This repository is dedicated to exploring how artificial intelligence can help transform regulatory policy into machine-readable code. Our starting point is the new NEPA regulations, provided as an XML file (https://www.federalregister.gov/documents/full_text/xml/2025/07/03/2025-12326.xml).
We're experimenting with using AI to analyze and translate complex policy documents into formats like JSON or YAML—making them actionable for software and automation. The analysis and code generation will happen in Databricks, and all results will be stored here for easy access and collaboration across the Department.
Because we have limited familiarity with both Databricks and policy-as-code generation (which is itself a fairly new and evolving concept) we'll also be leveraging AI to bridge gaps in our experience. We'll be sharing and learning as we go, and documenting our journey for others to follow.
- Check out the NEPA regulations XML file in this repo.
- Open the notebook (
NEPA Policy As Code.ipynb) to see or contribute to the analysis and code generation. - Please log a GitHub issue to share your findings, generated code, or notes to the repository so others can learn from your work. You can also follow along in the AI Code Gen Early Adopters Teams channel
Anyone in the Department interested in policy automation, AI, or learning about Databricks. No prior expertise required—just curiosity and a willingness to experiment.
This is an open, collaborative experiment. Mistakes and learning moments are expected and welcome. We'll be documenting what didn't work, as well as what does. When sharing, please do the same: document your process and share insights so we can all benefit.
This project is for internal experimentation and learning purposes only.