This post was written by Konstantin Lekh, Sasha Zinchuk, and Eugene Sergueev from Flo Health, and Liza (Elizaveta) Zinovyeva from AWS.

In this post, we share how Flo Health’s engineering team turned a proof of concept (PoC) from the AWS Generative AI Innovation Center into a production-grade, AI-powered medical content review and generation system built on Amazon Bedrock. This system reduced review time by 60 percent and tripled content throughput without expanding the medical team. We cover four areas: (1) adapting the PoC architecture for Flo Health’s content pipeline, (2) implementing specialized AI Judges for different review dimensions, (3) building an AI content generation system with Retrieval Augmented Generation (RAG), and (4) lessons learned from prompt engineering and production deployment. Part 1 of this series covers the initial proof of concept.

This post reflects the perspective of the Flo Health engineering team.

At Flo Health, we create diverse content to support millions of users on their health journey: from in-app stories and articles to onboarding flows and marketing materials. Every piece of content, whether textual or visual, must meet our rigorous medical accuracy standards outlined in our guidelines. While we’ve successfully integrated AI tools to accelerate content creation for editors and designers, medical review remained our critical bottleneck.