Case Study: Rox achieves 99% accuracy in outbound email evaluation with Snorkel AI

How Rox achieved 99% Accuracy with Snorkel Evaluate

Rox, an AI-powered sales productivity platform, needed a more reliable way to evaluate its outbound email personalization models. Off-the-shelf evaluation methods and its initial LLM-as-a-judge did not give the team enough confidence to ship critical customer-facing email features with the level of accuracy required.

Using Snorkel AI’s Snorkel Evaluate, Rox scored its judge against human experts, improved alignment, and uncovered issues in its outbound model, including incorrect recipient names. With Snorkel AI, Rox achieved 99%+ accuracy with specialized evaluators and improved the shipped critical outbound email feature by 24 points, enabling the team to deploy with greater trust.

Rox

Shriram Sridharan

Co-Founder