Enterprise application strategy is at a turning point. Paul Saunders, GVP Product Marketing at SAP, joins Julia Kloppenburg and Benedikt Gieger to explore what is actually changing in enterprise software and what will endure. From AI-first design principles to clean core strategy, from the three-gear model of transformation to a frank assessment of why best-of-breed architectures created more problems than they solved, Paul traces how the industry arrived at this moment and where it is heading. The conversation covers the gap between the vision of autonomous enterprise and the practical reality facing most customers, why trust in your underlying data is the prerequisite for any AI investment to pay off, and how the next generation of interfaces is already making today's UX look dated. A grounded, experience-rich conversation for anyone navigating enterprise technology decisions right now.

Enterprise applications are being redesigned from first principles, but the transformation is not moving at the same speed everywhere. In this episode, Paul Saunders, GVP Product Marketing at SAP, joins hosts Julia Kloppenburg and Benedikt Gieger for a wide-ranging conversation about what is genuinely changing in enterprise software and what will remain constant longer than most people expect.

Paul draws on 35 years in the industry, including time as a CIO and as an industry analyst before joining SAP, to explain why the shift from record-and-report systems to AI systems designed around sense, decide, and act is more profound than a technology upgrade. It requires rethinking process and organizational culture as much as the software itself.

The episode covers Paul's three-gear model for enterprise transformation: technology, process, and people. He argues that organizations trying to move all three at once typically stall, and that the most successful transformations sequence them deliberately. The banking analogy he uses throughout: a traditional bank installed ATMs and took tellers off the floor, reducing one form of friction while creating another. A mobile-first bank redesigned the experience entirely. Both happened, but at very different speeds and with very different results for customers.

On clean core strategy, Paul explains the underlying logic: separating what a business must standardize, what it can differentiate, and what it should treat as experimental innovation. He uses the BMW and Mercedes comparison to illustrate why two companies that look like direct competitors may need fundamentally different technology approaches based on where their competitive advantage actually lives.

The conversation also addresses the practical limits of AI adoption. Paul is direct about the trust problem: a CFO who walks into a board meeting and says "AI thinks revenue could be two and a half billion" will not stay employed long. Reliable AI outcomes require reliable underlying data, and most enterprises have not yet done that foundational work.

In the final section, Paul answers the question of what will look most outdated when viewed from 2030: best-of-breed architectures that gave each department the best tool for itself but left nobody accountable for how the pieces fit together or what the data actually meant across systems. What he expects we will be grateful for is simpler: an enduring focus on the customer's customer, and on technology as the stage crew that enables performance rather than the star of the show.

Topics covered in this episode:

AI-first enterprise design (sense, decide, act) · Clean core and pace layering · Three-gear transformation model · Trust and data quality as prerequisites for AI · Best-of-breed versus integrated platforms · Semantic data and the limits of API integration · Generational differences in UX expectations · The concert crew analogy for technology's role · What 2030 will reveal about today's enterprise architecture decisions

Most agentic AI pilots in manufacturing produce impressive proof-of-concept results and then quietly stall. The data is not clean, the processes are not standardised, or the organisation simply does not change how it works. So what does it actually take to get from a compelling demo to an agent running in production, handling real decisions inside real SAP processes? Shashikant Walimbe, VP and Global Head SAP Practice at LTM and an SAP partner deeply embedded in manufacturing operations, gives a practitioner's answer: start with one decision that costs you money every week, measure it, put an agent on it with clear limits, and see whether the numbers move inside a quarter.

In this episode, Shashi walks through what agentic AI genuinely means for a manufacturing operation (and what it does not mean), why the skepticism of plant managers and COOs is earned rather than ignorant, the five specific reasons agentic projects fail to reach production, how LTM designed trust into a live deployment at a semiconductor manufacturer, and his honest view on when SAP's AI capabilities win on merit versus when specialist vendors or hyperscaler platforms are the better answer. The conversation closes with a direct discussion of workforce impact: which roles shrink, which change without disappearing, and what responsibility a deploying partner carries in that transition.

Episode Chapters
00:00 – Introduction: SAP Autonomous Enterprise Podcast
00:30 – Guest introduction: Shashikant Walimbe, LTM
01:10 – What agentic AI actually means in manufacturing
02:24 – Why plant managers and COOs are right to be skeptical
03:03 – The honest answer to the COO: start small, measure fast
04:33 – Commercial tension: how LTM draws the line on advice
06:23 – Why most AI pilots do not reach production
07:02 – Five reasons agentic projects stall before scale
08:40 – Semiconductor manufacturer case: agents in the SAP backbone
10:14 – Designing trust into agentic systems
12:53 – SAP versus competitors: when does the SAP route genuinely win?
15:29 – Workforce impact: which roles shrink, which change
18:52 – Closing thought: start with one decision that matters

Listen and Subscribe
Subscribe on Spotify
Listen on your preferred platform

Why 80% accuracy is not enough in enterprise AI. SAP Latin America President Adriana Aroulho on complexity, talent, and partners.

Latin America is one of the fastest-growing technology markets in the world, and the autonomous enterprise is not a distant vision for the region. It is the next stage of business transformation. But making AI work at enterprise scale requires more than deploying a capable model. It requires business context: the actual processes, policies, rules, and operational reality of each company. In this episode, Adriana Aroulho, President of SAP for Latin America and the Caribbean, joins hosts Julia Kloppenburg and Benedikt Gieger to explain why the region's regulatory complexity, from Brazil's tax reform to cross-border trade agreements, makes it one of the world's most compelling proving grounds for business AI and intelligent agents.

Adriana maps the three-step adoption path she sees across the region — modernising the digital core, establishing trusted data and governance, then scaling to cross-functional AI — and makes the case that autonomous enterprise is not a large-enterprise-only play. She also addresses the talent gap, the role of regional partners in co-creating industry-specific agents, and the question every leader is asking: do you need to complete every maturity step before you can move? Her answer is clear: the foundations cannot be skipped, especially the data piece. But companies do not have to wait until everything is perfect to start generating value.

Episode Chapters
00:00 – Introduction: SAP Autonomous Enterprise Podcast
00:51 – Guest welcome: Adriana Aroulho, SAP Latin America
02:02 – Why now: converging trends driving LatAm transformation
05:02 – Where AI agents deliver the most immediate value
06:33 – Cloud adoption paths: from installed base to business AI
09:20 – Is autonomous enterprise universal across company sizes?
12:14 – The talent and skills gap: a global, not just regional, challenge
15:43 – Partner ecosystem: local complexity and co-creating agents
19:07 – Leapfrog versus sequential maturity: can steps be skipped?
20:46 – One thing every leader must get right: business context
24:20 – Closing and thanks

Listen and Subscribe
Subscribe on Spotify
Listen on your preferred platform