Beyond Boundaries S1 E6 - Beyond automation: Designing the future workforce

In the season one finale of Beyond Boundaries, Gareth Workman, Kainos’ Chief AI Officer, is joined by Mohan Rajagopalan, Senior Principal Product Manager at Workday, to explore how AI agents are reshaping not just the way we work - but the very design of work itself.

As digital workers move beyond back-office automation and into decision-making, collaboration and customer engagement, organisations face a fundamental shift. This isn’t just a technical challenge - it’s a leadership one. From evolving job design and redefining roles, to building the right mindsets and structures, Gareth and Mohan unpack what it really means to scale AI across the workforce in a way that empowers people and unlocks value.

What’s the opportunity when humans and AI collaborate by design, not by accident? What needs to change in how we lead, organise and build trust? And what practical steps can leaders take today to move from pilot to practice?

Tune in for grounded insights and forward-thinking strategies to help you lead the next phase of intelligent work - with clarity, confidence and purpose.

The full episode transcript is available here.

Available on your favourite platforms:

Transcript

Please note: the following transcript is auto-generated and may be subject to error. 

Start of episode

Gareth Workman 
Welcome to Beyond Boundaries, the podcast from Kainos that helps business leaders navigate the fast evolving world of AI. I'm Gareth Workman, Chief AI Officer at Kainos. And in this final episode of season one, we're looking ahead to the future of the workforce. So over the past five episodes, we've explored how the relationship of people and AI is changing from collaboration and agency to leadership, adoption and trust. So today, we're moving beyond automation to ask, what does it really mean to build a workforce where humans and agents work side by side?

AI is no longer just working in the background, it's influencing decisions, joining conversations and reshaping how work gets done. That opens up real possibilities and opportunities, but also challenges us to rethink roles, leadership behaviours, and how we scale AI responsibly across the organisation. To unpack all of this, I'm joined by Mohan Rajagopalan, Senior Principal Product Manager at Workday. Mohan brings a clear, strategic perspective from one of the world's most forward-thinking companies, one that's been embedding AI into work long before it became a headline. Mohan, welcome to Beyond Boundaries. It's great to have you with us and a perfect way to wrap up the season.

Mohan Rajagopalan 
Awesome, I didn't realise I was the season finale, but thanks Gareth and the Kainos team for the invite. We're excited to have a great conversation over the next hour or so, and hopefully I can learn something as well.

Gareth Workman 
Fabulous. And on let's just start with the big picture. So, we talk about digital workers or AI agents. What does beyond automation actually mean in practice for businesses? What's your take?

Mohan Rajagopalan 
Yeah. So it's funny. I was listening to our Chief Product Officer, Gerrit, speak the other day and he was asked a similar question and he said, we really don't know all these primitives out there about agents, about tools, about how they're being built. We're in the early innings, in the early stages. So we can all hypothesise and evangelise, but we're learning every day. Every day there's a new startup doing something new, a new large company figuring out ways. But the way we see it as is, where's the collaboration method?

How can we get these automated machines, agents, to work alongside our workers today to make their lives better? To focus on the work they're passionate about and get them out of the tedious work that drives everybody insane or makes people more tired at the end of the day than really the work they're excited to do when they come in and start every morning and what they graduated university to work on.

Gareth Workman
That's, that's really cool. I think, you know, as you say, this is early days and there's lots of things changing and there's lots of confusion around what agents are and aren't. So what are some of those common misconceptions that you hear kind of day to day?

Mohan Rajagopalan 
I think we're still coming up with what it is. I think right now, the way we think about agents is agents have skills and capabilities. And these skills and capabilities today seem rather simple. And tomorrow they could involve reasoning and needing more context about the decision making. But today is just about something that it can move us from point A to point B a little bit faster. And we always like to draw analogies to the consumer world. And if we think about how

automated cars are all over the place. I was in San Francisco just the other day and I got into a car and there was no driver. We sat in the back seat and we went about four miles. But if we think about that journey, it really started with us controlling the car. Then a little assistance, then some partial automation, then conditional automation, starting to really understand context of what's going on. And then from there, it would just keep evolving until we finally hit full automation.

So I think when we think about agents, agents today are here to assist us. And they may partially help us do some of our work. In the future, probably who knows when, there may be full scale of being able to actually do the work we all do. But for now, I think we're in the early stages of assisting us, maybe helping us here and there on some of the partial work, but that's kind of where we are.

Gareth Workman
Yeah, we're very much in the foothills of the possibility and how do we remove some of the, maybe the tedium from our lives or things that want-to-do for want of a better analogy. But in that near term, where do you see AI agents or digital workers delivering the most tangible value? Is there business functions you think or sectors that are maybe further along in this space?

Mohan Rajagopalan 
So I think the earliest sectors where we're seeing it is really where there's a lot of reading, summarisation, and where the context already exists, but it just may be too dense to get through. So we did a great acquisition here at Workday for Evisort, which really does work around legal contracts. These are, for the most part, structured documents. And if you look across Workday or Kainos, we have many thousands of customers. Somebody may change one or two clauses. And for an AI agent to read through all those contracts, identify...

how many of them are all identical and follow the same terms to how many of them have special terms. I think that really is where it would take a human too long to read that and too tediousness, but we're able to use the power of compute against that problem. It really unlocks a lot of opportunity.

Gareth Workman
As you say, it's where this kind of maybe process intense or labour intensive that actually, it is coming into its own and helping people get through that breadth and depth of content.

Mohan Rajagopalan 
Yeah.

I think it's labour intensive. And you said it, right? But also like the data set is pretty well structured. We're able to give a baseline around what we submitted in terms of like maybe a contract negotiation. And then the system is quickly able to tell us where all the anomalies are within the vast breadth of negotiations we would have done. Workday has been around 20 years. So you can imagine the different ways we've talked to contracts, but it also just transcends depending on our customer examples.

Gareth Workman 
That's really cool. So like one of the things I'm gonna pull a thread on here is like, as AI agents do become more capable, how do you ensure that they actually improve employee experience without adding noise or friction or cognitive overload?

Mohan Rajagopalan 
Yeah. I think it's interesting. We've always seen this in different forms, whether it was like RPA automation or workflow automation. I started my career as an Accenture consultant. And one of the first things we were taught to do is what's the as-is flow, what's the to-be flow. And I think right now where we are is we have the way we're doing work. We have the way we think agents and AI can help us do the work. And in six to eight months, once we've worked with that agent, we see how we're actually accomplishing work then. And I think being able to

A, look at that progression. You'll be able to identify some early stages of value. But then on that end state, are we really going to be able to tell that our work has been done better and that's led us to free time? So I think it always comes back to the basics. How are we doing it today? Let's map it out in a process flow. Then let's see where we can tweak it with agents. And then let's see how we're able to hold and maintain that new process in the longer term. But I'm sure this is something Kainos runs into with its customers. When you

go into a new implementation, how do you help them identify, do they wanna build their current workflow process or do they wanna try to transform their workflow process for the implementation?

Gareth Workman 
Yeah. And as you said, there's always a balance to be struck in terms of making better what you have or going out and saying let's just re-imagining everything. Cause with re-imagining things brings unintended consequences as well, along with kind of benefits.

Mohan Rajagopalan 
For sure.

Yeah. And I think the other one which we depending on where you're on the organisation, sometimes those outlier processes or complicated processes are there for a reason. It's something which we see a lot. Somebody has tribal knowledge that knows that it can be a nuisance later. So you don't want to reimagine too much.

Gareth Workman
And sometimes things, they've been in existence for so long, the original reason for why they exist has often been forgotten about. And it's that piece of how do you unpick some of those challenges.

Mohan Rajagopalan 
So I think some of this process stuff, this is process re-engineering, we're going to be doing it continuously for the next 20, 30 years. And agents are just going to be a new tool in that tool set.

Gareth Workman
So Workday has been known for embedding AI in its core platform. So, from your perspective, how has Workday's approach to AI agents evolved? What have you learned about designing them in a way that truly enhances the human experience at work?

Mohan Rajagopalan 
I think first and foremost, we've always been deliberate and humble in how we rolled out AI. We've always wanted to assist our workers. We really believe in human and machine collaboration. And that's from our early days of anomaly detection, rolling out skills cloud, to now all the agentic announcements you'll see at Workday Rising in November in Barcelona and September in San Francisco. But how our thought has evolved is really just understanding

what people are doing, what's the context they need to make those decisions, and then from there, being able to see where can we augment technology to help those decision-making. If you imagine it, Workday celebrated its 20th anniversary today. So we have some customers with 10, 15 years of data about their workers within their Workday experiences. And now we're going to help them navigate that with agents so that, Gareth, we know your entire progression at Kainos. We know where you started, where you came. We know your passions. We know your performance reviews. We know... all that - but now, what's the insight we need there to make you excited to work at Kainos for the next 10 to 15 years and how can AI help that to bring it up to your manager, to your leadership group and really be able to identify the right opportunities for you.

Gareth Workman
Yeah, and I think the bit that comes through loud and clear is just how human centric your approach to it has been. It's always been about helping people achieve more and do more than just a technology play. It's the human at the centre of the wheel.

Mohan Rajagopalan
Yeah.

We truly believe that like people are centre of companies. At Workday, our number one core value is our employees. We want to make sure we have happy employees all the time. And having happy employees is about preparing them with the right tool sets that they need to get their jobs done. They want some parts of their consumer experience here within their business. They want some of the business tools to also be up level to a very consumerist experience. And the same thing is going to be true with agents. We interact with agents all the time in our personal

Gareth Workman
So maybe delving into sort of practical ways Workday help customers adopt AI agents, not just technically, but in terms of kind of culture and mindset as well.

Mohan Rajagopalan 
Yeah, I think first and foremost, we're really proponents of the idea that we have to take our crawl, walk, run to this. Trying to do a large scale transformation across thousands to, in some of our customer base, hundreds of thousands of employees is just very difficult. So start off with a business unit, start out with a team, really go through as we talked about earlier, we're all getting marketed to with different agents. We have employees who want to build agents. We have developers within IT teams who want to build agents.

So there are some things we know agents are really good at. We talked about earlier, retrieving information, summarising information, being able to organise work. So try to find those areas where there's enough scale to be able to build and test into it, and then identify the right person. Is it a build where we're working with our IT team? Is it a partner where you're working with an external consultant who can provide you bespoke or expert advice? Or is it somebody who has IP in that area and actually has built a product?

And then bring that on board and just start it small. Pilots work great, then roll it out. I think we're also in a world where we know regional regulations may change or may enforce different things. So the idea is start small, get the early movements, show the ROI, then expand it out from there.

Gareth Workman 
As you say, I think it's that piece of just, walk before you can run and kind of show that bit of confidence. And you can use that flywheel effect to do greater and bigger things, but get that momentum to go in a safe, manageable pace that you can, you can deal with  in your organisational context.

Mohan Rajagopalan 
And I think the other thing which we can't forget is what we've learned all the times while being in business. There's no way to build something and forget about it. So make sure you have a plan to maintain it, to update it. You're understanding that because how we see our businesses today are not how our businesses will run tomorrow. So we need to have all the tooling, all the agents we have really involved with that story.

Gareth Workman 
Absolutely. And I think you've touched on something like, businesses are going to look different in future. So in terms of, you know, whenever we look at integrating digital workers into teams, what do think the opportunity is to rethink job design, collaboration models? How do you see roles evolving alongside AI?

Mohan Rajagopalan 
Yeah. I don't have a crystal ball on this one. For the customers we talk to, for the developers we talk to, for the hyperscaling platforms we talk to. We don't know. Every day, every AI announcement I sit on the edge of my seat waiting. I think right now we know what we have. We have the ability to create great content through gen AI, we have the ability to really do a new user experience through the conversational experience. I think that to me has been very surprising.

that I have to be better at asking and being able to type it out and ask. And if you think about it, the last large UX transformational shift we had was about 20 years when the iPhone came out and it was all touchscreen. So now this chat conversation is our new experience. So going back to your question, we're learning it. I think we just need to start with what is AI good at? Where is it a deterministic answer? Where are we, where if the answer is wrong,

it's the lowest risk of having an impact? And then start testing it. Once we see the significant scale of that, we can move it out of the job description, move it into the AI, but also more so taking any savings we have and reinvesting back into those employees. AI is not a story of cost savings. AI is a story of being able to reinvest in growth areas of your business and growth areas of your employees. And that's something we want to always remember.

As much as you're moving out, you're taking those dollars and you're reinvesting it back in your business.

Gareth Workman
Yeah, no, a hundred percent. Whenever you chat to most people, like their to-do list is always too long. Nobody ever gets through their to-do list. And the high value things they want to get obviously are out of reach because they're kind of trying to keep up with the pace. So as you say, the world is normalising in a different way and it's going to create different ways of working and different roles and entirely different opportunities as a result of that. But it's a, it's a moving kind of target as it were at the minute.

Mohan Rajagopalan
The time savings I'm really excited to see is that person who may spend an extra two hours after their family goes to sleep to clean out their inbox or their messages. If that's now 30 minutes or an hour, that's really the type of savings we're excited about, but that's also built on a larger story.

Gareth Workman 
Yeah, so I suppose this is another kind crystal ball question, I'm already starting to think about this.

Mohan Rajagopalan 
You're pushing me to the max here, Gareth. You're going to come back, we're going to do this in a year, you're going to tell me how much I got this right and wrong.

Gareth Workman 
I think there'll be some things that neither one of us predict will happen as well. But in terms of, kind of, what should leaders  be thinking about when it comes to that full life cycle of a digital workforce from kind of onboarding them, monitoring them, or even retiring an agent, whenever it may be, something better comes along. What's your early thoughts on that?

Mohan Rajagopalan 
So, we like to think of it in four great stages and our functional architect likes to remind me of this. First off is just registering the agent. So understanding that the agent exists within our footprint and it navigates through our enterprise stack and enterprise technology. Second is about configuring the agent. Really do we understand what skills within the agent we want to turn on and off? Most of these agents, you can almost think of them as like super orchestrators. They will have sub skills that they're doing and you can... either opt in or opt out of them. And once you're able to opt in and out of skills, understanding what are their security credentialing. Where in your enterprise architecture are they able to navigate to and what are they able to interact with within that data set? If it's a Workday specification, what are the security groups or contextual security they're inheriting? So now we've been able to configure an agent. We were able to register an agent. The last step is activating agent. How are we able to activate it and then start looking at those agent-driven metrics around what are they doing? What's the scale at which they're working? What's their success rate? Being able to look at traces and observability around what do they actually do? And then lastly, when it comes time to suspend an agent because you don't need that use case anymore, or deprecate or retire an agent, those use cases will come. I think we still need to understand around what it means to retain datasets around that. So are we purging or just retiring and maintaining their existence? But we really see it going around the ability to register, configure, activate, then track in analytics. And then lastly, trying to understand when and if you need to suspend or retire the agent.

Gareth Workman 
But I think in that kind of process, it forces you into some deliberate thought process to think the journey through from the outset and you don't just go, let's activate and see what happens. 

Mohan Rajagopalan 
Let's activate and let it run wild, because the scariest thing is always rolling it back. We all know database people who the script they least like to write is to be able to roll back something which they previously did.

Gareth Workman
And another thing that's terrifying is sometimes fixing forward and not sure when the fix might arrive.

Mohan Rajagopalan 
Yeah, one hundred percent. And imagine this too, at the scale of which we all work with businesses today. It's large. You're trying to push fixes out to multiple clusters across different regions with different data retention rules. That's why the being deliberate, being able to approach it in stages and being able to track it with the ancient metrics and analytics is really powerful.

Gareth Workman 
Yeah, absolutely. And like for these systems to scale, they need, kind of, you know, buy in and trust. So maybe what, what role does, you know, HR play in helping you integrate digital workers, you know, for that transparency and team and organisational morale?

Mohan Rajagopalan
This is all a partnership. There will be groups and councils really impressed, the office, the office of the CIO, the CIO role has evolved since my existence, like working in businesses. It used to be that they would give you the tool set and you're like, this is the tool you'll use. But now they really are stewards of the business. They really manage the data set. They take that responsibility seriously, but they also look themselves to be partners, partners with the office of the CFO, partners with the office of the CHRO.

So when we think about your role, what's the office of the CHRO in this? It's two parts. One of them is being able to identify their organisational design. What are these agents doing? Where are we doing the reinvestment? What are our employees doing? And then as people who are able to uptake agents themselves, understanding how they want to use it. What's the shape? What's the process? Where do we want them to do it? Where they want us to notify. So everybody plays two roles in this. Like even the CFO, they'll look at the ROI or on the investment of agents. But also look to see where agents can help and assist and collaborate with their own team members as well.

Gareth Workman 
Cool. So you've already hinted at this earlier in the conversation, but just to kind of bring back a topic that we touched on. So where do you think businesses should draw the line between decisions made by AI agents and those made by humans? And how do you make that boundary clear in day-to-day work?

Mohan Rajagopalan
Yeah, I think right now it's probably more conservative and it's going to be a spectrum. So the spectrum is no decision making to full autonomous decision making. We're probably in stage one or two where we determine what the decision is. We determine how many times is that decision made and what scale and then what's the goal of that decision making that we want the machine to do there. It could be something simple as if we go back to our legal example, I just need some things, review a large stack of documents and give me anything that where there's an anomaly. There may be a high miss rate. They may identify some stuff with anomalies that didn't be identified, but you would rather have that decision be more conservative than more liberal. And as you learn and you're able to provide feedback loops, let's not forget that a lot of these agents, a lot of these interactions, there are going to be human feedback loops to say, was this correct? Was this not? So if you're engaging in those feedback loops and you're a little bit more on the conservative side, you're training those agents to really evolve with your business. And then maybe in two or three years, it's a little bit more autonomous and it's a little bit more autonomous. I still go back to view it as a scale. We're on zero to one right now, maybe one to two, and we're trying to get to five, but it's not a race to five, it's a deliberate process to five.

Gareth Workman
Yeah. And some of the things you talk about is, it's very natural to us when we talk about it kind of helping humans improve, whether it's through getting their feedback or applying that type of thought process to drive better outcomes, as you say, but it's, it's important that that is, you know, factored in and done so.

Mohan Rajagopalan 
And the other one too is being patient. We will have to be patient with the machines and it's harder to be patient when, it's hard to empathise, because you're looking at a blinking cursor or looking like a chat product versus a person, but we will have to be patient because we really want it to help us at the end of the day because we all can lose perspective of the goal which is to free up time to do the work we're passionate about.

Gareth Workman 
Absolutely. So one of the things, if you've been following Beyond Boundaries this season, we always like to leave our audience, with something practical, something leaders can take away and apply. On that, Mohan, if you're looking, kind of, to move beyond pilots and starting to scale digital workers across the organisation, what's the one practical step that you think makes a real difference?

Mohan Rajagopalan 
That's a good question. I think as we think about agents and we're really trying to scale, it's really just trying to put in the right guardrails around where and how it's playing. Like know the systems it's interacting, have that tribal knowledge about why the system exists, why the process exists, and then be able to deploy the agents in ways that you know, like kind of where, which direction they're going to go. The scary part is, and I think the fear we all have is agents are just going to run wild, but that's only if we really let them.

But if we're able to build the right set-I have a two year old. When I take him to the sandbox, if I don’t go like, ‘walls are there,’ okay, I kind of know which parameters I need to be watching in. Then I’m just looking in the middle for something unexpected. And I think that's really the tidbit. There's no race to get to full autonomy, having agents everywhere.

It's all about human machine collaboration. So with that, set the guardrails up right, really provide the machine's an opportunity to collaborate with the human's where works actually getting done. See the results and always just be ready to be agile and move with it.

Gareth Workman 
Yeah, thanking you. Brilliant. Thanks for that. So this conversation has brought out some kind of big themes - like the changing nature of work, the evolving role of AI and what it means to be leading a more AI native world. But before we close, I want to leave our listeners with something that kind of sticks, something that they can take back to their teams. So here's my final question. If you had one thought from today's conversation, you'd want every business leader to hold on to, something that would maybe challenge assumptions or sparks action. What would that one thing be?

Mohan Rajagopalan 
Yeah.

So I have three thoughts and I've highlighted them throughout this podcast, but I think it's really important. I think first and foremost, these primitives are new. The connections are new. What's being developed is new. So there is a patience. We need to be patient as we roll these agents out. We need to be patient as we train them. The second thought is really around the idea that whatever we see as savings, whatever we see as the value being driven, it's really about taking that and reinvesting it back into growth. And growth comes from two places. Like, can you identify ways to upskill your team members? Can you identify new product lines your business should go interact? Let's really take the promise of what the agentic future is and put it back into the business to scale for growth. And lastly, it's really, and we're passionate about this, it is about human and machine collaboration. There's no business where machines are gonna run the day.

Humans really have the knowledge, the context, the empathy, the skill set to drive the growth and the agents are here just to support that in a collaborative manner.

Gareth Workman 
Yeah, I really like your point about, about patience. It's a really well-articulated thing. It's almost this race to go somewhere, but actually figure out where it is you want to go before you start racing off. It's the fear of missing out almost. And so it's almost drives a...

Mohan Rajagopalan
100%.

We're in a hype cycle right now. I actually think the trough of disillusionment, once we come down, is going to have some of greatest innovation and it'll be the transformational innovation we all see. But for right now, it's early innings. These primitives are new. We really want to reinvest everything we see into growth and it's about human and machine collaboration.

Gareth Workman
Perfect, so look, that's a fantastic and a brilliant takeaway to end on. So that brings us to the end of this episode and it's the final conversation of season one of Beyond Boundaries. Mohan, thank you for sharing your insights so freely with us today.

Mohan Rajagopalan 
Yeah, awesome. Gareth, this was a pleasure and hopefully we'll be back in season two to check out some of this crystal ball work we did.

Gareth Workman
Absolutely. And for our listeners, the message is clear. AI isn't just reshaping how we work, it's reshaping how we think about work itself. So whether you're experimenting or scaling, now is the time to lead with clarity, confidence and purpose. So thank you for listening to us today.

Mohan Rajagopalan 
Thank you.

End of episode