Sapphire Orlando '23 - are SAP's AI ambitions in line with customer priorities? Inside the SAP SuccessFactors generative AI news
- Didn't get enough about SAP's AI plans from the keynotes? Want more specifics on the strategy and timeframe? You've come to the right place, as I dig deeper into SAP's generative AI news for SuccessFactors with SAP's Meg Bear.
When I go to enterprise events, I try to go with an open mind. One of my biggest jobs? Identify the gap between what customers are talking about, and what software vendors are promoting on stage.
There will always be some kind of gap there - the only question is: how concerning is that gap? How big is the disconnect?
No surprise: SAP had a lot to say about AI in their Sapphire Orlando keynotes. That's not the only topic SAP covered by any means - another year, another batch of press releases so vast they require a table of contents. To make sense of it all, I taped two different podcasts, and still didn't cover all of the content:
- Making sense of Sapphire 2023 Orlando - with Josh Greenbaum
- Sapphire Orlando ’23 review - Brian Dennett has strong takes on SAP Datasphere, cloud business transitions, and the future of TechEd
Is AI a priority for SAP customers? ASUG members weigh in
But is there an AI priority gap between SAP and its customers? On the day prior to the Tuesday keynote kickoff, Josh Greenbaum and I spoke to a gathering of SAP customers via ASUG's Executive Exchange program. The pain points in the room centered around: data silos, persistent integration challenges, managing multiple SaaS software solutions, and concerns about moving to S/4HANA - in particular, finding the right partner, and upskilling internal team members (essentially, developing an S/4HANA skills and talent plan).
Does that mean AI is not a priority for SAP customers? Not necessarily. Though AI was not cited amongst the top issues in the room, we did an informal hand poll asking: how many planned to own their AI development - and how many were counting on trusted vendors like SAP to handle the AI innovations/risks/development investments? From the show of hands, only about 10 percent wanted to primarily focus on their own AI development and data science teams.
That puts the pressure of AI development - and customer results from those investments - directly on SAP. SAP certainly seized that opportunity during the Sapphire 2023 Orlando keynotes, with a slew of AI announcements, focused on what SAP calls Business AI. The irony of enterprise software AI? Some vendors have been pressing ahead on AI for years now. But the surge in generative AI consumer adoption (and large language model tech), has put vendors on their heels a bit. If you try to build all of this tooling in house, you're going to be in the slow lane when it comes to delivering that functionality. Clearly, SAP is determined to avoid that problem.
SAP and generative AI - partnering with Microsoft on SuccessFactors use cases
SAP is no stranger to AI; it's been engineering AI into products for at least eight years. During Christian Klein's keynote (replay here), he said SAP has "already embedded AI in many of our applications, being used by over 20,000 customers."
But generative AI is a different story. In my view, SAP's positioning around Business AI is about: SAP focusing on what it does best with AI (focus on business and domain content), and turning to "ecosystem partners" where needed for AI tooling. As SAP put it in the AI for Business press release:
SAP embeds artificial intelligence across our portfolio, leveraging the best AI tools in the market and combining them with our extensive industry-specific data and deep process knowledge to deliver the solutions that power our customers' most critical business functions – on day 1. Customers can deploy with full confidence and trust because SAP Business AI is built with responsible business practices.
This approach allows SAP to embed new generative AI features quickly:
On top of the new SAP Business AI functionalities launching today, from making procurement processes more intuitive to personalizing customer engagement, we also announced a generative AI partnership with Microsoft to improve how customers attract, retain and develop their most vital resource – their people.
One of the early use cases for generative AI in SAP's applications: SuccessFactors. As per this news, SAP and Microsoft Collaborate on Joint Generative AI:
The companies will collaborate on integrating SAP SuccessFactors solutions with Microsoft 365 Copilot and Copilot in Viva Learning, as well as Microsoft’s Azure OpenAI Service to access powerful language models that analyze and generate natural language.
Two early scenarios involve recruitment and learning. Which leads to my second fun job at enterprise shows: nail down details on what products will be available when, including licensing implications. And, ideally, find out the "why" behind the fanfare. Perfect time for an Orlando sit down with Meg Bear, President and Chief Product Office, SAP SuccessFactors. So, why SuccessFactors, and why now? As Bear told me:
With the introduction of AI, I don't think anybody is missing the beat that this is changing how we work. And there's a lot of energy around that. This was a really good opportunity for both Microsoft and SuccessFactors, to show not only how we're both innovating in this area, but how our innovation really complements, what we're each doing.
Okay, but what are the pain points and opportunities? Because I don't think a typical HR professional really cares if they are using generative AI nor not. But they all want to be smarter on talent, and less burdened by HR admin chores. Bear spoke to the issue of individual HR employees first:
From an individual point of view, there's two things: How do I remove the the overhead and the busywork? Microsoft's reports have some really good data about how much time we are wasting with non-productive work... It's very easy to see that things like writing assistance, things like distillation of conversation, things like understanding and catching up with a conversation [can help]. Maybe you're new to join a team; to be able to catch up: what's the conversation that has been what's going on? So accelerating your time to understanding and competence, and then leveraging your time more effectively.
This technology, however, brings both opportunity and a touch of fear: will it help me, or outperform me? Bear says we need to tackle this head on:
For an individual concretely, I think there's both enthusiasm, as in, 'Oh, maybe this will make it easier for me.' And there's maybe a tiny bit of anxiety of, 'Hey, does this mean that my job is under threat?'
And so what we're very keen to do is to help ease that burden a little bit, and clarify how things will get easier: what kind of things we can see coming very quickly to reduce complex tasks and make them simpler? And then, how can we help with the skill requirements of the jobs changing, as the nature of work changes, so that people feel like they can be equipped quickly, and can keep up, because this pace is going to require change.
Behind the first SuccessFactors generative AI use cases - why recruitment and learning?
Let's drill into the use cases. Start with recruitment - as per the news release, SAP plans to take advantage of Azure's OpenAI API:
Through an integration between the SAP SuccessFactors Recruiting solution and Microsoft 365, people leaders will be able to fine-tune job descriptions using Copilot in Microsoft Word with additional content and checks to detect bias. The final job descriptions will then be published in SAP SuccessFactors solutions to complete the workflow, without people leaders having to leave their normal flow of work.
Writing job descriptions are a sensible generative AI HR scenario, earning my rare use of the dreaded "low hanging fruit" buzzword. Job orders are tedious to write, can tolerate a degree of the type of written errors these AI systems are prone to at times, and have a number of specifics that generative AI is good at compiling (human supervision, of course, should catch/correct any of the AI system's overreach). The automation around the SuccessFactors job description process is appealing also. As Bear pointed out to me, this type of generated writing is also ideal for those, for whatever reason, find writing cumbersome or physically challenging (examples: dyslexia or English as a second language). She adds:
[Job descriptions] also don't have any private data. So it's very easy to take it on without concern. When we look at bringing in AI technology, we are very thoughtful of ethics, privacy, and transparency for individuals. And so as you say, job descriptions are a really nice starting point, because they they are something people understand; they have a lot of value; a lot of time is wasted.
And yet, even something as seemingly innocuous as a job description can contain a lot of bias. Bear hit on a theme I heard a few times from SAP at Sapphire: can we utilize AI not to compound human bias, but to correct it - perhaps diversifying our applicant pool?
We know that job descriptions can be a source of bias, and can cause certain parts of the candidate pool not to be seen. The better we can get at this, we can not just improve the time for the individual, but actually speed up the time to finding the right people to come join your organization.
Remember, when you're hiring, you're hiring for a reason - you're hiring for a business urgency. And so the faster and the better job you can do, bringing in those diverse skills and talent, the more you're going to be able to get on with the business of doing the work that you're trying to do.
That's the fascinating thing about HR: seemingly straightforward AI use cases carry big ethical implications. Then, this new SuccessFactors AI scenario is definitely one where the data privacy questions must be fully addressed:
SAP will also leverage the Azure OpenAI Service API to offer prompts to interviewers within Microsoft Teams with suggested questions based on a candidate’s resume, the job description and similar jobs.
That is one where we're having a conversation right now with our ethics and privacy group, to make sure that we put the right guard rails, the right transparency and the right clarity behind that - both for how we enable it, but also so that our customers can decide what makes sense and where to use it.
This is a really good use case for how we're thinking about all of our AI innovations. We're looking at: how do we solve big problems? How do we do this with a lot of transparency? How do we have a lot of understanding for the fact that this is going to evolve over time, so that the burden is not held with our customers, but that we can absorb both the upkeep and the compliance sides of this, to provide good guidance for them to make decisions on where and how to use these tools.
My take - on timeframes, and the ethics of AI potentials
When you think about the customer priorities we heard about from ASUG members, including the problem of data silos ("operational islands") and integration pains, those concerns may not be as far off from next-gen AI as it appears.
Your "AI" is only as good as the data you put into it. If SAP helps customers sort their data issues, in turn, that puts customers in a much better position to absorb AI functionality across SAP's product lines (mostly it's the cloud product lines that will have access to SAP's embedded AI, but that's another potent topic). SAP would argue that their aggressive push into Datasphere is a big piece of the puzzle here; that's beyond the scope of this already-long-enough piece.
As for timing on general availability, Bear says SAP is targeting the "2nd half" SuccessFactors release, which is typically early November. It's possible not everything announced will be ready then - that's about getting the issues in this piece right before it's rolled out.
You can criticize SAP for many things - I certainly do - but SAP is definitely vigorous about the data privacy aspects that inform these use cases. Bear is right: SAP customers will count on SAP to deliver this functionality to them - but also to properly de-risk it and handle the AI experience transparently. During Q/A sessions at the show, SAP repeatedly noted the opt-in aspects of AI functionality.
That will be especially important when partners are involved, especially if any customer data is leaving enterprise walls or SAP controlled areas. Too many vendors are taking an unsavory "opt-out" path here, rather than the proper opt-in, or: they are looking to monetize AI with a premium apps charge. Whether SAP is going far enough with its SAP pursuits remains to be seen. Some believe that ERP needs to be totally re-built with AI as the core. I'd argue SAP may be trying to re-invent ERP with sustainability woven into granular transactions as the guiding principle.
If so, perhaps that will set up fascinating comparisons in an ERP market where vendors would ultimately make different bets on which emerging technologies matter the most to modernizing ERP. I'm not saying SAP isn't serious about AI. But, having heard a number of vendors announce their AI plans this spring, I wouldn't say SAP's AI messaging is differentiated. Some at SAP may hate to read this, but if it gives any comfort, I only heard one vendor talk about AI in a differentiated way, and even that pertained to future possibilities, not current functionality.
This does not mean SAP isn't differentiated on a product side with AI, but that requires deeper product demonstrations, comparisons, and customer validations. At any rate, the differentiation aspect is a bit overrated, because every vendor, SAP included, needs to thoughtfully integrate AI capabilities into their products. Serve your own customers - that's far more important than getting some kudos for appearing to be further along.
What about more radical changes to user experience, provoked by generative AI? At multiple shows this year, I heard discussion about the possibility of a prompt-based front end for HR. This is not as far away as some might think. I have a friend at a major SI who is already able to automatically complete a number of HR tasks (like provisioning a new employee), without having to navigate any screens.
However, as I said to Bear - and also to an SAP AI panel - prompt-based software access raises a huge batch of fresh questions about bias. This applies to consumer-search like Google and Bing as well. As soon as an AI system is showing me just one result, rather than letting me scroll results, there are concerning implications for exclusion, bias, and the dominance of the incumbents that dominate the top search position.
In other words, if AI designers inside and outside of software vendors get too headstrong about disrupting legacy user interfaces, they could fast forward past serious ethical (and legal) considerations. I found SAP's answers to this pesky question useful. SAP, including Bear's team, is looking hard at the issues of bias. Bear points out that the bias of how you ask a question, such as "show me the best applicants for a grocery management role in the southwest," can bring its own biases into the mix. The best way forward, as SAP sees it, is to design AI systems that compensate for and correct human biases. For now, Bear sees such an interactive HR interface as something that could be another way to work with the system, not the only way.
If we can do that, I like our chances. Notice that Bear didn't talk about massive job losses. These changes feel more like enabling HR pros to do their jobs better, not lose their organizational standing. That said, the issue of unprecendented change cannot be ignored. Bear says one way to handle that is to look for opportunities, and "bring yourself forward a little bit." If AI surfaces an opportunity, push to be considered. As per the AI-enabled learning announcement, if your personalized learning path surfaces something useful, grab it.
And: you should be able to update or request a change to the skills profile AI might build for you. If the AI misrepresents or underestimates you, you need to challenge or correct it. In my view, properly designed enterprise AI will enable that. Bear:
I don't want to leave out that there is collectively a lot more learning to be done. And so the most interesting and the most important bit is to look at the results. And for us to be evolving these solutions as we learn what is happening in real life... I do believe we're going to find things we didn't anticipate.
We're leaning in across every dimension, to ask ourselves: how do we remove the burden, how to elevate the role of the individual and how do we bring that value to the ultimate work experience that we're trying to deliver.
A worthy ambition - stay tuned.