« All posts

Coral: An LLM-Native Harness for Production Recommender Systems

Coral is an LLM-native harness that enables continuous optimization of recommender systems.

Coral is an LLM-native harness that enables continuous optimization of recommender systems. Traditional methods allow engineers to test changes slowly and reactively, while Coral automates this process through a closed-loop agent. A/B tests on two large social platforms demonstrate that this system increases user engagement while reducing costs.

This synthesis was produced from its source by AI; there is no human editor or manual review step. How we work