Like any good organization, there's a healthy pipeline of things you're looking at. What is your recommendation for leaders who are earlier in, in their journey, who are now at the beginning of 26? Just starting to look into this and say we should be doing something with agents. How do they start with a pilot? How do they go from pilots into production? What is your recommendation?

Mo Jamous: 23:22

Yeah, look, I think there are, I think a lot of organization are getting stuck in the. I say tool selection. Oftentimes people think we should only have one agent framework or one tool or one capability, and I think that's a little bit unrealistic in my opinion. Imagine if if today, forget about agents for a second. If someone comes out and say, I'm only gonna have one sas. Product within my organization and I'm gonna only use that. And I think, people will laugh at them, correct? How realistic or really ridiculous that is. But the same thing. I think my advice is not to get stuck in and to, into selection and just understand that we, one day you will live. In a world where agents are getting created in different areas, some agents will, maybe you'll be using copilot studio to create them. Other agents you might be using a tool like L Chain, some, other agents might be getting created in Agent Force. So I think, we don't get too stuck in, in, in tool selection. But I think it's important. Maybe think about an architecture where you can create. A single glass pane to be able to understand and monitor all of the agents that are being created within your enterprise. So I think if you're able to build your own orchestration layer that's able to track all of the different agents within your organization so you can understand. The different capabilities and have an agent registry will go a long way. So just I think long I'm trying to say don't get stuck in, in, in tool selection. Be open to the idea that you will have multiple agents coming from different sources. And then second, from a people perspective, I think it's also education and upskilling your teams on really kinda what are agents, they can create it and what are what you can do with them. And start small. You don't really have to have, like I said, we started with this code reviews agent, which is really not super complicated to do, but had a huge impact. So find an area where you can create straightforward, one or two agents. You can have quick wins and start to create momentum building those one or two agents to begin with.

Andreas Welsch: 25:41

Mo, thank you so much for joining us today and for sharing your experience and your expertise with us. I certainly learned a lot about how to bring AI agents into business, what it takes where the opportunities are, where we are in the industry. Sounds like it's still nascent, but good to hear from leaders like yourself that you're looking at this, you'll bring this in, into production, into lots of good learnings, so more. All the best for an exciting and successful 2026, which is at the beginning. And I would love to reconnect and hear how things are going in the future.

Mo Jamous: 26:11

Alright. Thank you so much, Andreas. Really enjoyed it. And we'll stay in touch.