It's been a few months since you last heard from me, and here's why. I've been focused on solving a significant challenge in GenAI projects: managing the uncertainty of outputs from large language models (LLMs). Check out this week's video or our Github project for an overview.

Unless you're tackling something trivially simple, your GenAI project will get stuck. You'll struggle with performance metrics, error prevention, degradation detection, and project timelines. Teams find themselves in perpetual experimentation because traditional software engineering methods can't handle the unpredictability of these models.

Over the past three months, I've developed a new methodology, Performance-Driven Development (PDD) to address these challenges. It's not hard--just a different way of working--you can get started in about 15 minutes with generated data and spreadsheets. If you're curious, I have a RAG solution example on the PDD GitHub project.

For a quick overview, check out this week's video - enjoy!