A Meta study shows that pairing two AI coding agents to review each other’s patches improves bug detection more than giving a single agent a bigger budget. Meta’s RADAR tool reviewed over 535,000 code changes, with over 331,000 merged into the codebase, cutting median review time by 35%. Production incidents dropped to one-fiftieth of those from manual reviews. Meta’s Engineering Agent fixed test failures, with 80% of generated fixes undergoing human review and approximately 25.5% accepted into production. The Just-in-Time testing framework produced a fourfold increase in bug detection across over 22,000 generated tests, reaching up to twentyfold for significant failures.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
