Laurel leverages large language models to automatically generate the assertions needed for program verification, reducing the burden on engineers.

• Solves a key bottleneck in formal verification by automating assertion generation
• Uses innovative domain-specific prompting techniques to improve LLM performance
• Demonstrates how AI can enhance existing verification tools like Dafny
• Represents a practical application of LLMs to complex software engineering tasks

This research matters because it makes formal verification more accessible and efficient, potentially increasing adoption of verification tools in industry and improving software security at scale.

Laurel: Unblocking Automated Verification with Large Language Models