Claude Opus 4 + RAG: Build an AI App with Embeddings & Knowledge Graphs | MVP Unplugged

Michael Washington walks through a real-world build of AI Story Builders, an open-source (MIT) AI writing app that combines structured story data, embeddings-based retrieval, and a knowledge-graph layer to improve consistency and reasoning.

Overview

In this MVP Unplugged episode, host Justin Garrett interviews Microsoft MVP Michael Washington about the design and implementation approach behind AI Story Builders.

What the episode covers

The problem: why models struggle to “just write the next chapter”

Structuring story data for better context

Michael describes breaking a story into explicit components such as:

This structure is then used as the basis for retrieval and for asking targeted questions about the story.

Embeddings + cosine similarity for retrieval

RAG as the core pattern

Knowledge graphs and tool calls

Demo and distribution details mentioned

Resources mentioned