SAP tech leaders met at ASUG Tech Connect at a pivotal time. But what got them on a plane to New Orleans? AI is a corporate imperative, but S/4HANA deadlines also loom. Here's my take - with views from customers and on-site interviews, including SAP's Juergen Mueller.
SAP TechEd 2023 served up a slew of announcements, and the emphasis was no surprise: generative AI.
Thanks to candid (virtual) discussions with media and analysts, SAP brought more clarity to its AI strategy than I was expecting (I reviewed that last week in SAP TechEd 2023 news analysis - what did we learn about SAP's AI strategy, pricing, and developer engagement?)
But how does that match up with customer priorities on the ground? I had a chance to find that out last week - via a series of jugular conversations at ASUG Tech Connect 2023 in New Orleans - including an interview with SAP Executive Board Member and CTO Juergen Mueller, and in-depth conversations with ASUG and DSAG leadership (DSAG's Jens Hungershausen, Chairman, Board of Directors, was also on site in New Orleans).
SAP tech leaders at the AI and S/4HANA crossroads
Tech leaders are facing a crunch time of competing priorities. Mueller told me one CIO said he was "swatting a thousand mosquitos." The demands of generative AI, cloud transitions, and - in the case of SAP CIOs, S/4HANA go-lives - bring an onslaught of technical challenges.
Oh, and while you're managing all the risks/demands of these projects, make sure you deliver continual wins to the business - and create a technical architecture flexible enough for direct-to-consumer and subscription/service business models. Sound like fun?
I believe the urgency of these challenges accounts for the strong attendance for the inaugural ASUG Tech Connect event. You'd have to be some kind of superhero to get all of those things done, which may be why the marketing theme of ASUG Tech Connect revolved around becoming an SAP superhero. But the emphasis in New Orleans was really about strengthening teams: attendees were on a mission to get the info they needed, for whatever their teams were facing next.
I'll get to those priorities shortly, but first, for those who are wondering, ASUG Tech Connect was not intended as any kind of TechEd replacement in the US (SAP TechEd was on-the-ground in Bangalore, complemented by a global Virtual SAP TechEd event). But ASUG, in collaboration with SAP, perceived an opportunity to give SAP tech leaders a chance for peer-based learning and roadmap discussions with SAP - the kinds of experiences that are very hard to achieve via a virtual event (though, in my opinion, not impossible, but that's a debate for another time). As one attendee told me, "I got the TechEd news last week; now is my chance to learn what it means for my organization."
Behind the SAP TechEd news - SAP's Juergen Mueller shares his view
But what about those who weren't able to be in New Orleans - or at last week's TechEd events? What should they make of the news? That was my first question for my sit down with Juergen Mueller. Mueller responded:
Every developer is expected to be versed in generative AI. What we've done is we really made it simple for development roles. They don't need to be an expert, but to become a generative AI user, I would say. In SAP Build Code, you'll see it, and in the Generative AI Hub you'll see it, and then of course, the most technical announcement was the HANA Cloud Vector Engine.
But via the Generative AI Hub, we make it very, very easy. So if you saw the keynote on TechEd day two, where basically for every document you have, you can say, 'Embed this, embed this, embed this' - and that's all you need to know basically, to be able to use gen AI in combination with companies' documents.
SAP believes it now has the right data solution for AI also. That is crucially important to enterprise AI, where Large Language Model data goes stale quickly:
We talked about Datasphere significantly in the keynote as well. So we will put all SAP company-specific data there, and we'll be able - and our partners and customers will be able - to refer to that when they build Large Language Model-based applications.
Yesterday, GPT-4 Turbo was announced. That's really the latest model that has been announced. The training data has cut off in April. So it doesn't know anything about what's happened since April, and certainly it doesn't have access to any company's internal information. And that is where we come in, to really help build the bridge, and combine both worlds - the LLM world, and also company-specific information.
Mueller also shared the "concrete GA date" for the HANA Cloud Vector Engine: Q1 2024.
But what about SAP customer priorities? What we learned at ASUG Tech Connect
But are customers on the same page with SAP's AI emphasis? On the ground in New Orleans, the answer was mixed.
SAP's generative AI roadmap session was well-attended, as was ASUG's early reveal of its BTP customer survey. But SAP's "clean core" session was incredibly crowded. I believe that reflects the imperative of moving to S/4 effectively - and managing SAP's technical footprint in a more cost effective way. But don't overlook this clean core objective: the ability to capitalize on new business models - without as much technical friction
After the keynote, Josh Greenbaum and I did a session on decoding SAP, moderated by ASUG's David Wascom. During the talk, I did a hand poll, asking customers to say whether they were at Tech Connect primarily for S/4HANA, or for AI info. About 97 percent said S/4HANA.
Though this show was not primarily for hands-on developers, the future of SAP development was a hot topic here too; SAP customers want to make sure their development teams are on the right skills trajectory.
One of my favorite lines in the entire show came from the session by Paul Modderman and Jelena Perfiljeva, aka The Boring Enterprise Nerds. During the provocatively-titled The SAP Skillpocalypse: How To Not Become Extinct, Perfiljeva advised:
Use internal teams for strategic work, outsource mundane tasks (not the other way around).
Bingo! But in order to do that, you may need additional developer training, a topic I took up with Max Wessell, Executive Vice President, SAP Learning, who was also on site for a guest keynote (more on our conversation shortly). As for customers, here's the sum of what I heard:
- Direction on S/4HANA is a core pre-occupation
- The "clean core" is arguably SAP's most potent technical concept, and
- AI better come with a business case. Oh, and
- How is effective AI possible when our so-called "transformation" is stuck in data silos?
This exact issue flared up at Constellation Connected Enterprise a couple weeks ago. We were in the midst of a sexy debate about generative AI possibilities, but a conversation about data governance broke out instead (CCE 2023 - the great AI debate shifts from content creation to...data governance?)
During my session with Greenbaum, it didn't matter whether we talked about AI, S/4HANA, or broader transformation - it all came back to data. As one customer told us:
We had a digital transformation roadmap first, to achieve our company's initiative. And we have all the other pieces underneath. One of them is our S/4HANA journey, which we are kicking off. And with that, we also have our data strategy. It all goes hand in hand, because you can't take the crap data we have right now in our ECC, and put it into S/4 0 harmonizing our data, cleaning our data; even knowing what data we have.
But data governance projects shouldn't just be about AI-readiness; they should be delivering other benefits. In the case of this particular customer, one of those benefits is data self-service. On the other hand, a steady diet of acquisitions makes data governance even more daunting:
We're creating that data dictionary, so that people can self-serve, and get the information that they want - and it's coming from one place. We're a company that's had a lot of acquisitions; we don't even use the same terminology for things. So we're building a glossary, so that when we call something 'this,' we can holistically, across the same company, have the same context. Those are the things we're working on. It's really hard. It's overwhelming. But we're just getting started.
One of the attendees said their AI pilot project was a fail. Why? Their data simply wasn't good enough:
I was just curious on your perspective on how much data cleansing efforts are needed. We did a small POC recently with AI on our blueprints and functional specs in Solution Manager - just throwing them all out there and then putting an AI engine on top of it. It was truly 'Garbage in, garbage out.' We have stuff out there, different versions of blueprints, giving wrong answers based on that. I'm dealing right now with a BI data load that was done years ago with bad data from a BI system. It was just masked by another load that had a timestamp above it. So I'm just curious on how much people were estimating: what is the data cleansing strategy before you start feeding an AI engine, before you start to expose that?
Another attendee explained the data/analytics obstacles versus business priorities:
We're expanding our business into some new markets, and we're struggling with capturing the data - the information we need. Our analytics - nothing is real time. We're exporting and importing spreadsheets, so we fundamentally believe we have to get to this data faster - if we want to be an AI-driven company. And to do this data strategy, we have to have governance around that.
Josh Greenbaum is one of the grouchiest analysts I know when it comes to AI, but what's important is why. As it turns out, there's a glass-half-full angle to Greenbaum's AI discontents: What if generative AI is the corporate push our neglected data quality efforts need? During our "Decoding SAP" session, Greenbaum hit on AI data issues early. As he said during our ASUG Tech Connect review podcast:
I was sort of joking when David Wascom introduced us and said, 'Well, I have this question about Gen AI.' And I said, 'Yeah, that's the only AI question I'm going to answer,' because I didn't want to talk about it. But you know, now I'm deciding this is a really good aspect of gen AI: whether you're for it or against it, whether you see the business case today or the future, the first thing you're going to have to do anyway is clean up your data.
Really do that data governance - that data management thing that probably should have been done a long time ago in order to be able to have clean data. Like that one attendee who said, 'Garbage in garbage out,' that's what you're gonna get. So if anything calls attention to data quality and data governance, Master Data Management, maybe I'll be a little more positive about what we're all doing with gen AI.
How should tech leaders respond to the AI/data imperative?
If we accept this imperative, where does the CIO go from here? Hands-on developers must now contend with ABAP Cloud, BTP, SAP Build Code and other low-code tooling. But the CIO has to find a way to lead that effort - and that's not all: 'Okay, I need to upgrade S/4HANA? I haven't done that yet. The deadline is coming. But I also need to address AI.' How do you think about all these things at once - and still serve your business stakeholders, who won't wait for a ticket to be heard? Mueller responded:
Yeah, it's a good question. Because a lot is coming at CIOs or Chief Digital Officers, and I'd say you need to pick a few strategic partners. Everyone is now coming at CIOs in a positive way, sharing that they want to help with bringing gen AI for this niche solution and for that niche use case, etc.
In the past, the pressures to transform seemed to vary by geography - but not so much anymore. In the past, I recall hearing from DSAG board members that the imperative to "transform" was more of a vendor sales pitch than a customer reality (and cloud was hardly a top priority). But during our talk at ASUG TechEd, DSAG Chair Jens Hungershausen told me that across the countries he visits, from the US to DSAG's home turf in German-speaking countries, the agenda is pretty much the same: make sense of your business transformation amidst economic pressures - and figure out how SAP fits into it. If you simply do a technical S/4HANA upgrade, you haven't necessarily transformed anything. Cloud may be the right answer for some projects - but a knee jerk move to the cloud doesn't necessarily "transform" your business either.
That raises the question of SAP innovation parity between on-premise and cloud - and whether SAP customers must move to cloud - and RISE - for innovation, such as SAP "Business AI." (I'll get to that further in my upcoming DSAG interview piece. For a flavor of that, see my April 2023 DSAG interview).
During our 1:1, I asked ASUG CEO Geoff Scott how this all stacks up. After all, Scott's day one keynote with Mueller and Microsoft's Scott Guthrie focused heavily on generative AI. But Geoff Scott believes SAP tech leaders need to think about this differently:
I believe that what's coming at us as technology practitioners and professionals is a world that is cloud-based. A world where we are consuming software as a service, and a world where the more customized you are, the less likely it is going to be for you to take advantage of innovation, and be able to be on the forefront of change - versus being on the way, way back end of it. So that is a stark commentary - and for most organizations who are in SAP and by the way, I don't think this is an SAP related thing; I think this is across everything.
I think if you're sitting there looking at where you are today, what I just said is a staggering, frightening concept because you have years and years and years of technical debt.
Here's a curve ball: Scott doesn't buy into the thinking that all technical debt is bad.
We have this notion that all technical debt is bad. I'm not of that opinion. I think there's good technical debt, reasonable technical debt and bad technical debt. I don't think people go into anything saying, 'You know what? Today I'm going to create technical debt. I think people make decisions based on the best information they have at the time - and we don't know what the next 20 years are going to bring.
Scott believes generative AI has disrupted tech leaders - because it forces the issue of tech modernization. He arrived at this position via feedback from ASUG members:
If you'd asked me that question a year ago, if ChatGPT hadn't been around and generative AI hadn't been around, I probably wouldn't have given you this answer. But as I'm looking at this and watching this, I'm listening to what people are saying. The only way you get to any of this is you've got to have, cloud-based, software-as-a-service based, low customization. It doesn't work any other way.
Is there such a thing as being "ineligible for AI" - if you are mired in an overly complex, on-premise landscape? Scott thinks the answer might be yes. This opens up the question: how should customers integrate SAP's ambitious AI plans into their own roadmaps?
Part of this is clearly understanding SAP's AI strategy and roadmaps. But as Scott alludes to, another central theme is emerging: there is such a thing as AI readiness.
My take
As usual, I've run out of space before I've run out of ideas. Was ASUG Tech Connect a one-off thing? I don't think so. Despite a quick push to go-live, Tech Connect had a clear identity - and a palpable sense of community. There is room for even more informal, peer-led sessions at Tech Connect, but enterprise software vendors could learn from the Tech Connect playbook: concise one hour keynotes, then send you on your way, with bona fide time for networking (not lip service), and a choice between more conventional and more informal sessions (like the SAP Skillpocalypse).
Greenbaum and I talked to Mueller about an underrated TechEd issue: the need for reducing S/4HANA migration friction even further. Is this a matter of highlighting the partners making headway with innovative approaches, or does SAP need to go even further with its own tools? Mueller says it's both - and yes, there may be a role for AI here too, further automating the assessment of SAP landscapes pre and post-upgrade. And yes, we also talked with Mueller about public/private cloud innovation versus on-premise - a hot topic particularly with DSAG. I'll get back to that in my upcoming DSAG piece (for more context on S/4HANA upgrades, check ASUG Kris Turner's recent diginomica post, What are the biggest challenges to large scale ERP upgrades? Learning from SAP S/4HANA projects).
I was glad to see the presence of a BTP-only consulting partner at the show - the upstart BTP vendor sovanta (aka "the innovation factory for SAP BTP"). It was interesting to hear their take on how BTP has matured, what use cases are a good fit, and why BTP potentially stands out from other non-SAP options for app-building and integration (I was told for projects that leverage SAP data, or SAP business content, that BTP is often the best option, whereas customers are still using hyperscaler platforms for most customer-facing apps at this time).
This brings us to the future of SAP TechEd. For SAP to deliver for its customers' coding needs, SAP needs strong online education, along with hands-on skills training for both ABAP and new-to-SAP developers. It is my opinion that while ASUG Tech Connect should continue, SAP TechEd itself may need further re-invention. Perhaps there is one larger rotating TechEd, but logic tells me that you'll have much better luck reaching both new and 'classic' SAP developers by bringing coding/hackathon sessions to their local cities, SAP Code Jam style. Many of the developers SAP needs to reach are not flying to Vegas on SAP's behalf (TechEd Bangalore is different; thousands of developers live within driving distances).
I get the impression from Wessell that SAP is definitely looking hard at further ramping up its city-by-city code tours. Whatever the future of large scale TechEd, localized events are clearly a big piece of that puzzle. Add non-SAP and open source events into that mix also. If SAP relies too much on virtual training, it won't get there. Virtual events can also be more than they typically are, but that is a different blog post.
As usual, I've run out of word count before I've run out of topics, but I would be remiss not to mention a very interesting discussion Greenbaum and I had with Mueller about the potential for a different kind of RISE with SAP - one that brings in a strong integration between Signavio, the Lean IX acquisition, and Cloud ALM. Greenbaum has been championing the possibilities of this type of holistic process and landscape management tooling; we get into why in our ASUG Tech Connect review podcast. Part of the thinking here? Process debt is just as real as technical debt.
Though I have not heard anything from SAP directly on its intentions here - except some strong Signavio talk - if RISE went beyond hyperscaler management to include that type of integrated approach to process/landscape management, I don't think SAP would have to do the hardcore RISE pitch anymore. That type of product would sell itself, and make a much more persuasive argument for asking SAP to manage cloud hyperscaler footprints.
Whether that will materialize remains to be seen. Greenbaum thinks all the pieces are there, and he's looked at this closely; I've been burned before by imagining SAP will unite its disparate tools into something grand - and no, Solution Manager doesn't qualify, neither did NetWeaver - so I'll wait and see.
Only time will tell if SAP is making the right decisions on the future of TechEd. ASUG Tech Connect is an encouraging development, but now the needs of hands-on SAP developers need addressing. Wessell said that definitely happened at scale in India, at the largest SAP TechEd ever (4,000 hands-on tutorial sessions took place on the show floor terminals).
How will SAP address this across North America, and globally? I believe you can get a lot done with innovative online training - Wessell's team has taken SAP's online training to a significantly higher (and more accessible) level. SAP's Virtual TechEd also has a place - 30,000 registered this year - though there is still plenty of room for improvement in terms of imaginative online interactivity and Q/A. Either way, you need in-person training to bolster those online efforts; there is too much at stake for SAP and the ABAP developers who have more than paid their dues. It's too early to say how this part of the story will play out.