The following Utility Voice was authored by Marc Rosson, Community Connector at Utility.Community.
A series built on hard questions can leave a wrong impression: that the vendor and the customer sit on opposite sides of the table. For the last four articles, we have pressed on gaps—whether anyone had a better governance answer than SAP’s, who secures an enterprise where agents do the work, whether the data those agents read is trustworthy, and whether the customer they see is real, current, and consented. This month we wanted to do something rarer in our industry’s discourse: document the agreement.
Because when you lay three very different documents side by side—SAP’s own words, a DSAG utilities working-group survey from the German-speaking user community, and the influence lists our own community has been assembling—something striking emerges. We are all describing the same enterprise, in the same order, with the same warnings. The disagreements are real, but they are mostly commercial. The agreements are structural. And agreement, properly noticed, is leverage.
Three Voices, One Table
The three sources deserve naming. First, SAP’s official voice: Torsten Welte, SAP’s SVP and General Manager for Energy & Natural Resources, in his post-Sapphire interview with ASUG, alongside what SAP now publishes in the open on its Architecture Center—including the AI Golden Path, SAP’s stated starting point for building AI applications, maintained in public on GitHub. Second, the German-speaking user voice: a DSAG utilities working-group survey cataloging roughly two dozen member priorities, from webinar cadence to political pragmatism to demographic change. Third, our own: the “what we still need to influence” lists this series has carried from its first installment. Read cold, the overlap is not partial. It is the majority of every list.
Agreement One: Foundation Before Autonomy
Ask Welte what a utility CIO should do over the next 24 months and he does not answer with agents. He answers with a clean core and infrastructure that scales, data that is in order, and change management for an organization about to absorb AI. The DSAG survey says it in one German-engineered sentence: AI does not replace the cleanup of processes. And this series has spent months arguing that governance, security, trustworthy data, and a correct customer record come before autonomy, not after.
Even SAP’s own developer guidance now enforces the sequence. The AI Golden Path is built around a design-led lifecycle that begins with explore and discover—identify the business problem, define success metrics and evaluations—before anything is designed, delivered, or scaled. When vendor enthusiasm, German pragmatism, and North American regulatory caution all produce the same sentence independently, we can stop debating the sentence and start budgeting for it.
Agreement Two: Data Is the First Deliverable
Welte was blunt: if data is siloed or inconsistent, AI will hallucinate and behave unpredictably—and he reached for Peter Hinssen’s line that there is more fiction written in Excel than in Word. The Golden Path’s technology stack makes the same point structurally: the data layer sits beneath the AI layer, which sits beneath the agents. There is no agentic architecture in SAP’s published guidance that does not stand on governed data.
DSAG arrives at the identical conclusion from an unexpected direction: its members want numbers, data, and facts as the precondition for constructive dialogue with politics—objectivity as a data discipline. And this series made the argument that the data foundation is the foundation under the foundation. Three voices, no daylight. The only open questions are sequencing and funding.
Agreement Three: Publish, Don’t Whisper
The plainest item in the DSAG survey is also the most telling: information twice a year is no longer sufficient given the pace of change. SAP appears to have heard it. The Autonomous Enterprise arrived with 13 two hours webinars across multi-weeks. The Architecture Center now publishes reference architectures—the security operations pattern, the agentic AI blueprint, and the AI Golden Path—in public, versioned, on GitHub, under an open license, where any utility architect can read, cite, and even propose changes. That is a different posture: architecture as public artifact rather than account-team whisper. Welte described the engagement change the same way—forward-deployed engineering, capabilities co-developed with customers instead of features waiting six to twelve months in a back office.
Honesty requires noting where this agreement runs out. DSAG wants a roadmap horizon beyond ten years, because a CS-maintenance-scale decision cannot be made against a two-year view. Our community wants pricing commitments that outlive promotional windows. The direction—transparency—is agreed. The distance is not yet.
Agreement Four: Knowledge Is the Asset at Risk
Ask each voice what actually threatens the autonomous enterprise and a surprising answer recurs: not the technology, but the departure of the people who understand the current one. Welte calls it the silver tsunami and points out that much of that knowledge is already latent in system configurations, waiting to be harvested—documented and undocumented alike—then adds, to his credit: as a community, we probably need to do more. DSAG names demographic change directly and proposes the countermeasures: a searchable knowledge database of one-page white papers, and a forum for unresolved issues that assembles cross-company teams around them. Our community has been building on the same instinct—peer-authored, searchable, shared.
Put those together and the convergence sharpens into something actionable: institutional knowledge is the scarce input to the autonomous enterprise. A Company Memory is only as good as the knowledge we manage to capture into it. The tribal knowledge walking out our doors was, it turns out, the training data.
Agreement Five: The User Group Is the Interface
The DSAG survey is a document about making a working group matter: converting passive members to active ones, building communities for business departments and not just IT experts, networking across associations. Welte, for his part, described an SAP that develops capabilities with customers and does its community work through enterprise architects and user groups. And every ask this series has made routes through ASUG. All three voices share an assumption worth making explicit: the channel for shaping this platform is collective, not bilateral. No single utility moves the roadmap. User groups—amplified, aligned, and armed with facts—already have: DSAG’s formal demands on the API policy put contractual definitions and transition timelines on SAP’s agenda within weeks.
What DSAG and Our Community Share: A Direction Forward
Which brings me to the relationship this article is really about. The German-speaking utilities and the North American ones are running the same playbook, independently, without coordinating. They want a knowledge database of short, searchable white papers; so do we. They are building communities for billing, receivables, and reporting professionals—the business side, not just IT; that mirrors exactly how our community is structured. They are converting passive members into active contributors; that is our standing challenge too. They assemble numbers, data, and facts for their dialogue with BNetzA, BDEW, and the ministries; we do the same for our PUCs, FERC, and NERC. Different acronyms. Identical discipline.
And DSAG’s sharpest structural argument deserves an audience on this side of the Atlantic: SAP’s development capacity is finite, and country-specific fragmentation wastes it, because the physics and the market requirements are largely the same across borders. If that logic justifies harmonization within Europe—and it does—it does not stop at any country border. The measured next step is not a merger of communities; it is coordination between them: aligned requirement lists, exchanged white papers, the occasional joint session, and a shared influence agenda wherever our asks already overlap. Reading their survey against our lists, they overlap almost everywhere. We should start treating those documents as drafts of one document.
The Honest Caveats
Three cautions belong in the open.
First, agreement on direction is not agreement on terms. Joule for ECC customers requires a signed RISE contract. The API policy’s grace periods and migration paths remain under negotiation. Industry AI is moving to a commercialization model “centered on value” that has not been defined. Vision consensus can coexist with sharp contractual disagreement, and the warmth of the former must not be allowed to soften the latter.
Second, consensus does not ship software. A meaningful share of the agreed foundation is still roadmap—the Agent Gateway’s general availability, the LeanIX AI Agent Hub, agents in the SuccessFactors organizational structure. Our community’s discipline of distinguishing announced from delivered from adopted applies to happy convergence exactly as it applies to controversy.
Third, notice what none of the three voices resolves: the operational technology boundary. Neither the Welte interview, nor the Golden Path, nor—fairly, given its scope—the DSAG survey answers where agent authority stops relative to SCADA, ADMS, and OMS. An agreement list is drawn from what everyone talks about. The gap list is drawn from what no one does, and that silence should worry us more than any disagreement.
What We Still Need to Influence
- A joint user-group agenda: merge the DSAG survey themes and our community’s influence lists into a single set of asks, delivered to SAP through DSAG and ASUG simultaneously. The same request in two voices is twice as hard to defer.
- Obligation-to-implementation transparency: DSAG’s ask for legal requirements documented in the Regulatory Change Manager and our ask for NERC CIP mapping are the same product feature—a published trace from regulatory obligation to SAP implementation. We should ask once, jointly.
- Horizon commitments: a utilities roadmap view beyond ten years for decisions of CS-maintenance scale, and post-2026 pricing commitments for the agentic stack. Commitments, not promotions, remains the standard.
- A utilities lane in the AI Golden Path: the Golden Path is deliberately industry-generic. We need the utility-specific walk-through—IS-U data products, meter-to-cash agent patterns, and the OT boundary—contributed and maintained in the open, the way the rest of the Architecture Center already is.
- Community knowledge infrastructure: support—and restraint—from SAP so that peer-authored knowledge bases can flourish alongside Company Memory rather than being absorbed into a proprietary layer. The knowledge we capture together should remain ours.
A Call to Our Community
Across this series I have asked four questions: whether you had a better governance answer, whether you could tell your regulator who secures your agents, whether the data they read is trustworthy, and whether the customer they see is real. Those were the same question in different clothes. The fifth is different in kind: if the vendor, the German-speaking user community, and the North American utility community all agree that the foundation comes first—clean core, governed data, captured knowledge, an engaged community—then what, exactly, is your utility waiting for? The convergence documented here is the business-case memo your leadership keeps asking for, written by three parties from different perspectives with consensus.
And one invitation with a deadline. Find your DSAG counterpart—the working groups are a connection away, and the mail exchange between association speakers has already begun. Bring one one-page white paper on something your team has actually solved; ask for one in return. If enough of us do that before we gather in San Antonio this October, the transatlantic agenda will not need to be proposed. It will already exist.