Subscribe
Share
Share
Embed
The WorkOps Podcast is your weekly conversation with HR leaders and People Ops practitioners doing the real work.
In every episode we dig into one story. A process that went sideways, a system that just didn't work, and what someone actually did about it. Packed with practical lessons you'll want to bring back to your team. Whether you're supporting 500 employees or 5,000, this is how the best People leaders are building for what comes next.
nancy_luschkowski-e78d76fe-ea44-47f4-8bca-1-CFR
===
[00:00:00] Welcome to the Work Ops podcast. In every episode, we dig into one story, a process that went sideways, a system that just didn't work, and what someone actually did about it. It's packed with practical lessons that you'll want to bring straight back to your team. This podcast is brought to you by Kinfolk, the AI service desk built for HR.
I'm your host, Jeet Mukherjee, and with that, let's dive in.
Jeet Mukerji: Hey, everyone. Today, I'm joined by Nancy Luschkowski, the director of HR Infrastructure and Operations at PagerDuty. Nancy, thank you so much for joining us today. Before we jump into things, can you tell us a little bit about yourself and how did you choose HR?
Nancy Luschkowski: Yeah, I would love to. So I have spent about 13 years now in the tech and SaaS sectors working behind the scenes as an operator. So really building and scaling HR infrastructure through various stages of both pre- and post-IPO growth. So my sweet spot really [00:01:00] lies at that intersection of strategy and execution.
So whether that's navi- integrations or implementing various HR tech stacks, designing total rewards programs, or even just rethinking how we scale core processes, right? Which is what we're, we're doing right now at PagerDuty. Lately a lot of my focus has been on how we can thoughtfully bring AI and automation into the employee life cycle and to drive true value, right?
Not just chasing efficiency. So for me personally, as a PMP practitioner, I really like to approach HR as a product. I really strongly believe that it needs to be efficient and data-driven and above all else, really employee-centric, right? So I'm really excited to be here today.
Jeet Mukerji: Love it. Those were music to my ears, HR as a product, Nancy. And really great to have you , on the podcast. I'd be curious to hear, before we start talking about your dysfunctional story, which we're gonna get into, how do you think about [00:02:00] HR as a product? Or how do you define HR as a product?
Nancy Luschkowski: Yeah. So really just treating our programs and our workflows like living products, right? Rather than treating a process or a program rollout like a one-time project really mapping baseline data gathering continuous feedback, user feedback and identifying where operational time like leaks, right?
And then iterating our, products, our programs, our processes over the entire life cycle
Jeet Mukerji: That's really good to hear. So effectively, you're a team of internal product managers that's building out the employee experience. That's great to hear. I was also curious to hear about your title. It is not only HR operations, but you have infrastructure in there. Can you tell me of how you see those two things as potentially separate but connected?
Nancy Luschkowski: I, yeah, I do feel they are separate, but very interdependent, right? And I think what PagerDuty, we recently just did this, so we just formed this [00:03:00] new function. It- people operations, talent operations and various operations were siloed within the different HR specialty areas. And what we noticed is while each team was doing amazing work building great programs and really processes within their function, we were finding that , there were some gaps when you looked at the process as a whole.
So there was redundancies, there were tech stacks , technology decisions I'll call it that were being made worked great for that team, but there were available features in another part of another piece of technology we had. So just really making sure that we are when we're designing a process even if it's, just in performance management or even if it's in the compensation cycle, we-- when we're process mapping and we're designing, we are really thoughtful about the end-to-end design so that we can account for any interdependencies and really make sure that it's a great experience
Jeet Mukerji: Love that. So Nancy, let's dive into your [00:04:00] dysfunctional system or process. What do you have for us today?
Nancy Luschkowski: Right now the process of adopting AI, and I'm calling it a process because I'm operational. Everything is a process for me. But that adoption for AI in HR is very, is a very challenging task. There's a lot of pressure from the industry and leadership to adopt and execute AI and just be more efficient.
So for me, there's, as I was thinking about this podcast and getting ready, I was really thinking about what really-- I was process mapping my own challenges currently. And I think there's four, key areas. The first one really being this internal tug of war human touch versus over-automation.
And I think over-automated processes where you lose the human touch and valuable engagement , it is a risk, right? So I'm really trying to figure out how to automate it, automate more of the ro- those robotic parts of HR without losing the [00:05:00] human touch. And I think the friction that this causes is or o- the friction that over-automation causes anyway is when a process is 100% automated, employees may feel like, more of a number in a database rather than an actual team member.
And with that, AI and automation right now really follows a binary logic, right? So they don't really understand nuance, empathy, or exceptional circumstances that are real things in a in the employee journey. So I don't think we should use AI to really replace that human conversation.
We should be really using AI to buy back the time for more human conversation when it matters.
Jeet Mukerji: Yeah.
And when you're thinking about those robotic parts of a process, it sounds like what you're saying is "Hey, we're not gonna automate the entirety of the process, but maybe like 40, 50, 60% of the process could be automatable." What are some of those kind of low-hanging [00:06:00] fruits or processes that you're thinking about?
Or are you early in the journey of doing that discovery?
Nancy Luschkowski: I think anything administrative that - the team currently does repeatedly that is more of triggers within a process or data entry those type of things. I think an example could be, when somebody is off-boarding, right? So s- an employee has been with the company for five years.
They're leaving the company and then rather than a personalized exit, right? If we automate every single workflow and send a bunch of robotic emails that can feel cold and not give that person the appreciation that they deserve after five years of service. So maybe it's gonna look something like you automate the exit, the process and workflows and reminders but you customize the exit interview, right?
Where you take the time to appreciate them. You are [00:07:00] customizing notifications to various teams and Slack channels , that does the individual the justice. And having a very thoughtful team farewell, right? Where it's an in-person appreciation. So things like that. You can obviously automate the payroll, final payroll execution, their last paycheck, or you can automate the, instructions on benefits, right?
Your benefit termination, your COBRA, things of that nature. But just really giving the team the space to then curate those really personal moments.
Jeet Mukerji: Yeah, and spoke about a process which a lot of HR technologies claim to already do, and yet there are some gaps. And I think what we're seeing now is that there's automation and then there's AI, and I'm curious to hear how you think about what is something that falls in the automation bucket versus something that falls in the AI bucket.
Or actually when you're mapping those processes are you not [00:08:00] separating and are you looking at just what is the best way to solve it?
Nancy Luschkowski: we are. What we actually recently did was we just took a request. We went out to the team and said "Hey, if you wanted to launch AI, what is it?" So we can start doing an inventory of all of the things that we want to put on our roadmap. And, I think about 90% of the requests that came in were actually for automation, right?
So it-- I think there's a piece of understanding of how, what... i'm sure you can use AI to do a little bit of automation for certain points but it was for automation. So I think for us, it's really defining that for the teams. And how I'm thinking about it right now, and this might not be the right way, right?
I'm just going through my own process, is we want AI for us at PagerDuty to be able to connect multiple sources of data points, right? Where it's just not a single trigger, a single request. It's, recalling from sep- a lot of different areas. I think phase two of that is gonna be [00:09:00] then how do we close the loop, right?
How will AI be able to not only recall, but actually complete a workflow, meaning an address change or, some basic things. But in order to do that, which kind of leaves me to my next talking point is really around You also cannot drop AI into a process that is broken and expect it to do that better.
Solve, right? We have to go back to what are we solving for, right? Do we need to fix our processes and make them more simple? And yet the answer for us at PagerDuty is yes. We have to start with that. One really cool AI use case that we are talking about is really getting more data and metrics for our hiring teams, meaning can we pull data in from multiple work sou- data sources to say, "Okay, where is the talent that we're looking for?
How do we determine if they are top talent? Like where is their geo?" What compensation are they looking [00:10:00] for?" Are there more insights as can we get insights from LinkedIn or other sources to say, "Why are people leaving the roles that they're in, and what are they looking for?"
That's a more intangible. That's that, those are the things that, that I'd like to see the company benefit from AI.
Jeet Mukerji: Nice. And the day-to-day job obviously doesn't stop, but what we're also talking about is getting things ready for AI and also understanding what is AI in and of itself so that we can then differentiate where do we point it at. That feels like a lot on top of the day-to-day job of just running HR operations.
How are you motivating the team? How are you carving out time for the team? What does that look like today?
Nancy Luschkowski: Yeah. So we, right now we are motivating the team by allowing people to experiment, and what I'm working on right now is giving everybody access to the tools in order to do that, and building definitions on what is the safest way, right? If you're working with employee data you're [00:11:00] in HR, what is the safest way for you to experiment?
Really exciting news that I just found out, so in July all employees at PagerDuty are gonna be given an enterprise AI tool. So that's coming soon. A smaller group of us have just recently got access to that, so we're gonna be doing our own testing for that and then giving really clear guideline...
with the good part of having an enterprise AI system is that you will have the same access and permissions to any source that you're trying to tap into that you would otherwise, right? So it's a little bit more con- safer, I wanna say. I don't wanna say controlled, but safer. So that's really exciting that we get to do that.
Other than that - We have Gemini for the enterprise currently, and for my team, I've been really encouraging them, even when they come, ask Gemini first. That should be your first place to go and ask before, you come to me and ask. If you wanna ideate something, your AI agent, your AI tool is a great partner to do that.
So that's how we're [00:12:00] encouraging. We're also talking about making the space and having, Oh my gosh, what is it called?
Jeet Mukerji: Like a hackathon?
Nancy Luschkowski: ~a... Yes, a hackathon. Thank you. Monday morning, haven't had enough coffee yet. So having hackathons~ within the people team, so we're-- I'm gonna intentionally be setting up time for the team to do that on a monthly or quarterly basis.
We're still determining like what makes sense. I I don't wanna put too much pressure on the team because exactly what you said, it's like we have a very... Like most organizations, we are running really lean teams, and so finding time to do everything, so I don't wanna put too much stress on, and pressure, but I wanna make sure that there's a designated space for teams to do that.
So we're gonna be doing that going forward.
Jeet Mukerji: That's awesome to hear. Yeah, hackathons seem to be the number one way of getting people to adopt AI within the business because they can really see what they can do. The other question I had for you was, yes, there's the AI learning piece and then there's also the we need to improve our processes.
So there's two things happening at the same time. And what you said at the top of the call was there's pressure from everywhere to [00:13:00] do something with AI and automate more and do a little bit more than with less or the same. How long can you give yourselves before the first piece of automation or the first AI agent that's gonna be launching?
Or is it more kind of circular? Is it gonna be three months of process improvements? How do you phase that, and how do you combat the pressure that you're facing to be like, "We gotta do something yesterday"?
Nancy Luschkowski: Yeah. I think for me as an operator, I feel like process improvement you're always doing that, right? It's a part of the product mindset. You really can't build an AI strategy until you map out your baseline data and your actual workflows, right? That's so we have to do that.
But I don't think you ever stop doing that and looking at them at every stage, like every point of change in the business, every stage of growth and reversely, right? If something happens in the org where there's attrition and you have to downsize, you should also be looking at this. So I think HR really needs to start, thinking like a product team, right?[00:14:00]
Measuring where that time is actually leaking and how much value. We really have to like double-check ourselves and make sure that the AI solution is really giving us the value that we need before we invest in another tool. I don't think that's just a one time and we're done.
I think we will, this is the evolution of technology, and it's gonna require us to look at our processes constantly.
Jeet Mukerji: When you're thinking about AI strategy, how does the AI strategy of your team interplay with the AI strategy of the company? Do you guys have a, quote-unquote, "seat at the table," or is it more everybody does their own thing?
What's a model that's working for you guys?
Nancy Luschkowski: Yeah. So ours is our actually our OKR cascade model. So for the company our OKR is related to AI, is really to expand AI operations, right? And also AI native operations that increase our [00:15:00] velocity, efficiency, and our customer value, right? Company-wide. So how that's transforming into our people team objectives is really transforming that employee experience with AI automation.
So really embedding AI-driven automation across all of our programs and systems and processes where we can with the goal of, giving employees time back for that high-value work.
Jeet Mukerji: Gotcha. So it's really thinking about your, again, with a product mindset, the-- your end user in mind and making your customers how do we improve the customer experience for HR, which is employees? There's a lot of different projects that can be done here. Do you guys have like a top-level set of metrics in terms of understanding how AI or automation is delivering value to employees or is it more project basis?
The reason I ask is it's often really hard to actually look at, hey, [00:16:00] this is the automations that we're doing and this is the direct time savings. And there may be better metrics out there. So I'm sure our listeners would be curious to hear your thoughts on this too.
Nancy Luschkowski: Yeah, I think baseline data we're using for, at least for cycle time is our ticketing system. But yeah, this is something that we're gonna really be focusing on our offsite is how do we define the value? And I think it's gonna be initially through each project, right? We have a lot- various projects for the remainder of the year that are really dedicated to identifying automation and AI.
I think, again, this is a work in progress, and we're spending all of this week actually working on this. Where I'm leaning more towards is our data, right? How do we get data in front of employees, in front of teams, managers, leaders, so that the business can make really informed decisions more quickly?
I think that, that is a huge win that how do you measure that? I think that'll be when we get there, right? When we're able to [00:17:00] get really insightful data. I'll say in the current state right now, that is a challenge for us based on our HCM and our, on our current data is our people analytics team and actually each HR practitioner, managers it's challenging to get the data we need.
There's a lot of manual manipulation that needs to happen to data to get it to a point where it's really insightful for us. So that's top of mind for me is how do we get real value.
Jeet Mukerji: Got it. Got it. And we're now seeing with things like MCPs or with your more, dare I say, traditional APIs, there are different ways of pulling data and then manipulating that data. And I think this is one of the things that we touched on a little bit before we hit the the record button, which is around build versus buy, and how do you decide what to use to be able to pull out that data and get it in front of people?
Where is your stance on that never-ending question?
Nancy Luschkowski: Yes, that's a good question. It's two fundamentally different AI architectures [00:18:00] essentially, right? You have that one, we'll call it the build your own kind of horizontal level desktop agent, right? It's really flexible. You can use the agent to really master that unstructured knowledge work with, messy local files and cross-application workflows using the MCP, right?
So on the other side you have our vertical record agents, right? This is our source of truth, very, deterministic, really deeply embedded in that vertical powerhouse which is fortified by our system of record. And it's en- they're engineered to be highly compliant and native for HR, really HR transactions only.
So I think the problem for me really is I think where I was struggling was I initially I felt like I had to choose one or the other. And HR teams really face that operational friction when we're trying to like force E- either one of these tools in the wrong way. So [00:19:00] I think for me it's not one or all.
I'm coming to realize it's gonna be a hybrid strategy on how we use these two models
Jeet Mukerji: It's great to hear. Yeah. The more and more leaders we speak with, it's it is a combination. It's build and buy and sometimes borrow
Nancy Luschkowski: Yeah. And, I think our source of truth is a really good way to your HCM agent or the native built i-in your source of truth is really secure. It's that back-end execution for those really heavily transactional work, comp changes, leave of absence things like that.
And then the homegrown is more of your kind of front end, what do we wanna call it? Employee concierge type
Jeet Mukerji: Right
Nancy Luschkowski: where it's in your Slack or Teams or wherever your team-- however the organization works. And it acts as a way to handle, complex triage in a sense, like reading through existing documentation it, understanding the history of prior questions asked.
Jeet Mukerji: [00:20:00] Yeah. And I have a feeling the answer to the next question is gonna change as you explore what's possible with AI even further, but I'm kinda curious to hear your thoughts. You mentioned employee experience a couple of times, and HR's product means that they're your end customer. What is the ideal experience for your end customer, the employees that you wanna be able to create?
Are we living in a world where they never have to log into an HCM ever again and everything happens through Slack? Or what does that look like for you?
Nancy Luschkowski: That is a good question, and I think that is changing, right? When we're thinking of, like, how the entire workforce operates. And this is a question I've had with for a long time, is for the employee, right? Regardless of what they need they're not thinking "Okay, this is an IT request, I have to go here."
Or, " I need something out of Salesforce, I go here." Or, " HR, okay, now I need to go here." It's like they just need something, right? They need something to do their work. They need [00:21:00] something to do their best work, right? And so how can we facilitate that in a way that's fluid, regardless of what they need? I think that's where AI is really gonna change, and that's how I'm- It's forcing me to also rethink that service model and really, have those conversations with partners in IT and other areas of the business to say, "How do we wanna accomplish that?"
And I think the complexity is that depending on... We might have to have multiple channels, right? Because depending on your role, how you work, it could be different. It could be Slack. It could be maybe people work directly in Salesforce, right? If you're on the go-to-market side of things. So we have to do that discovery to say like how does our population really work when we're talking about, like, when they're getting the stuff on their desks done.
And then building that. So yes, I... That, that is a, that is one of the challenges of this process. I'm calling it a process
Jeet Mukerji: Yeah. And that means a lot more conversations with other departments and [00:22:00] making sure that you're aligned on the AI projects. Do you guys have a-- We hear this a lot, like an AI council or something to be able to align all the different projects. How is that working for you
Nancy Luschkowski: Yeah, we're forming... So the organization is forming a AI center of excellence. And we're at the very early stages of what that's gonna look like, but I'm hoping, I think this is gonna be definitely , in the conversation and in the planning. It's like we need to understand all aspects of the business and how everybody works before we can make those really thoughtful decisions.
Jeet Mukerji: Sounds very intentional, which is great to hear. And when we're talking about like how the business operates, that also has an impact on your own team's operating model and the other parts of the HR team that you're also supporting, HRBPs and performance and L&D. How do you see the operating model for maybe the services team changing as you allow employees and managers to be able to self-serve more and more of some of those requests that you mentioned?
Or is that something that you think is gonna get figured out as you run [00:23:00] more experiments?
Nancy Luschkowski: I think that will get figured out as we run more experiments. Our whole goal in whatever service model we end up executing is going to be making sure that we can be super efficient in the triage, and that so that we can identify... We can offer self-serve everywhere where it makes sense faster.
And I think right now we don't have the automation designed in our systems to facilitate that. So that's like my P0, is like in order for us to do that, we have to go through a process health check through pretty much all of our HR processes to say do we understand what the trigger are, triggers are?
Do we have a really simplistic way to push that workflow through approvals or necessary additional steps? And then what is the output, right? Is that clean? Or is everything outside of the system, right? And so it's like manual touch. So we-- I think that is P0 so that we can enable more AI and so that we can then design those personal touch moments.
But I think [00:24:00] P0 is the inventory, the cleanup,
Jeet Mukerji: foundations and first principles. That is very good to hear.
Nancy Luschkowski: 100%.
Jeet Mukerji: So, Nancy, if you wanted to leave folks, our listeners today with one piece of advice on anything that we haven't covered so far for anybody who's starting this journey or halfway through their journey on how to implement AI as a process what would you like to share?
Nancy Luschkowski: So I think to avoid kind of analysis paralysis, my recommendation is that you-- it doesn't have to be either/or. You don't have to have your strategy fully mapped out and designed for you to implement some low-hanging fruit. So I think where you can optimize and automate quickly within your processes, start there.
And implement AI in your organizations by enabling your users first and enabling the skill. The skill that we want our workforce, our HR teams to have is the understanding of how to use AI, right? And we don't [00:25:00] need to create learning paths. We don't need to do a lot of enablement.
What we want them to do is use AI to teach yourself how to use AI your desk, right? Ask the questions. And I think that's a great start, and then enable your teams to build agents and then share with each other, right? That's how we're gonna, we're gonna create, AI products, is that build something within your own desk that helps you with your work, right?
Use AI to ask questions, even if it's, "Okay, AI agent, how do I use you? This is my job," right? " how Can you help me?" Even if it's just starting there and then really focusing in on this is a self-teach moment. This is how you need to maximize AI and then sharing, and then hopefully that sharing leads to experimentation and testing, and then something that you can apply in production, right?
The-- I think that's the big, that's most value that I think our HR [00:26:00] teams can do right now as we're, the bigger, longer, scarier hairier strategy and execution. Don't let that paralyze you, I think is my biggest
Jeet Mukerji: Love that. So it sounds like what you're saying, yes, we need to focus on cleaning up our processes and make sure they are correct in our systems before we layer our AI on top, but also look for the quick wins. Don't get into analysis paralysis forever. Use AIs to self-teach, and then crucially share those learnings with your team so that you can elevate everybody else.
Nancy Luschkowski: Yes
Jeet Mukerji: thank you so much for joining us on the "WorkOps" podcast and sharing where you are on the journey so candidly with AI as a process. And to everybody listening, we will catch you on the next one.
Nancy Luschkowski: Bye everyone