How Southern Company Gas Uses Regulatory AI to Prepare for the Witness Stand
Preparing to testify in a rate case is among the most demanding tasks in utility regulatory work. A single proceeding can run more than a year, with filings arriving in successive rounds. By the time a witness takes the stand, they may need to reconstruct their thinking on questions they last addressed five years earlier.
John Cogburn has lived this experience many times. As Director of Rates, Tariffs, and Regulatory Planning at Southern Company Gas, he has prepared testimony, defended positions on cross-examination, and managed company filings across multiple jurisdictions. At LET'S GO! 2025, HData's annual customer conference, Cogburn sat down with HData Enterprise Customer Success Manager Selwyn Yeh to talk about how he uses HData's Regulatory AI during one of the most important phases of a rate case: the final days and hours before he takes the stand.
The Long Arc of a Rate Case
A rate case rarely feels linear when energy professionals are working on it. After a company files, the proceeding goes quiet, but only briefly. Then come data requests from intervenors and commission staff, testimony filed by intervenors and staff, rebuttal testimony from the company, and, finally, the evidentiary hearings convened by the commission. By the time a witness prepares for cross-examination, months have passed since they last sat with their own original arguments.
“You have a clean theory of the case on day one,” Cogburn said, “and it's very easy to lose that over the course of a case.”
The challenge is not only remembering what a witness said. It is anticipating the questions others will ask, understanding how opposing parties have characterized the company’s positions, and revisiting issues that may not have surfaced in years. And always present is what Cogburn called the “Perry Mason moment”: the late-hearing surprise where someone pulls a piece of evidence the witness forgot to review.
How Cogburn Uses Regulatory AI in Practice
In the days leading up to a hearing, Cogburn turns to HData for a handful of specific tasks, finding that AI tools can dramatically streamline witness preparation.
Returning to his own words. Before one hearing, Cogburn asked Regulatory AI to summarize how he had characterized a key allocation change in a previous Virginia rate case. The result surfaced the central theme, gradualism, more clearly than he remembered articulating it himself.
“I knew all that, but I kind of said it better in the last case,” he said.
Rather than scrolling through transcripts to reconstruct his own reasoning, he could anchor himself in his prior testimony in minutes and walk into the hearing grounded in his prior reasoning and aligned with how the company has historically framed its positions.
Tracking how intervenors frame his arguments. Intervenors often reference Cogburn’s positions in passing in a sentence or footnote. Those mentions are easy to miss and easy to regret missing. Regulatory AI surfaces those references and points him directly to the source.
“It gets me to the key points,” he said, “and takes me right to it in their testimony.”
Because Regulatory AI provides detailed, downloadable citations for every response, Cogburn can verify each reference at the source without rebuilding a binder index.
Refreshing context on long-dormant issues. During one proceeding, an intervenor raised renewable gas questions rooted in a case from five years earlier. Cogburn anticipated being pressed on the same ground during cross-examination. In the past, he would have spent hours digging through filings to rebuild his familiarity with the record. With Regulatory AI, he pulled the relevant testimony, reviewed the highlights, and reconstructed his mental model “in minutes instead of hours.”
Aligning with policy witnesses. Most rate cases pair technical witnesses with a policy witness whose testimony establishes the broader narrative but may contain references that need technical backup on cross-examination. Cogburn uses Regulatory AI to identify those moments in advance.
“It helps you understand what they may be opening themselves up to,” he said, “and what they need to punt back to you.”
That coordination tightens the company’s overall case and prevents inconsistencies that intervenors are quick to exploit.
A Broader Shift in Witness Preparation
As Regulatory AI becomes more deeply integrated into how parties prepare for proceedings, the bar for preparation will rise across the table. Witnesses on every side will be expected to have read, anticipated, and surfaced more because their counterparties will have.
That shift is already visible in how Cogburn’s team approaches the final stretch of a case. Rate cases will continue to be long, complex, and high-pressure. But for witnesses willing to use Regulatory AI as an accelerator and not a replacement, they no longer need to be a test of endurance.
Cogburn’s full session and other videos at LET'S GO! 2025 are available for download. To learn more about how HData's Regulatory AI supports witnesses, researchers and regulatory teams, visit hdata.com.
About HData
As the AI-native operating system for energy regulation, HData serves the largest customer ecosystem in regulated energy, helping utilities, regulators, advocates, advisory firms, corporates, and energy technology companies navigate regulatory complexity. Through centralized data, domain AI, and purpose-built applications, HData accelerates the research, analysis, and workflows critical to how the future of energy is decided.
