A San Francisco-based startup, Theorem, has secured $6 million in seed funding to create automated tools that verify the correctness of AI-generated software. The company aims to address the challenge of ensuring the reliability of code produced by AI coding assistants from companies like GitHub, Amazon, and Google, which now generate billions of lines of code annually. Theorem’s approach combines formal verification with AI models to automatically check proofs, significantly speeding up the verification process from years to weeks or even days.
The core technology behind Theorem not only catches bugs that traditional testing might miss but also helps in scaling software oversight efficiently. For example, the company was able to turn a 1,500-page specification into 16,000 lines of trusted code for a customer looking to enhance performance without introducing errors.
As AI continues to shape software development, the need for rigorous verification tools becomes increasingly critical, especially in sectors like AI research labs, electronic design automation, and GPU-accelerated computing. Theorem’s focus on scalability and correctness guarantees sets it apart in the market, aiming to provide a viable solution for ensuring the reliability of AI systems embedded in critical infrastructure.
Source: VentureBeat