'We have to come up with a plan for them' - Workday leaders on the fate of workers displaced by AI

Yesterday, Workday rolled out a multi-purpose, self-service AI agent as part of its new Sana for Workday AI-powered user experience. This is expected to autonomously complete many of the requests and actions that people have traditionally taken to an HR helpdesk or shared service center — and other administrative and routine tasks in finance. This of course has serious implications for the people currently employed to do this work, a consequence that Workday’s leadership acknowledged when speaking to media about the Sana roll-out. Aneel Bhusri, the company’s co-Founder, who recently stepped back into the CEO seat, concedes:

I do think a lot of low-level HR work is going to get replaced by agents. There's no way around it, and what the industry needs to own, including Workday — and you'll see more from us on this topic — is we have to figure out a way to take care of the employees that are, you know, dislocated for no better word, because of AI. We have to come up with a plan for them.

Because what you saw with Sana, there's a lot of manual work that was done in weeks that now gets done in minutes, and there's no company that's not going to want to do it that way — it's the better way. But we have to find a solution for the people that get displaced.

A big part of what Sana also does is, and Workday does with Learning, is retraining, and we've got to double down on the retraining side.

He adds:

I hope there's a world where AI is complementary to humans. We have to find a path to that.

His colleagues were quick to point out that AI automation isn’t just about replacing the work that humans currently do. Gerrit Kazmaier, President, Product & Technology, cited the experience of NetApp, a customer that used the Evisort contract intelligence agent to analyze 90,000 contracts and found an estimated $2.5 million savings, mainly in procurement. He explains that this was analysis that simply wasn’t previously economic to carry out:

There was never the question, should we hire 600 lawyers for that? It now becomes economically viable to do things that were not conceivable, before we had AI, to actually apply reasoning to those problems. 

Now, with AI being available, the far greater opportunity, and the far greater growth opportunity, is going to be in doing all the work undone today because of the changed unit economics of reasoning. And I think that's the big opportunity that creates space for higher-value services and higher-value functions for people impacted.

Return of the polymath

The upshot is that, while repetitive administrative tasks will be performed by machines rather than people in the future, there will be new jobs needed to supervise and act on the work of agents that take on those tasks that were previously uneconomic. This means that, instead of learning how to do a certain thing and then just doing it repetitively, people at work will prosper when they’re able to quickly pick up new skills or work across different domains. Joel Hellermark, formerly CEO of Sana and now SVP and GM of AI at Workday, expands on this theme:

My hope is that this era will result in the return of the polymath, that will see a renaissance of sorts, where people can very quickly upskill into new domains, and will remove some of the specialist needs, and people will increasingly become generalists.

I do think there's demands that will become increasingly automated, but I do think we're also reducing the time that's required to get into some of these roles, such as software engineering, and hopefully we'll be able to reskill folks into those much faster.

Kazmaier speculates that as AI takes on more repetitive tasks, this will lead to a redefinition of jobs and roles from which those tasks are removed but then others are added — effectively an unbundling and rebundling of traditional roles to create completely new combinations and permutations. He comments:

At a more fundamental level, what AI does is, it's basically unbundling the job architecture, because today, humans are the only form factor that reasoning is available in thus far. And that means that the jobs that we have today, like coding, like legal and reading contracts, like hiring people, like managing payroll, they're all limited in how many people can you bring at what costs?

But now suddenly, we can rethink the definition of a job itself, right? Because we can decouple writing code from developing a good software architecture. Or we can decouple finding information and contracting and procurement from having a great oversight on a procurement strategy. Because the one we can scale via AI, and the other one is still innate human reasoning and creativity and oversight.

This new 'job architecture' will introduce an operating model in which people's capacity is augmented by what AI can do for them, where they will collaborate more extensively and their job definitions will change more frequently. He predicts:

We're going to see a definition of job architecture that is separating everything out based on tasks. We will see teams which are not static anymore in long-running organizations, but far more fluid, and a workforce planning cycle which is continuous based on unlimited reasoning in the form of AI agents supporting people doing the work. And so companies are getting far more agile.

Finance adoption

In the immediate future, Bhusri comments that AI will accelerate the pace of IT modernization in enterprise functions such as finance that are still in many cases running on legacy systems. Typically these functions previously had less incentive to move to the cloud because they didn't have as broad a user base as applications like HR and CRM. But as AI becomes capable of taking on many tasks that previously had to be done by people within these functions, the potential benefits of adopting AI will force a re-evaluation. He explains:

Finance applications still are primarily run by the finance team. So if they had a general ledger that was working and had customized it to make it work for their business, and could generate all the statutory regulatory reports they needed, well, maybe there wasn't that same impetus for change, because they didn't have thousands of users. It was pretty limited.

But in the case of AI, finance, of all places, is looking for cost savings. There's a lot of work done, whether it's by third parties, like financial auditors, regulators, or any work, that now can be done by agents, whether it's the financial audit agent, or just done in a better way. [For example,] the financial test suite agent, just make sure that you're ready for your audits, almost on a real-time basis. You can almost do an audit at any time. That was not possible with either the legacy on-premise systems or the newer cloud systems. They were still business process automation systems, not reasoning and probabilistic engineering systems, the way that we've talked about today.

So for the CFOs, they look at it as, 'Hey, this is the new way to differentiate how we do business. We need to embrace AI. That's the reason to finally move from my legacy system to an AI-driven cloud system.'

My take

It's good to see Workday's leadership thinking seriously about the impact of its technology on existing workers and their roles. Introducing AI agents that are able to automate many tasks that existing workers have previously carried out will require careful change management and leadership across every enterprise. Those organizations will be looking to HR vendors like Workday to help support these complex transitions in people's job roles and the associated reskilling and redeployment of the workforce. Rolling out the technology is the easy bit — the bigger challenge will be helping people adapt to their new roles and responsibilities.