Case Study: Lavender cuts latency by 50% with Vellum

How Lavender cut latency by half for 90K monthly requests in production

Lavender, an email intelligence company, needed more effective tools for prototyping and managing their LLM prompts in production. They wanted to engage their entire team in prompt iteration, a process that was previously restricted to developers within their codebase. They were also looking for a platform to help with testing real-world scenarios and ensuring model availability, which led them to Vellum.

By implementing Vellum, Lavender was able to manage over 20 LLM features, including their Email Coach and Personalization Assistant. The solution enabled collaborative prompt engineering, easy capture of live data for optimization, and integration with an Azure client. This partnership resulted in Lavender cutting their average latency by 50% for their 90,000 monthly production requests, providing a significant performance improvement for their users.

Lavender

Sasha Boginsky

Full Stack Engineer