The Department of Energy (DOE) expects two internal artificial intelligence (AI) platforms to generate an estimated $74 million in annual operational savings, as the agency rapidly expands AI adoption across its workforce and mission operations.
Speaking on July 16 at GovExec’s Government Efficiency Summit in Washington, D.C., Bridget Carper Arnone, deputy chief information officer for architecture, engineering, technology, and innovation at DOE, said the department’s Joulix generative AI suite and Quanta data platform are transforming how employees work.
“Right now, we’re looking at $74 million in annual operational savings by leveraging the two tools between Joulix and Quanta, and that’s just two,” Carper Arnone said. “With our use case inventory, there’s a lot more.”
In 2024, the department introduced EnerGPT, a secure, department-specific chatbot. Today, Carper Arnone said that effort has expanded into Joulix, a broader generative AI platform serving 21,000 users across all 88 DOE department elements, including the department’s 17 national laboratories.
Carper Arnone said Joulix has grown beyond a single chatbot into a suite of AI applications tailored to specific business functions, including human resources, procurement, executive order compliance, and departmental policy review.
For example, she said DOE built a custom application that generates department-approved position descriptions, while another helps acquisition officials summarize lengthy contract proposals.
“What was taking us months, we were able to get that down to weeks, and to be able to award quicker because we were able to leverage AI,” she said of the procurement application.
She said the department has also continued to improve Joulix by adding web grounding in 2025, allowing users to retrieve internet information with roughly a 20-hour delay while keeping DOE data inside its secure environment.
However, as adoption increased, Carper Arnone explained that employees wanted AI to analyze far more information than Joulix could process at once.
That led the department to develop Quanta, a data platform built with Databricks that organizes structured and unstructured data before applying AI models.
One of Quanta’s earliest deployments came when DOE’s Office of Electricity needed to search roughly 1 billion documents under a tight executive order deadline.
“They needed to go through a billion documents, and they had three weeks,” Carper Arnone said. “So, we uploaded those into Databricks – 12 minutes, and it took the 200 gigabytes of data down to four.”
Since then, Quanta has expanded rapidly across the department.
Carper Arnone said adoption of the platform has grown by 4,900% in eight months. Quanta is now deployed in 44 offices, and it contains more than 500 data sources, with another 100 in the implementation backlog.
DOE aims to expand Quanta to all 88 department elements by the end of the fiscal year, according to Carper Arnone.
For Carper Arnone, the success of Joulix and Quanta ultimately isn’t measured only by hours saved or cost savings.
“Mainly for us, it’s more of the mission enablement,” she said. “If we do this right, how do we advance the mission? That’s another way that we look at it.”