June 2024 Summaries

LaunchDarkly is a tool that makes it easier for developers to switch between different AI models based on application contexts such as user email domain, device, zip code, region, or custom context. In this tutorial, we built an example demonstrating how to use LaunchDarkly with Python and Flask to switch between AssemblyAI model tiers based on a user's context. The tutorial covered setting up the environment, coding a Flask app to serve transcriptions via an API, and adding feature flags to swap between model tiers.

LaunchDarkly introduces new AI flag templates for its Feature Preview interface to help users release, monitor, measure, and iterate on models and prompts for optimal performance. The AI model flags allow iteration on both models and configurations, while the AI prompt flags enable adjustments in prompt configurations. These features are available for Starter, Pro, and Enterprise plans and can be accessed through the Feature Preview interface.

This tutorial teaches how to set up, run, and analyze funnel experiments in LaunchDarkly. Funnel experiments calculate the best versions of a series of events or user flows and help identify where users drop off in a process. The tutorial covers creating metrics, setting up feature flags, building funnel experiments, analyzing data, drawing conclusions, and implementing changes based on experiment results. It emphasizes that funnel experiments are essential for understanding and optimizing the entire user journey, as opposed to single variable changes like A/B testing.