This post was originally published on this site.
I’ve already given vendors a demerit for downplaying their best customer stories this fall – in favor of “AI-agents-are-coming!” superlatives. In my event halftime review video with Brian Sommer, that was my top bullet:
1. Vendors are overhyping AI agents
– this hype reduces clarity on AI feature rollout, customer adoption and ROI
– but it also obscures the truly cool things customers are doing now, with mature tech/apps
However, I give vendors like Workday credit: they were able to address burning questions, via my second takeaway:
2. Enterprise AI architecture matters for better output/results
– think smaller, customer-specific models, RAG, foundation models, industry LLMs, “grounding” with customer data
Vendors don’t need to pitch AI so bombastically. Customers are ready to kick some serious tires, but they want to learn more. Workday has now taken this further, via two (free) AI education courses. At Rising 2024, I worked those courses into how Workday’s enterprise AI architecture has evolved, and how AI connects to finance transformation.
Workday Rising 2024 customers – stories of gen AI in action
But there was another notable highlight of the fall season: customers are starting to use generative AI features in an enterprise context. When it comes to on-the-record stories, it’s still early days, but it’s informative to hear about the learning curve, as well as the business case. Yes, we’ve documented plenty of “classic” deep learning use cases, e.g. predictive AI, recommendation/personalization, etc., but generative AI is a different animal, and needs use case scrutiny.
At Workday Rising 2024, I had the chance to hear generative AI field stories from two notable customers: DataStax and Hewlett Packard Enterprise (HPE). For Sadie Bell, VP, Innovation of People Systems, HPE, her vision is to apply AI to create a different kind of HR than we’ve had in the past – one where HR leaders bring innovation forward, rather than getting bogged down in the administrative HR that’s all too common, even today.
Before we dove into the interview, I treated Bell to my rant on how technology should result in a more inclusive approach to talent – not the exclusionary, mostly rules-based systems that screen out talent today – while HR leaders talk about skills shortages that are, at least partially, of their own making. How does an HR leaders like Bell react to that? She responded:
I think technology, AI specifically, is helping to illuminate the problem and also solve it, or at least create a path, because the solution is in the action – and people’s trust and willingness to try what the data and information is suggesting. And so I think one of the important things is that it’s easy to judge a book by its cover, right? It’s easy to look at a resume, and notice your favorite college or university or a 4.0 GPA – which could mean nothing or can mean everything – at the exclusion of those who do have the skills and the talents or the experience, maybe not in the same area.
How does HPE use gen AI in HR?
Bell’s comments struck a nerve: I am intrigued by AI’s largely untapped potential to challenge human biases, but it can be controversial to apply that to a high risk area like HR, where such tools can cause real professional harm when misused. Then again, I believe a lot of applicant screening technologies cause such harms today, so this is not, in my view, a new peril. But how can AI support a more inclusive type of HR? For Bell, it starts at a personal level, where gen AI tools have been a big time saver in her work correspondence. But it doesn’t stop there:
I mentioned how I use it to write emails. It gives me a starting point when I might have writer’s block. We see this in the form of Copilots. We see this in the form of generated job descriptions. We see this in the form of, ‘Hey, sometimes we have biased language that’s being used,’ and we don’t realize it, but we can use Artificial Intelligence to look at it and say: ‘Hey, you know this word can make a certain group of people never look at your application. From the start, you have already excluded a huge workforce of abilities, skills and assets that you actually need.’
HPE’s HR transformation – how data drives skills and talent changes
In my last Workday Rising piece, I noted how data-rich areas like Skills Cloud are a potent entry point into AI – something Workday could have showcased even more at Rising. HPE is a terrific example. Bell told me about their move to Skills Cloud three years ago. That move helped to enable a shift away from degree-based hiring criteria:
A few years back, we shifted to being more skills-based hiring, and really looking for untapped areas – and places where we could reach a wider audience of potential team members, and not at the point of exclusion because they don’t have a four year degree from the best college on the earth, which is debatable, right? But they have the skills or the experience to fill our needs. As part of that journey, we have to make shifts in HR to be able to support that need.
One of those shifts was implementing Workday Skills Cloud, among a few other products and technologies in learning, and in the Career Hub, and so making the connection both internally and externally. Bringing in the Skills Cloud is really particular to internal, and not necessarily what we use for the onboarding process or pre-boarding process. But really looking at: what are the skills that we have in our organization today? What are the skills we need in our organization now and in the future, and how can we upskill, reskill, and attract the talent to fill those gaps – so that we continue to accelerate and be at the cutting edge of technology and driving a best-in-class solution?
This made mentor matches possible, at a bigger organizational scale:
We’ve had pockets of places where we would try to do this before, but Skills Cloud really helped us to seamlessly turn it on, identify skills that people may have, let them confirm those skills, add new skills associated with their current role or previous roles, and then connect it to opportunities to learn, for opportunities to mentor, and for opportunities to have other internal jobs or gigs within the organization. We have seen people applying for gigs across business, across function, to learn, grow and develop.
Especially in a remote-enabled work environment like HPE, it can be hard to make mentoring connections on your own – now that’s changed: Bell:
We have mentorships where now you’re able to connect, because the system recommended this person who has the skill you want to grow in, learn more about and develop. It’s made connections across the organization that we didn’t necessarily have as easily before.
The mentorship program has become so successful that at times, Bell told me she has to turn her mentor availability off due to how many opportunities are surfaced. Bell says there is a good KPI story emerging also:
We have some pretty good metrics. I wouldn’t say all of this is attributed to turning on Skills Cloud, as I think across most companies, we’re seeing attrition be lower this year. But internal mobility is up 40 percent the last two years, and so we’re seeing a lot of benefit.
“I want to automate the heck out of this” – DataStax applies gen AI to Accounts Payable
On the Workday cloud financials side, I picked up on a similar pattern: AI is an emerging benefit from a data and process discipline already in place. During my on-site interview with DataStax, CAO Arnulfo Sanchez shared how they do scenario planning in Adaptive Planning on the fly, making tough/important decisions about where to allocate resources, not as an annual exercise, but whenever circumstances change. Sanchez looks forward to using Adaptive Planning AI to do a comparative five year forecast – and this isn’t far off.
As Sanchez told me, DataStax is already moving ahead with some generative AI initiatives. Example? He specifically cited Auditoria’s Intelligent Automation for Accounts Payable (AP), which is fully integrated into Workday (here’s a link to the details Auditoria presented at Workday Rising). As I see it, things like AI for AP are on an automation continuum, with the goal of eliminating as much manual work as possible. This is not as simple as it sounds, when you consider that different business partners may have different payment terms, which in the past may have been addressed via custom fields, or manual attention. Sanchez’s put it best:
I want to automate the heck out of this.
But as Sanchez explained, that also means working with the Accounts Payable team, to make sure the automation fits the needs of their clients. That’s the big headline from this fall: winning gen AI projects thoughtfully integrate this tech into human workflows, sometimes automating, sometimes “assisting,” sometimes monitoring (e.g. looking for outliers), sometimes handling initial output (first drafts). The change should be met head on, transparently. Sanchez:
One thing that came out of this project with Auditoria is: you have this mindset of, ‘Okay, I’m going to use this tool. We’re going to get more efficient quickly.’ The one thing maybe we didn’t appreciate as much is: we also have to change our own processes in order to make this this work… I can’t overstate that enough.
What makes these AP automations “intelligent”? One example: Auditoria’s solution is set up to handle Accounts Payable information requests by email. During this early phase of model training, Sanchez says that the AP team is reviewing these emails before they go out, but he is confident that a significant time savings will be achieved.
My take
Sanchez acknowledged that AI/automation projects can stoke internal fears of job loss. DataStax has an instructive way to handle this: address the future of work at head-on, and: give employees a flavor for new roles they could take on. Some companies are doing a poor job of communicating what AI means to them. DataStax is going for a different tone:
Our CEO talks about – and I’m not going to exactly quote him – says: ‘AI is not going to replace humans. Humans that know how to use AI will replace humans that don’t know how to use AI’
That’s what we talk within the finance organization: ‘Hey folks, we all need to be open to this. We all need to understand the tools that are out there to make our jobs easier.’
I’ve been critical of the emphasis on generative AI productivity. It’s a mechanistic framework, and a tough metric for any technology to achieve. DataStax is asking a more compelling question: what will employees do with the time saved? Though it’s still early days with these AI automation projects, Sanchez is confident enough to start doing job rotations:
That’s what we did within our 30 person finance team. We recently did job rotations, because of some of the efficiencies that we expect to come out of these projects that we’re doing.
An AP person rotated to AR; others went to R&D, marketing and sales. This ties back to the Workday Rising theme in my last financials piece. As I wrote:
All the shiny AI in the world won’t disguise poor collaboration skills. Workday’s steward -> partner CFO evolution shares that same narrative… Over at DataStax, Sanchez told me he uses that same “partner” language.
HPE’s Bell hit on the same theme:
We’ve seen this be a big transformation within the organization, to say the least, and we’re not so far on our journey. We still have a long way to go to really capitalize on the potential that we unlock.
Spare me the “we’ll get rid of our call center” posturing, the “one person will be able to run a billion dollar company” fantasies, and the absurd notion that AI allows you to rip out your ERP and CRM. The best generative AI stories will emerge from organizations that can tell a better narrative about their employees – and how they apply that hard-won data.