Workday VP of Talent Acquisition Adam Godson shares how conversational AI reinvented the frontline hiring process, and how his framework for process reinvention applies across broken legacy workflows.
Emily Faracca
Multimedia Content Writer
Workday
Too many AI strategies are built entirely within the vacuum of comfortable corporate offices by people who spend their days at desks. When boardroom assumptions dictate technology design, the resulting tools often clash with user reality—forcing a hybrid workforce to navigate rigid, multi-page forms that represent digitized bureaucracy rather than actual innovation.
Workday's General Manager of Talent Acquisition, Adam Godson, shares how to escape this top-down trap by aligning technology with actual human behavior in the field. He breaks down a practical framework to strip away legacy workflows, isolate core operational decisions, and deploy seamless conversational AI. Tune in to discover why moving past boardroom assumptions allows you to build a frictionless experience that ultimately makes hiring more human.
Adam Godson
Hi, I’m Adam Godson, general manager of talent acquisition at Workday. I served as the CEO of Paradox before our acquisition by Workday.
Today, I want to talk to you about the automation paradox. It’s the realization that in an age of infinite AI scale, the most valuable thing you can offer isn't more efficiency—it’s a return to human connection. Last year, the systems my team built scheduled over 32 million interviews.
Think about that scale for a second. That is 32 million human connections, career shifts, and economic opportunities, all powered by AI. But to understand how we reached that kind of scale, I have to take you back to a version of recruiting that couldn't have been smaller. 23 years ago, my recruiting platform wasn’t cloud-based. It was a physical sign I hung by the side of the road that said, "Now Hiring". I was a recruiter for a manufacturing plant, and on a good day, I could manually interview around 20 people.
That sign represents the legacy era of business. It was static, it was one-way, and it relied entirely on luck and manual labor. I started building technology because, frankly, I couldn’t clone myself, and I wanted to solve a scale problem. How do you give everyone an assistant when the person hiring has 100 other tasks to do? Fast forward to today, and we aren't just talking about digital versions of that roadside sign. We are in the era of agentic AI. To give you a sense of the scale that comes with it, our technology helped nearly 5 million people get hired just last year.
So yes, we solved the scale problem. But the deeper lesson is not just about volume efficiency. It’s about how you use AI to remove friction from a process that is, at its core, deeply human.
By the end of this masterclass, you’re going to learn how to build from behavior instead of assumptions, how to reinvent a process instead of simply digitizing the old one, how to learn from customers without letting them pull your roadmap in every direction, and how to build with conviction when user behavior tells you where the world is going.
Underneath all of it is one big idea: the best automation should create more room for human connection, not less - the greatest paradox of all. Let’s get started.
If you’re a CIO or a CHRO, you’ve likely spent the last decade perfecting the employee experience for people who sit at desks. But 80% of the global workforce—the people in your hospitals, your retail stores, your restaurants - kind of like the one behind me right now—don't have desks. When we look at frontline workers, their "office" is typically a mobile phone. If your AI strategy requires an employee to log into a portal and fill out a 12-page web form, you aren't innovating; you’re digitizing bureaucracy.
We have to stop designing for the process and start designing for the person. AI shouldn't be a destination people have to travel to; it should be a conversation that meets them where they are. In fast-moving environments, your competitive advantage isn't your database, it's how fast you can turn a stranger into a connection through a frictionless, mobile-first interaction. When we realized this, we stopped building better forms and started building 'Olivia'—a conversational agent that lives in text, or messaging platforms to communicate in real time.
We didn't just change the tech; we changed the behavior. And that’s what we’re going to break down next: how to observe where people are getting stuck, and how to have the conviction to build something simpler.
So, how do you actually do this? Let’s use frontline hiring as the case study. In the early days of building Paradox, we spent time observing how companies were operating by default. We went to physical stores and restaurants asking to apply to jobs.
The most common answer we got was people handing us a paper application and telling us to leave, or asking us to leave and apply online. Of course you’d never do that with the cash register!
We also observed the chaotic life of a store manager - serving customers, managing employees, and trying to manage the hiring process themselves through it all. We thought, “this person needs an assistant! What if we could put a recruiter in every store?” That’s the first lesson: if you want to build better technology, start with behavior. Go to the place where the work happens. Look at what people actually do. And don’t confuse buyer assumptions with user reality.
Now, let’s move onto the hardest part of the framework. It’s not the coding or the data science. It’s the conviction to stay the course when your customers, and even your own team, try to pull you back toward the safe way of doing things. In the era of AI, it’s actually pretty easy to improve a process. The technology itself can give what used to be an impressive uplift to many things, but that isn’t the same as transformation.
To achieve true transformation, you have to dig deep into a process, understand the why behind it, and find ways to change, eliminate and re-invent by getting to its core. In the early days of Paradox, that meant spending time in the field with customers actually observing and learning up close. That was particularly true as our product gained early fit in hiring frontline workers. People that bought the software typically worked in the corporate office - they weren’t frontline workers. So the observations and problem statements they expressed were from their point of view, but it was an echo of someone else’s point of view. We felt strongly that we needed to understand the challenge from the point of the view of the end user.
The persona understanding was a key point for us as we invented something new - not just what the product would do, but who our users were at a detailed level. That led us to a simple framework for process reinvention: Observe, isolate, and deploy. First, observe. Go watch the real workflow. Don’t rely on the documented version or the office version. Go to the place where the process actually happens and see it for yourself. Watch what people do. Watch where they get stuck. Watch what they ignore. And pay close attention to the needs of each persona involved, because most broken processes are really a pileup of different needs that have been flattened into one bad experience.
For the frontline, this led us to completely re-inventing the traditional job application. For a hundred years, people have filled out job applications by listing their qualifications, turned them in to be batch processed, and waiting.
But that process didn’t exist because it was particularly useful, it existed because communication wasn’t fast enough. Once we realized that, we started asking a much better question: what is the manager actually looking for when they read this thing? Which brings us to the second step: isolate.
Reduce the process to its core elements. Ruthlessly question everything else. We would literally stand behind managers reviewing a job application and ask them, “what are you looking for?”. And the answer was almost never, “I need every field on this form completed.” It was much simpler. Are they eligible to work? Can they do the job? Can they work these hours? Once you understand the real decision, you start to see how much of the process is just inherited baggage.
The third and final step is to deploy. Rebuild the process around the real decision, and use technology to unlock a better version of it. We re-invented the process to just ask those key questions up front instead of making a candidate fill out the form and wait. We hung a sign on the wall that said, Text Olivia to apply. The candidate answered the core questions the manager actually cared about — are you eligible to work, can you do the requirements of the job, can you work these hours — and then we scheduled the interview right then and there.
As we broke down the needs for each stakeholder, we realized it had become a process that didn’t serve anyone well. Candidates thought companies were slow, companies thought candidates were flaky, but in reality the process was just bad. By reinventing the hiring process, we took an early client’s time to hire from 21 days to 3 days. And nothing valuable was happening in those 18 days…it was all waiting.
So at that point we knew we had unlocked something special. It’s pretty common for people to think that it was the technology change in AI that made the difference, but it was actually the technology change coupled with process re-invention. And that’s what made it hard for our competitors to copy us. Everyone thought it was the technology and company after company tried to copy us without success, but it was really the deep work with customers to re-invent the process with technology that helped us win.
Reinventing the hiring process was a huge part of our early success, but we could not have done that work without the right customer relationships around us. There is simply no lab like the real world and your customers are full of insights and ideas. Now, they are also full of distraction and bad ideas, so the magic is in knowing the difference. For us, strong customer relationships came down to a few principles.
First, talk to a lot of customers and be intentional about the kind of relationships you build. The core is around expectations and how to balance the incentives. Our strategy here was to build deep, personal relationships and eschew the formality that would normally come with them. This wasn’t for everyone. Customers looking for documented statements of work, contractual commitments, hourly billing - those structures were not going to fit our vibe and we had to be okay with that.
Second, look for the co-builders. I can think of lots of times that we gave up potential sales in order to focus on the kind of customers that were willing to co-build with us. Having the right relationships also helped us a lot as we got things wrong. And we got a lot wrong, but had earned deep trust with our customers that allowed them to give us some grace.
Third, learn to separate the signal from the noise. The danger in building with customers is getting too deep in a narrow use case or point of view. Having many co-build relationships was helpful for us here to discern signal from noise. As we heard patterns emerge, we were able to get to the core signal, cross-reference problem statements and avoid building for niche problems.
And finally, understand your one-way doors. Some decisions are easy to unwind. Some aren’t. And when you’re building with customers, you need to know the difference. The deeper or harder-to-reverse the decision, the more important it is that you’re acting on a real pattern, not a narrow request. One of the clearest examples of that for us was the evolution of our own vision. Originally, the company was recruiting.ai, and the ambition was narrower. But over time, we kept hearing a version of the same question from customers: if you can do this for candidates, why can’t you do this for employees?
That mattered because it wasn’t just one request. It was a pattern. And that pattern helped expand our thinking beyond recruiting into onboarding, retention, and broader employee communications.
Some of the best product insight you will ever get comes from being close enough to customers that they trust you, and from talking to enough of them that you can tell the difference between noise and signal. That combination — closeness and discernment — shaped a huge amount of our success.
Building with customers was an essential part of building with AI, but having conviction in our point of view was also a key part of our success story. Keep in mind this was in 2017 - years before popular LLMs like Claude or ChatGPT became widely accessible household names. We fundamentally believed that conversational AI was a transformational technology that would change the way people communicated.
The trends of the world - the explosion of text messaging and mobile device use and the data we saw in our systems only hardened that conviction. And yet, sometimes customers were equally adamant that people wouldn’t like it. It was often because they weren’t the end user. They weren’t the frontline worker, so their convictions were based on their own assumptions about human behavior, not the behavior we were seeing directly. I recall one story where this came to a head with a large and influential co-build customer. Our point of view was Conversational Apply - the idea that applying for a job should feel like one person talking to another person, only using AI.
They were adamant that it should be optional, and the system should also have an Apply button that could lead a person to fill out a typical web form application like the one they had always used.
We said no.
It was a case where we stuck to our belief system. We had an overriding belief in our design principle - conversational AI was the main thing, not the side thing, and we were trying to move the world beyond the old experience, not preserve it inside the new one. The apply button would open the chat, but we were moving the world beyond forms. And yet, I think we also learned some balance in how far we could push change with our customers. We talked about the concept of skeuomorphism, where we can design things to be familiar to help people adjust to change.
Sort of like how the save icon on a computer looks like a physical floppy disk - the new thing that replaces the old thing has familiar elements of the old thing. So we used some familiar words around applications and interviews to describe the new experience while not backing down from making it new and more effective.
Governance is part of that balance too. Having conviction doesn’t mean removing judgment from the system. It means being clear about where you want the technology to lead, where you want humans to stay in the loop, and which decisions need stronger guardrails because they’re harder to unwind.
In hiring especially, the stakes are real and we took that responsibility seriously. And frankly, there were so many technologically cool things we thought we could do…but as we looked at the processes there was so much friction to remove, that the most impactful thing we could do was connect people to other people.
The other unlock for us was being all in on messaging as a medium instead of email. The consumer behavior trends supported this, but so did our direct observations of the frontline persona. For many people, the phone was the only computer they really had. It was where their time was going. It was the quickest, easiest, most natural interface available to them.
At the time, many people didn’t design for messaging first. Back then, even Twilio, which powers much of the world’s text messaging, was in its early days and things were harder. It had limited design options…like…basically none. But we liked the simplicity of it. Many builders didn’t use SMS because it was expensive at the time. Remember when you got charged 10 cents for every message? But, we didn’t care…we knew that it would commoditize over time and prices would come down. We held to our belief that if we built something valuable, we would figure out the economics. And we were right. Today, we send over 3 million texts a day, which seemed unfathomable back then. That 10 cent text that once felt like something you had to ration is now one of the most common and cost-effective communication channels in the market.
The same pattern happened with LLM’s. Early on, they were expensive enough that a lot of teams understandably got fixated on cost. But my team heard me loud and clear: I don’t care about cost - I care about making the best product. And now we’re seeing the cost of foundational models driving downward. Once behavior tells you where the world is going, you have to build toward it with conviction. If you can already see the shift clearly, you can’t let today’s constraints keep you trapped in yesterday’s design.
All of these lessons point to a bigger truth. The goal of AI in hiring was never to automate recruiting.
In fact, the paradox we’re named after is exactly this: our goal is to automate the administrative stuff so we can improve human connection. You buy software to make the experience more human, not less. That matters because recruiting is not online shopping. Recruiting is emotional. It’s tied to someone’s self-worth, their economic circumstances. It’s tied to possibility. And the moments that matter inside that process — the moments where someone feels seen, persuaded, welcomed, or excited about what comes next — those should not be designed out. They should be designed for.
So when I think about process redesign now, my recommendation is actually to start the other way around. Assume the technology can automate almost everything. Because it probably can.
Then ask: what human parts do you want to introduce? Where do you want there to be a real emotional connection point?
In some environments, that might be a phenomenal onsite interview experience. In frontline hiring, it might be as simple as a welcome call every time someone gets hired, just to say: we’re excited to work with you. But start there. Find the human touchpoint that matters. Then work backwards into the constraints, the policy questions, and the automation. Start with the human, not the automation.
AI is re-inventing hiring and the entire process is being re-designed. It’s an amazing time to be building - we get a chance to re-imagine a process as it should be to make it better for everyone. I can’t wait to build a brighter future together. Thanks for joining me.
Report
A gap between effort and impact is the real ROI problem. Discover the solution.