Building the workforce for agentic AI: Skills, teams and what leaders need now

You've got the platform and executive buy-in, but do you have the people to deliver on agentic AI?

In this episode, Micah Smith and Kate Ressler talk with Mike Reynolds, Business and Technology Executive at KeyBank, about what it actually takes to build, retool, and upskill automation teams for the shift from task-based RPA to goal-driven agentic AI.

Mike shares how he invests in his team to build automations, how those citizen developer skills are evolving for the agentic era, why ideas that were rejected years ago are now greenlit, and the leadership mindset required to invest in people when the pressure is on short-term results. 

Whether you lead an automation team, build solutions, or are figuring out what skills to invest in next, this one's for you.

Transcript

Highlights

SPEAKER_00 0:00

So, what we said no to years ago, we're saying yes to now. When I think of releases on these big systems, maybe a couple of times a year. Now I log into the hyperscalers console. I see new features every day being called out. We trained about 45 people, probably 80% from the line of business. And what we quickly found out, they have a great aptitude, they can challenge the business process. So in learning, there's kind of a flip side to it. What if you don't invest in your people and they stay at the same level? That's a problem too, right?

Welcome and Guest Intro

SPEAKER_02 0:34

Welcome to a Gentic Edge, where we explore the frontier of AI agents, enterprise orchestration, and the architectures that are shaping tomorrow's intelligent enterprises. As always, I'm joined by my co-host, Kate Wrestler. In this episode, we are taking on a super hot topic that every, I would say, executive or leader is at least thinking about or wrestling with right now, which is you've got a platform, you may have executive buy-in, but do you have the people with the right skills, the right mindset, and the right training to actually be able to deliver on the promise of agentic AI? And today we have a special guest who is going to help us unpack this and share his vision for how he's taking this on. We are joined by Mike Reynolds, who is a business and technology executive at Key Bank, where he specializes in service digitization group. And he oversees automation platforms like Automation Anywhere. He's exploring low-code solutions with things like OutSystems, and he's managing the architecture and infrastructure teams to make it all happen. He has over 20 years of delivering technical solutions to business problems in the financial services space. And Mike brings a really rare combination of both hands-on technical depth and executive perspective on talent, team building, transformation, and what it takes to actually get an organization ready for what's next. Mike Reynolds, welcome to Agent to Gedge. Thank you for having me.

Defining AI Agents

SPEAKER_02 2:07

Mike, I want to start off with the same question we ask pretty much everyone who joins us on this show, which is how do you define an AI agent? Oh my goodness.

SPEAKER_00 2:15

Yeah. You do need a level set with everyone that you talk about. Like you've got agents, they're just going to answer questions from documents, right? Down to the agent that does work that interprets your question or ask to complete a task. I lean towards AI agents do work. Programmatically, alerting people, maybe even asking for permission. Hopefully that's in line with what some other people are thinking too.

SPEAKER_02 2:48

No, I like that. And I think that's always been something for me is it's got to actually take action. Answering questions is great. And I think there's definitely a spectrum on AI agents, but being able to take action, a certain level of autonomy, big win

Pace of Change in AI

SPEAKER_02 3:01

with me.

SPEAKER_01 3:01

Okay, Mike, it's super good to see you again. You have been leading automation programs at Key Bank for years. What has been the single biggest shift you've felt on the ground as AI has entered the picture?

SPEAKER_00 3:14

Yeah, no, this is a great question. The energy that AI has, the fast pace change that it's unprecedented, right? So when I think of releases on these big systems, maybe a couple times a year. Now I log into the hyperscalers console, I see new features every day being called out, new releases of model. So a huge shift for us is keeping up with the pace of change.

SPEAKER_01 3:42

Yeah, completely agree. The pace of change is unreal, which takes me to another question

Moving Fast with Risk

SPEAKER_01 3:48

here. You are in a heavily regulated industry, financial services. So when you hear move fast in the context of agentic AI, what does that actually look like inside a bank or inside an org that has a ton of regulation you have to deal with?

SPEAKER_00 4:04

Yeah. So moving fast, everyone wants to do this. And your risk partners, people inside the business, they have ideas. It's like, how do you take that idea and turn it into a solution that really does the work and adds value? The great news is that RPA has been doing this for a while, right? So we have that model to build upon. Now, from a fast perspective, I think making sure that you have the right risk, appropriate controls in place, those are the challenges that we see in getting people comfortable with those results. People are understanding AI at different levels. And now you have to bring those levels all together so that people understand your testing results as their drift. How do you manage those kinds of activities? And every day we're seeing more and more tools helping with the guardrails. Those are some of the challenges that we have to overcome. And I think we'll drill down into some more of those as we go on.

SPEAKER_02 5:06

Yeah, Mike, that's really interesting. I have heard from other executives from banks in Charlotte, where I live, that no one is adopting AI. So it's interesting to me to hear that you're leading the way in this space with Key Bank. Talk to me about how this has changed your opportunity pipeline, the aperture of what you're taking on. Because you're getting in opportunities from the business that say, hey, we want this kind of solution. Are they envisioning you using AI for that? Are you helping them to explore what's possible with AI? How are you drawing that line between this is a traditional automation type use case, this is an agentec use case, this is something that fits somewhere in between?

SPEAKER_00 5:46

Yeah.

From Copilot to Agents

SPEAKER_00 5:47

First off, we do have a great team in place today to help ideas grow. You may call it like a Gen AI Council, you may call it just a helper organization, and a lot of people piling on to this. And what does that really mean? Are people doing the work? Are they helping you do the work? A baby step in the right direction. We implemented copilot for work at web. That really was a baby step in the right direction because people could see AI in their jobs, starting to leverage, hey, I have an email that has this kind of message. Summarize these emails. Which emails did I not respond to? All those kind of basic activities, once you get those underneath your belt, then you can demand that AI does more for your organization. And so those baby steps build momentum and they put capability in the business user's hands. Not saying that we have the magic secret sauce to get everything over the line. We're ultra focused on internal uh, you know, employee usage, AI use cases at this point. We do have some uh touch points uh external, but uh very focused on internal.

Reskilling the Team

SPEAKER_02 6:59

Mike, I want to shift gears a little bit to talk about the people side of things because the whole topic here is thinking about how do we look at reskilling, upskilling for people who are actually building some of these solutions. And then I would also say who are consuming some of these solutions. So talk to me a little bit about what your team structure looks like and responsibility-wise, let's say 12 months ago versus today, and how you see those things changing.

SPEAKER_00 7:27

The face of the team has changed. And when I think of ideas coming to fruition, a year ago, we had, let's say, a handful of people, part-time gigs, right? It's like, hey, is this starting to get real? Can you help out here in exploring the talent to see if we had a right mix to really move these ideas forward? When I think of business partners being involved, there's certain business partners that are very forward-thinking, others just want to make things faster. You need a combination of both. The level of complexity also mandates that you have a discovery section, a proof of concept or proof of value to see if you can really do it and how valuable are the results that you're really seeing. So you do have to make these agents, you know, tangible. People got to be able to touch them and feel them and see how they're going to interact. It's like you want both, but which comes first. I think those are challenges that we see.

Citizen Developers Evolve

SPEAKER_02 8:27

We've talked before about your team at Key Bank. And one of the things that always really impressed me was the fact that you had a team of citizen developers. And whether you still call them that or not, I'm going to leave up to you. But you had a team of citizen developers who were essentially building all of your automations and your solutions. Talk to me about how that team has changed and where you've seen gaps as you've started to adopt and drive for more AI.

SPEAKER_00 8:55

So early on, we decided, hey, we're going to train anybody who wants to build RPA automations. We'll get you in the door. We trained about 45 people, probably 80% from the line of business. And what we quickly found out, they have a great aptitude, they can challenge the business process, they look left, they look right, upstream, downstream from systems. They understand the processes really well. And that's a jump start over just bringing someone that has the technical skills to build these things. So that mix of having people with the business process builds trust with other business partners. How does that transition to agency? What we're seeing is people who are really good at RPA, they also understand agents and what they might do and what they might not do. And then I'm going to throw something out the reasoning of agentic processes, most of the time, people like the idea that hey, it's going to come up with the best plan possible. But then we also have the risk-adverse processes where we want things done exactly the same way every time. And so agents talking to bots, and hey, Micah, you got the MCP servers that we're going to be able to leverage from our agents calling bots. Huge thing because we like that repeatability and being able to do the same thing the same way every time.

SPEAKER_02 10:31

Yeah, I love that, Mike, because I think it demonstrates an understanding of some things within a business process have to be dynamic, have to be more probabilistic, and they're great for goal-based agents. But there are certain things that exist within a business process where I still want that repeatable every single time the same thing happens. And I think there's even some space for something in between where I would say it's kind of an intelligent automation where I might have a bot or an automation that's using an LLM to make distinct decisions within that automation, but it's only for a single purpose. It's not for designing the entire workflow or something like that. And I think it's really a combination of all three of those things.

SPEAKER_00 11:14

And also, I'll throw out one other thing: RPA from an expense perspective. Hey, I'm a banker, right? When you think of that cost, it fits in a very narrow cost efficient versus some of the AI you start to dig in, consumption costs, all those other it's also a way you can manage cost.

SPEAKER_01 11:34

That totally makes sense to us. And I think um aligns with some of the things Mike and I have been discussing around kind of the spectrum of what is possible with all forms of automation from traditional RPA to a gentic AI. Okay, so we're seeing that AI is changing what gets built, how things are getting built, and who is building it. How are you seeing that role converge at Key Bank?

SPEAKER_00 12:00

Yeah, um, more and more people are getting involved all the time. I hear, hey, last night I installed XYZ favorite product of vibe coding, right? And I created a website instantly. Have you seen this? And I get a link to someone's server that they set up overnight. And these aren't just like everyday kind of enthusiasts, these are executives, our head of risk built a site that was showing it to other executives. So I think there's a whole nother level of involvement where RPA brought in a lot of citizen developers and maybe even turned them into real programmers. So you're seeing these skill sets evolve in a natural progression, as well as in traditional academia. Universities have AI paths. They all converge when you get into an enterprise like Key Bank. And how do you get those different levels focused in the right areas and pull together a team that you can really enable and move forward fast? You also have to combine the risk partners. More energy has been spent on what's the risk than any other type of programming I've seen in my life at Key

Shadow AI and Innovation

SPEAKER_00 13:12

and prior.

SPEAKER_02 13:12

I'd say that what you're describing is not unique to key with that risk of what's happening, right? I think something really interesting is happening, right? Creative problem solvers have never been more enabled at any point in history to be able to build things, right? Kate actually built an app that our team is using. And that's awesome because people who have interesting ideas are able to bring them to life in a way they haven't been able to before. Coding agents are making building things using these no code and low-code apps super easy. But I think the other side of that is there's gonna be this tidal wave of new apps and services that come from inside of the organization's walls. And I think it's gonna be like this shadow IT, shadow AI. And I'm curious how you're thinking about not only approaching that, but harnessing it, right? Because there's gonna be cool stuff that comes up and you're saying, hey, this would actually be something that would be good at a broader scale, not just for this team or that department.

SPEAKER_00 14:10

At times I feel like I'm speaking out of both sides of my mouth. It's like, hey, here's the IT policy, can't do this. And then, hey, how do you change the policies and procedures so that you can take hold and grab that innovation and then turn it into something tangible, right? It's like the rules uh of yesterday are blending into something that we have to change for tomorrow. And when I see the innovation, the ideas, it's at a rate that we've never had. Been tracking RPA ideas since 2019. We would get four or five maybe a month in those times. Today, I'm averaging 20 RPA ideas uh a month and we're building those out. I also have control automations that that come into play. I have rejected hundreds of automation ideas over the years. Now we're revisiting those and seeing that AI agents are now a play. So what we said no to years ago, we're saying yes to now. And it's because more people can get involved. They see these things, they remember what happened and the why behind it. And now we're enabling them to get their ideas over the goal line. In some cases, sandbox environments where they're completely walled off from our normal IT environments. We're allowing business partners into those spaces that traditionally they haven't wanted or had little desire to get into. Now they're like demanding to get into

Skills, Quickfire, Recap

SPEAKER_00 15:43

them.

SPEAKER_01 15:43

Mike, we have a lot of people listening on this podcast or on LinkedIn or everywhere who are worried about what is next. What key skill would you have them focus on to remain relevant?

SPEAKER_00 15:57

In my career that you mentioned this decade, I think you're trying to call me old, by the way. I have seen a few trends. Problem solving skills are always a plus. The language of databases, yes, I'm actually gonna say SQL has been consistent. It's like understanding your data, always important, and it's probably even more dramatic now because AI accelerates that. When you're talking about policies and procedures and summarizing them or even applying, having agents apply your policies and procedures, we see a lot of conflict in those. I went through an applied statistics program. Everything was done in R, and we're like told, hey, fantastic. This is gonna be the language you need to learn. Back in my early programming days, Pascal was the language of academia, and you learn Pascal so you can learn of it. Whatever you learn today, be willing to relearn in some form or fashion tomorrow, is my long story short.

SPEAKER_02 16:56

I like that. Yeah, I mean, with how quickly things have continued to evolve, the thread that you mentioned there is someone who can understand a business problem, align to it, and then understand how to solve that problem, those people will always have a place. As you think about your team at Key Bank, and as they've gone through this progression from being kind of automation builders to now starting to build more agentic and AI-focused skill sets and solutions, how have you thought about upskilling the team? Like, what are you taking them through? How are they pursuing this? Are you leaving it up to them to, hey, figure out what you want to learn and how you're gonna learn it? Are you prescribing some paths? Talk to me about that.

SPEAKER_00 17:35

It's a combination of all the above. I think there's always the person who's curious. Those are the people you want on your team. They can jump in, learn things, they're gonna they're gonna learn things on their own. They invest in themselves. Fantastic. Then there's a prescriptive path, observability or evaluation type metrics. Uh, when you're looking at, hey, is there drift or is this model when we do an upgrade, is it gonna give poor results or is it too long, not enough detail provided, being able to measure those kind of things, that's something we can train, right? There are platforms, we can be very prescriptive about that. But again, some of the problem solving techniques that curiosity has to take over, the new kind of systems, the guardrails you can train, but the problem solving, you got to grow and invest in people. And hey, you've seen it every day you log in. The console is different, there's a new feature. You gotta be willing to click on that and start to understand what that does. And then how are you going to leverage that for your solutions in the future? I love that.

SPEAKER_02 18:41

Mike, I would say you were early on RPA, right? You were one of the first, I think, major adopters of really driving automation for your organization. But I would say that as I look at the arc of your career and as I've seen kind of your progression, you've proven that you're not just a technology leader, but you're also very business minded. I think that's kind of rare for people. How do you instill that in the teams and the people that you lead? Because how are you able to help them connect the technology to the business outcomes that matter?

SPEAKER_00 19:11

It's always that curiosity. And what I mean at an executive level, your people have problems. You're trying to enable those problems to be solved. Sometimes you have to jump in and roll up your sleeves and work side by side without any hesitation. It's like imagine you had a team of people and you have no hierarchy, and you're just jumping in and solving problems when you need to. Uh, being a good teammate, those are things that people see and that great leaders can jump in and find the right kind of mix of a team. And then, hey, when necessary, jump in, fly up to that location, get the business partners into the meeting, or even better, have your people sit side by side and do the job and see the pain that people have. When you understand people's pain, it's very hard to hold back and not give them a solution.

SPEAKER_01 20:08

Mike, for those, the leaders who are listening into this, they're trying to make the justification right now to invest in upskilling and really support the learning and development, just kind of as you're saying here. What would be your pitch to those people to convince them that it is absolutely worth investing, sitting side by side with those who are on the learning path?

SPEAKER_00 20:27

Yeah, I think there's an immense pressure right now to deliver valuable results. So in learning, there's kind of a flip side to what if you don't invest in your people and they stay at the same level? Um, that's a problem too, right? So, you know, which technologies are you going to invest your people's time in? Which ones are you going to encourage them to maybe spend some of their time? And then how do you do it, right? It's like short term, you may be able to bring in contractors that know a very specific skill set and can help the team and then transition. But long term, you probably need that on your team and for your organization to thrive, be able to replicate that over and over. So when I invest in when I'm thinking of my people and investing in training, and when people come to me, I try to make those decisions. Is this like a personal thing? Is this a team thing? How much time should we invest? Every week we allocate an hour, people can learn whatever they want. It's like go learn, watch a YouTube, pick whatever you want. We're obviously going to have specific trainings. Some of them are business training too, right? It's like understanding and leveraging that business process just as important being able to come up with an AI or technical solution. So hopefully that gives a little bit of insight into how I think about it and use the pieces that you need to make your team great.

SPEAKER_02 21:52

Yeah, I like that a lot. I think we're seeing that people who are really technology minded need to focus on learning the business a little bit more so they can clearly architect. articulate why it is what they're doing is important for driving outcomes. And likewise, the people who are really business minded need to stay abreast of what's available with AI and automation and those capabilities so they even know what's possible, right? They don't know what to ask for if they don't know what's there. So I think those are both really important. Mike, I want to wrap this up with a quick fire. I want to get your gut reaction to a couple questions or statements that I'm going to read. And then I want to hear how you feel about them. So first off, true or false all agents are created equal. False. All right. The single biggest gap in enterprise automation right now is keeping up with risk factors. Risk. I like it. Vibcoding has its place in the enterprise or it's a hack for demos.

SPEAKER_00 22:46

Ouch. It's getting a place in the enterprise, but I tell you what, great for demos and non-production code. We got a few more things to figure out on vibe coding like supporting agents to move things through environments before we really get there.

SPEAKER_02 23:02

Love that. AI governance keeping us all safe or a stick in the mud for innovation?

SPEAKER_00 23:09

You're trying to get me to take a position on this it's both and we all know it. It's like the guardrails are there for a reason. And when it comes to privacy some of the data it's super important to not expose. So I appreciate the governance but it's like I also need to figure out ways to be able to innovate. It's an enabler and then it can also slow you down at times. I knew that was going to be a tricky one.

SPEAKER_02 23:34

What's one key skill that most agentic training doesn't cover well?

SPEAKER_00 23:39

I would say creating evaluators right how do you know if you're like a prompt-based kind of situation that your prompts are getting the responses that you want. So evaluation and observability like in the real world what are you seeing actually being asked and should it be asked in creating those kind of pieces to help the agents work better. Just don't see that covered a lot. Okay.

SPEAKER_02 24:04

What's the biggest challenge most organizations are faced with when it comes to agentic adoption?

SPEAKER_00 24:09

I'm going to say tsunami of ideas coming at you and whatever kind of idea that you pick on to work there's so many ways to enhance it that you can get paralyzed with just the volume of agents that it's going to take to really optimize a solution for straight through processing.

SPEAKER_02 24:30

I thought we were going to head back to governance for that one. I'm over that all right and finally one piece of advice for a technology leader who's trying to drive more AI and automation within their organization.

SPEAKER_00 24:43

This is a good one. There's so many things that you could leverage and I'm going to go with the advice I would give myself today it's like have a thick skin you're doing the right things no one is ever fast enough but you build these core kind of platforms and building blocks that you'll be able to leverage later.

SPEAKER_02 25:07

Love that great advice for Mike when he listens to this back later but also great for every technology leader who's listening to this podcast. Mike thank you so much for your time thank you for joining us I appreciate the transparency and authenticity in all of your responses. Looking forward to more innovation coming out of Key Bank and I think you're definitely leading the way in that space.

SPEAKER_01 25:27

Thank you for helping us all right so let's do a quick recap here. What did we talk about today? We covered the concept that building for the agentic era is not just about platforms. It's really truly about people we have to equip our teams with the right skills. We have to foster a culture of continuous learning and we also have to ensure our organization has the talent to turn agentic AI from a buzzword into something that's real and impactful for the business. Mike gave us a front row look at what it takes inside one of America's largest banks and the playbook applied well beyond financial services.

SPEAKER_02 26:06

If you want to go deeper on any of the topics we discussed today be sure to check out the show notes. We've linked some great resources on building your AI and automation skills as well as developing your team's agentic AI readiness. And stay tuned because we've got some exciting things in the works on how Automation Anywhere is investing in next generation automation professionals. You'll hear more about this soon at our Imagine event coming later this year. Thanks for tuning in and be great