For more than a year, analyst Josh Greenbaum and I have been trying to advance the business case conversation. For SAP tech leaders in particular, this topic has urgency.

The latest example? This was the first interview topic aired by UKISUG Chair Conor Riordan, prior to UKISUG Connect: Hashing out AI, data governance, and the SAP migration business case with Conor Riordan. It also surfaced in my no-holds-barred appearance on Alexander Greb’s podcast (Jon Reed on the Real Work Behind Transformation).

At our ASUG Tech Connect sessions each November, Greenbaum and I heard this loud and clear. We heard it from enterprise architects - and we heard it from SAP project leads. They needed help; they needed to bridge that notorious gap with line of business leaders. **But why now? **

Why SAP business case urgency, and why now? Migrations and AI raise the stakes

With ERP migration decisions on the one hand, and high stakes AI moves to evaluate on the other, these issues take on new urgency. The office of the CIO is playing a different game now. Security, performance and operational support are a given, but if you’re not directly supporting the business, you’re not where you need to be. Heck you may want to be out in front of the business at times.

At a moment’s notice, you must be ready for a persuasive explanation of how IT investments are helping you navigate volatile supply chains, and deliver for fickle/inflation-sensitive customers. "Business case" has become the code word for tying tech to results. Yes, the tools and methodologies to build business cases have improved, yet Greenbaum and I felt something was missing. Bring on the adventures in workshop design.

ASUG’s events team supported our efforts, including an interactive session at the 2025 ASUG Annual Conference in Orlando. But while the customer discussion was vivid, attendees made it clear: they didn’t have enough to take away.

How could they bring this know-how to bear on their projects? We weren’t equipping them with enough resources. And: this wasn’t just about formal business cases.** Attendees wanted compelling business narratives, not lift-and-shift techno-jargon. **

ASUG Tech Connect 2025 - a business case rethink, including an LLM-based "coach"

Fortuitous encounters and cross-pollinations led us to a revised format. At this year’s ASUG Tech Connect, we had industry experts from SAP’s value advisory team on-site. We also had a customized, LLM-based tool to stir the pot, built by SAP Partner Basis Technologies. (The tool is now called ValueCoach, and it’s a capability within the Basis Technologies Klario "change intelligence" solution for SAP change management).

This workshop brought the unexpected. The biggest surprise? The role of the LLM in sparking (and supporting) our interactions. But first: here is a draft of how we envisioned the Tech Connect workshop.

Getting Business Stakeholders on Board with Your Cloud Transformation: A Hands-on, AI-assisted Business Case Workshop

Join a first-ever hands-on workshop that will help you define a business case for your SAP cloud transformations with S/4HANA, Business Suite and more. Our goal: give you insights and content to engage the specific business stakeholders you need to make your project a success. We’ll be breaking into small groups to workshop specific transformation scenarios – such as supply chain, finance, retail, and HR – assisted by a new partner-built AI tool.

Workshop post-mortem - LLM lessons from the retail group

Our retail working group brought disparate interests, without a classic retail customer. Retail in this context is really about connecting your end customer with the rest of your logistics/operational systems. The prompt that broke open the conversation was simple and jugular:

Our customers are frustrated with our e-commerce and in-store availability. They are not accurate to inventory.

The LLM output was effective and relevant:

• A framework and template/text for a business case around these scenarios, which could be adapted by the attendees after the event.
• Identifying the stakeholders that could be implicated in the scenarios we discussed, with coaching materials on how to best approach them.

But while the LLM output was useful, the biggest value was sparking/informing our group discussion. The output forced the issue.

One participant questioned whether they could actually take this content to their CFO. Why? Because it would reinforce a reactionary approach - one where they are already asked to justify their results. Our table’s SAP value advisor was able to reframe that. We spoke about the importance of developing pro-active business cases from the IT side, and shifting from a reactionary position.

This breakthrough happened because:

• The LLM output provoked the customer (in a good way), with a real-life example of how this could work.
• We had an expert advisor at the table who could place the output in a context that made the most sense to the customer (reinforcing my view: if you want to make the most of gen AI workflows, domain experts are even more important than critical thinkers).

LLM scenario from the utilities industry: "how should we present this to senior leadership?"

Another customer in our group had a different angle. As a utilities customer, the supply chain scenario didn’t quite work for them. But we were able to pivot the LLM session without rebooting it. We asked ValueCoach for advice: how should we approach this problem in our utilities division? The prompt:

I have another question: in our energy division, our customers are expecting more visibility into rate structure in billing, and the impact year over year of a rate case. How should we present this to senior leadership?

This led to the right output for our discussion. During the final wrap, this customer said they’d like to do a session that started with a general prompt about customer experience as a core strategy, such as:

Our energy division is dealing with more customer scrutiny over issues like rate structures in billing. How can we improve our customer experience?

At one point, the LLM business case "coach" asked us to go talk to the CFO about our inventory visibility issues. We came back with this response:

We had a good conversation with the CFO, but the CFO expects real-time inventory visibility. However, our inventory systems are not integrated with our ERP system in real time. How do we manage that expectation?

The LLM coach responded:

Thanks for sharing the CFO’s expectation for real-time inventory visibility. Given your current constraint (inventory systems not integrated with ERP in real time), let’s manage that expectation by translating “real-time” into business terms: decision latency requirements, financial impact of delays, and acceptable thresholds. We’ll quantify what “real-time” needs to be to protect revenue and margin, rather than promise a technical outcome prematurely.

The customer was struck by this example. Instead of just rolling along with our "real-time" buzzword, the LLM coach challenged us to go behind the catch phrase - and break out the components of "real-time" in measurable, business terms.

Identifying stakeholders - another key LLM benefit

Another key workshop benefit? Helping participants identify the stakeholders to engage with - both in terms of specific job titles/roles, but also casting a wider net, citing stakeholders that may not have occurred to the participants. In a report draft Greenbaum sent me, he wrote:

In the workshop example we’ve been discussing so far, ValueCoach first suggested the sales lead as one of the key business stakeholders with which to interact. Subsequent interactions also yielded the operations manager as an important stakeholder. But a query specifically asking which other stakeholders to connect with yielded the following additions:

• Procurement/sourcing lead
• Customer service/order management
• Plant manager/production planning
• Inventory control/warehouse
• Logistics/transportation
• CFO/finance director
• Legal/risk management
• IT applications owner (for inventory systems inefficiencies)

The initial response to this query also yielded a set of talking points for each of these stakeholders. ValueCoach also excelled at drawing out the importance of personnel issues and, by extension, the importance of having human resources stakeholders in the conversation as well.

Workshop LLMs surprise a critic

I’m often wearing the hat of an enterprise AI critic - that’s because I believe** the hype machine obscures the precision customers need for AI results**. This type of example shows why those who dismiss LLMs as word prediction machines are selling this probabilistic technology short.

While this is still a probabilistic system at its core, this level of semantic nuance shows why enterprises are not giving up on this technology, despite many false starts. Use case design is core to getting a better LLM result.

We were determined to have a business case discussion, not an AI tech session. The LLM was presented as a discussion starter and content generator (notice I did not say "creator" - those two are not the same). The resulting output is by no means a finished business case, but a template and head start. Nor will it create the stakeholder collaboration and buy-in necessary. But giving IT leaders a newfound confidence to engage in those discussions is no small thing.

Basis Technologies did a solid job of making sure ValueCoach was open enough to take the prompt in different directions, but still within a set of constraints (this required some tool testing and feedback sessions, to get the tool ‘loose’ enough to pivot, but not loose enough to lose focus). Tying this type of tool directly into a customer’s data and systems is an obvious temptation, but that’s beyond the scope of a group workshop format like this.

One fascinating aspect of the use of LLM coaches? The LLM output pays no heed to organizational politics. If an organization is open to new perspectives and data, this LLM output could compel organizations to reconsider their priorities/data/assumptions.

This is a bit ironic given how notoriously accommodating and encouraging LLMs can be to individual users, to the point of legal exposure. But if organizations use these tools in the right way, they could potentially help disrupt the politics of business as usual - and depersonalize provocative ideas for consideration by business leaders.

My take

Another advantage of the business case topic? LLMs are also trained on plenty of this type of proposal material. So, the context engineering for this use case had a big head start. Nor did it require any customer-specific context to be useful for this type of workshop, which would have been an obstacle.

Did this change my attitude towards LLMs? No - I’ve always felt this type of brainstorm/draft proposal scenario is an ideal fit for these technologies. I was expecting the tool to spark good discussions. I was hopeful ValueCoach would generate useful output for attendees to take with them, or refine later (it did).

But I wasn’t expecting the LLM interactions to embolden attendees to take business discussions forward in their organizations, and speak to line of business leaders more confidently. That opened my eyes, and made me eager to conduct further sessions. Watch this space - and in the meantime, Greenbaum should have his take on this workshop out soon; I’ll link to it from here.

End note: this workshop had a lot of moving parts and a fast design timeline. SAP, ASUG and Basis Technologies had to go along with a couple of mad scientists, and you never pull something like this off without a team of unsung masters of logistical detail behind these scenes as well. Thanks to all those folks...