Taking too long to build your LLM prompts? Struggling with internal tooling to manage prompt development? You need to read this.
Founded in 2008, Left Field Labs is a strategic technology partner to the world’s largest brands. They’ve worked with leading companies like Amazon, Google, Meta and Starbucks to build high quality digital experiences. Seeing the emerging potential of Generative AI, the team at Left Field Labs began developing their capabilities early and have already built several production use cases for their customers.
In the course of development of various projects they quickly recognized the need for better tooling. Everyone on the engineering team was doing their own prompt engineering, collaboration on Notion/Excel lacked version control, and there was no consistent way to compare prompts across model providers or test them robustly. When making changes to prompts, regression testing was an extensive process. A better solution was needed.
When Eric Lee, Partner & CTO at Left Field Labs saw Vellum’s HackerNews post in the early days of Vellum’s product evolution (5 weeks after first launch), he knew he wanted to give it a try.
About 10-15% of Left Field Labs employees use Vellum due to the productivity gains they experience. As an innovation partner helping to bring AI solutions to businesses and their users, Left Field Labs relies heavily on rapid prototyping to prove out technical feasibility and create UX and UI demos to test with users. Since partnering with Vellum, Left Field Labs has created dozens of such examples, allowing stakeholders to understand the value and potential impact of AI-augmented features and workflows before moving them into production. The quote below directly reflects the productivity boost Left Field Labs has seen while building these prototypes:
With Vellum, Left Field Labs has been able to optimize solutions across quality of response, speed of response, and cost, because prompts can be easily compared across model providers with ways to quantitatively test model quality.
Since becoming a customer, Left Field Labs has maintained close collaboration with Vellum. As they push forward the application of generative AI for their clients, Left Field Labs has discussed various architecture approaches and workflows with the Vellum team, often leading to the identification of feature requests, which Vellum deems worthy of investment, in order to enhance the value of their platform for Left Field Labs as well as other customers.
Left Field Labs initially focused on the Playground feature for prompt engineering in their usage of Vellum's services. With the introduction of Jinja templating by Vellum, the Left Field Labs team was able to create more complex functionality in their prompts. Vellum’s new Workflows feature has also been widely adopted by the Left Field Labs team, allowing for quicker orchestration of chained prompts and functions that previously required more custom engineering.
Left Field Labs values the partnership with Vellum, and the dedication of Vellum’s product and engineering teams to provide tools tooling & best practices so that both companies can work at the leading edge of generative AI applications. At Vellum, we love working with the Left Field Labs team because they push our product to be better on a regular basis.
Vellum has onboarded more than 70 paying customers who’ve improved their internal AI development processes. Sign up here to start exploring the platform for yourself. You will also schedule an onboarding call with one of Vellum’s founders or founding engineers who can provide tailored advice for your use case. We’re excited to see what you and your team builds with Vellum next!