Executive Summary
OpenAI has partnered with the U.S. Department of Energy's Pacific Northwest National Laboratory (PNNL) to explore using AI to accelerate the federal permitting process for infrastructure. Through PNNL's PermitAI™ initiative, they co-developed a benchmark, "DraftNEPABench," to test how AI coding agents perform on complex document drafting tasks required by the National Environmental Policy Act (NEPA). Initial findings show that these agents can reduce drafting time by up to 15%, signaling a significant potential for AI to streamline critical government workflows.
Key Takeaways
* Partnership: OpenAI is collaborating with the Department of Energy's Pacific Northwest National Laboratory (PNNL) and its PermitAI™ team.
* New Benchmark: The collaboration produced "DraftNEPABench," a new benchmark designed to assess AI model performance on drafting environmental impact statements and other NEPA-related documents.
* Technology Used: The study utilized "generalized coding agents" (specifically Codex CLI powered by a GPT-5 class model) that can read and synthesize hundreds of pages of technical content, verify facts, and draft structured reports.
* Performance Gains: The benchmark showed that AI agents have the potential to speed up NEPA document drafting by as much as 1 to 5 hours per subsection, equating to roughly a 15% reduction in drafting time.
* Stated Goal: The long-term objective is to reduce the approval time for federally reviewed infrastructure projects from months to weeks, thereby accelerating project development and strengthening U.S. competitiveness.
Strategic Importance
This partnership demonstrates a strategic move by OpenAI to apply its advanced AI to complex, high-value public sector workflows, positioning its technology as a tool for government modernization. Successfully accelerating bureaucratic processes like federal permitting could unlock a significant new market and showcase the tangible economic benefits of AI beyond consumer applications.