The Monthly Take

I had so many thoughts reading this a16z piece. I had to write this down.

The headline and framing sets you up for one thing (a serious, researched view on why SAP endures), but the body reads like a “directionally correct” thesis that skips the very nuance that makes SAP/ERP endure in the first place.

I understand that venture media has its own incentives. And yes , the essay is a narrative arc that conveniently slides into their portfolio-shaped opportunity. Still, for a brand synonymous with shaping the trajectory of cutting-edge technology, the lack of rigor is hard to miss.

Here’s where I agree and where I think the article under-delivers.

Let’s start with the part that’s absolutely right: SAP (and ERP more broadly) is not “going away.” Not because enterprises are nostalgic. Because enterprises are pragmatic. If you’ve lived inside a large companies like I have, you know this: ERP is not a set of screens. It’s the institutional memory of the enterprise.

It’s the canonical data model. It’s controls, permissions, approvals, segregation of duties. It’s posting logic, exception handling, audit trails. It’s the uncomfortable but necessary orchestration of end-to-end processes that keep the business compliant, reliable, and scalable.

And that’s exactly why the weekly Linkedin “SAP obituary” posts are so unserious. A new AI agent demo doesn’t dissolve decades of embedded process reality. It just changes how humans interact with it.

I also agree with the article’s core architectural direction: the “system of record” isn’t the main frontier - the “system of action” is.

In plain terms, People don’t wake up wanting “an ERP experience.” People want outcomes: onboard a supplier, approve a PO, post an invoice, close the books, ship product, handle a claim. And the friction is in the gap between intent (“what I need done”) and execution (“how the system needs it done safely”).

So yes, AI has a massive opportunity here: reduce cycle time, reduce training burden, reduce navigation challenges, and make enterprise work feel less like tab-hopping and more like outcome execution.

That direction is correct.

My issue isn’t the thesis. It’s the thinness. Because if you’re going to speak about SAP as the backbone of the world’s biggest enterprises, you can’t treat SAP like a static artifact. A few things stood out:

The screenshot used represents a SAP world many of us associate with the 2010s era.Yes, many organizations still run ECC. But it’s equally true that S/4HANA, Fiori, cloud transitions, and clean-core programs are not fringe topics anymore, they are the default modernization conversation for most SAP customers.

So when the article leans heavily on “look how painful SAP is,” it creates the wrong impression: that SAP is mostly a frozen legacy UI problem waiting for startups to rescue it. It misses the reality that SAP has been executing a very public evolution arc for 15+ years:

from ECC → S/4HANA, from GUI-first → role-based UX, from custom code → side-by-side extensibility, from monolithic upgrades → continuous delivery models in cloud.

You don’t have to love SAP to acknowledge this. You just have to do the homework.

The piece says (correctly) that these systems aren’t “complex” in the consumer-tech sense, they’re valuable because they encode how the enterprise works.

But then it stops short of the most important part: The “value” is not sitting passively in tables. It’s active, governed, contested. ERP value comes from:

- standardization decisions (what must be global vs local)

- process design choices (what is enforced vs flexible)

- governance (how you prevent customization sprawl)

- operating model maturity (who owns what, how decisions get made, how change is controlled)

This is why ERP transformations underdeliver, not because screens are ugly, but because organizations underestimate the human and governance complexity of process change.

So when the article frames the opportunity mainly as “AI will sit on top and make this programmable,” the question is: Programmable… under whose controls? With what approval logic? What happens when the agent hits an exception? Who owns the policy layer? How do you prevent a new “shadow process layer” from becoming tomorrow’s technical debt?

In other words: the hard part isn’t action. The hard part is trusted action at scale.

The piece implies that even after major upgrades, ERP often ends up being “read-only reporting” and hard to manipulate. Maybe that’s true in some environments. But it’s not a universal truth, and if you’re going to say it, it needs grounding, just like AI. In many enterprises, the whole point of S/4HANA and modernization is to enable:

- faster close and better real-time operational insight

- tighter integration across supply chain, finance, procurement, manufacturing

- cleaner data foundations and more scalable orchestration

- safer extensibility without breaking the core

If the claim is: “many programs fail to realize that value,” the analysis should go where it matters: why value leakage happens (customization bloat, weak data discipline, unclear process ownership, fragmented change governance). The system-of-record vs system-of-action demarcation isn’t new. What’s new is that AI makes the “action layer” dramatically more feasible, and more dangerous if it’s not governed.

As of now, ERP remains process-first: it protects integrity, compliance, and enterprise-scale orchestration. AI becomes task-first: it interprets intent, navigates complexity, and compresses cycle time. The winners will fuse the two with governance baked in.

In my view, AI is of today is best at intent, assistance, summarization, guidance, and task execution. ERP is best at end-to-end process integrity: controls, auditability, approvals, and scalable repeatability. That means:

- actions with previews + policy checks

- approvals and audit trails as default, not optional

- role-based access and least privilege baked into the agent layer

- exception handling treated as a first-class product problem (not a corner case)

The future isn’t SAP vs AI. It’s SAP & AI, properly blended. I am not sure if the long-term sustained opportunity lies in “SAP is ugly, therefore opportunity.” It isn't about rescuing users from screens. It’s making enterprise work trustworthy, faster, and safer, without breaking the core that keeps the business running.

From the Frontline