Why Build It Twice? How Two Utilities Are Sharing AI to Tackle Rate Cases
When El Paso Electric started looking for an AI tool to help with regulatory work, it expected to evaluate vendors and pick one. Instead, it found another utility that had already built what it needed and was willing to share it.
That raised a question the whole industry should be asking: if utilities share the same problems, why is each of us paying to solve them alone?
The problem every utility knows
A rate case means months of coordinated testimony across many witnesses and hundreds of data requests, each answered consistently with past filings. Staff research precedent across 50 states and federal agencies, then track every commitment in the final order. It all happens on tight deadlines with real financial stakes.
When AI arrived, regulatory work was promoted as apotential use case, and the hype outran reality. Technology alone can’t solve these problems. It takes subject matter experts, a utility’s own data, and tools built around how regulatory teams actually work.
What OG&E built
OG&E files rate cases in multiple jurisdictions, which requires internal staff to work efficiently to control timelines and cost. Our regulatory team’s frustrations fell into three areas:
- Consistency: aligning testimony with company strategy, other witnesses, prior filings, and past orders
- Research: finding relevant precedents quickly across every state and federal agency
- Responsiveness: responding to the shorter times and growing volume of data requests
- Tracking: managing schedules, data requests, and compliance with commission orders
With Diff Consulting, we built an AI solution on an open-source, containerized stack deployed in our Azure tenant and portable to on-premises infrastructure. We designed it as “plug and play”: reusable functions that can be assembled into different solutions. We did this to save development effort. We didn’t yet realize the same design would make the tool shareable across companies.
End users shaped the tool through short agile cycles. That kept development grounded in real work and built adoption from the start. In fewer than three months, we had a working capability to draft witness testimony. We named it Advocate.
What we’ve seen so far:
- Increase in quality of testimony written resulting in less rework and corrections
- Increase in volume of testimony that is aligned to overall objectives and not redundant
- Increase in confidence of consistency with previous regulatory filings
- Reduce redundancy and external review costs
What El Paso Electric was looking for
At El Paso Electric, we spent significant time with our regulatory, legal, and business stakeholders before looking at any technology. We identified 7 use cases, comprising 29 mandatory requirements, in 3 categories: regulatory research, calendaring, and testimony and RFI drafting. We wanted a phased approach that delivered value early and a human in the loop at every step. We also set a target ROI based on conservative savings estimates.
Then we saw Advocate. Our stakeholders responded right away: we need something like that. Advocate already met 12 of our 29 mandatory requirements. Another 7 will be delivered within six months of implementation, which brings us to about two-thirds. The remaining 10 are now part of an expansion plan we are building with OG&E and Diff Consulting and will be shared between the two companies.
We had been shopping for a product. What we found was a partner.
The hard part: making sharing work
Agreeing to collaborate was easy. Making it work across two regulated companies took some effort.
The agreement. OG&E owns Advocate outright and licenses it to EPE to run in EPE’s own environment for internal business use. EPE contracts separately with Diff Consulting for implementation and support. Diff Consulting develops enhancements; OG&E owns them. Because the structure is a standard software license, other utilities could join on similar terms without negotiating shared ownership.
The technology. Each utility runs its own instance in its own environment. We share the application: code, configuration, and improvements. We never share data. Because Advocate is modular and containerized, adapting it for EPE is mostly a matter of configuration rather than a rebuild.
The governance. OG&E owns and maintains the roadmap, funds development, and sets priorities with input from EPE. Because OG&E works in agile cycles, enhancements are released regularly to both utilities.
The obstacles. We expected legal review to be the hardest part. It wasn’t. Because Advocate runs in each utility’s own environment and no data is shared between us, our legal team found nothing unusual and reviewed it like any other software license.
Keeping humans accountable
Can commissions and stakeholders trust AI-assisted testimony? The witness always owns the testimony. Advocate drafts, researches, and flags inconsistencies. People review, verify, and sign. To assist Advocate users, Advocate has several safeguards such as citation linking, document grounding, multiple question and testimony reviews and human in the loop approval. The tool makes experts faster. It doesn’t replace their judgment.
A model for the industry
Utilities rarely collaborate this way, even though most of us don’t compete with each other. The nuclear industry learned long ago, through organizations like INPO, that sharing lessons makes everyone stronger. The rest of the sector can do the same with technology.
Our model:
- Each utility keeps full ownership of its data.
- Each utility controls its own security and governance.
- AI applications are shared and improved together.
- Every additional utility lowers the cost of development for the group, saving them money.
- New capabilities benefit every participating utility.
Every utility that joins brings new use cases, expertise, and ideas, so the tools improve for everyone as costs fall.
Join Us
We’re looking for utilities that want to help shape what comes next. You can contribute use cases, subject matter expertise, or pilots of new features. In return, you get a proven AI application built for utilities, by utilities. It runs in your own environment under your own governance, your data stays yours, and every improvement is shared with all participants.
If your regulatory team faces the same pressures ours do, we’d like to hear from you. Contact Kevin Lagge, Director – Technology and Innovation (laggekj@oge.com) at OG&E or Nadia Powell, Director – Analytics & Data Strategy (nadia.powell@epelectric.com) at El Paso Electric.
There’s no need for each of us to build the same tool separately.
About the Authors
Kevin Lagge – Director Technology & Innovation at Oklahoma Gas & Electric
Kevin Lagge is an executive management professional with 30+ years of industry advancing leadership. Significant experience in organizational and business development in executive management, strategy, information systems, operations, and production management using problems solving, organization development, performance improvement, and innovation. Experience spans the energy, manufacturing, electronic, automotive, and plastics industries.
Nadia Powell – Director Analytics & Data Strategy at El Paso Electric
Nadia Powell serves as Director Analytics & Data Strategy at El Paso Electric, where she leads the enterprise analytics roadmap, data governance, data management, business intelligence, analytics enablement, and advanced analytics. She brings together business priorities, technology, and people to strengthen how the utility uses data to make decisions, improve performance, and serve its customers.
Powell has helped expand analytics from a reporting function into a strategic enterprise capability. Under her leadership, her team has delivered solutions that automate energy efficiency document processing, improve program and spending visibility, support regulatory reporting, and apply predictive analytics to supplier diversity. Her team also supports executive and Board of Directors dashboards, connecting enterprise performance measures with information leaders can act on.
She is advancing the next generation of El Paso Electric’s data and AI capabilities through cloud data platform planning, modern data architecture, and the development of trusted, reusable data assets. Her work spans an enterprise information governance framework, data ownership and stewardship, data literacy, and a data platform initiative. She has also helped establish the company’s AI Working Group and AI policy, while advancing practical AI applications across business functions, including Legal and Regulatory. Central to her approach is building the organizational foundation—skills, accountability, and collaboration—that makes technology investments deliver lasting value.
Since joining El Paso Electric in 2009, Powell has held leadership roles including Senior Regulatory Case Manager and Director of Palo Verde Generating Station Management. This experience gives her a distinctive perspective across utility regulation, generation, enterprise strategy, and technology.
Previously, Powell served as Consumer Affairs Officer for the City of El Paso and Regulatory Project Manager with Southern California Edison. She holds master’s degrees in Economics with a Regulatory Concentration and Business Administration from New Mexico State University, and a bachelor’s degree in Industrial Technology from Colorado State University. Originally from Brussels, Belgium, she brings a global perspective and a commitment to continuous learning to her leadership in the energy sector.
Diego Dora – Founder & CEO at Diff Consulting
Diego Dora is the Founder & CEO of DIFF Consulting and a leading voice on generative and agentic AI for the enterprise. He has spent 30 years in software development, more than 20 of them at the point where enterprise systems meet emerging technology. Today he is one of the few leaders who has turned generative AI into working software that runs day-to-day operations for utilities and enterprises across many industries.
Diego and the DIFF team have built more than 75 generative AI applications, now live in production, across multiple business units, including AI-powered tools for SAP ERP migration. Along the way, they developed a set of solutions and a methodology for Agentic Software Development and Continuous Improvement.
Diego is a recognized expert in AI agents. He and DIFF have worked many years for Google as experts in agentic AI for the enterprise, building solutions on the intersection between Google’s LLMs, Google Cloud Platform and Enterprise Systems (like SAP, Salesforce, Oracle).
Before founding DIFF, Diego was Managing Director and Member of the Management Team at sovanta AG, a German SAP innovation firm, where he helped set the company’s strategy for more than a decade. He built from scratch sovanta’s development lab in Argentina and led sovanta’s SAP Business Technology Platform practice and served hundreds of European customers, among them Adidas, Audi, BASF, Bayer, Porsche, Volkswagen, Lidl, Red Bull, Zurich, Swisscom, SAP SE and German utilities like EWE.
Earlier, at Accenture, he was Global Development Lead for the firm’s internal SAP program, leading about 30 developers across three countries, and was named a Top Performer four times. He also spent five years in academia, conducting open source research and serving as an assistant professor.
SAP has named him an SAP Mentor, one of fewer than 50 worldwide that SAP’s product teams and executives turn to for direct advice. He is a frequent speaker at enterprise software conferences, including SAP Sapphire and SAP TechEd, recently spoke on generative AI at ASUG SAP4U 2026, and is a regular guest on technology podcasts.



