If you want to understand inequality in American education, look at decodable books. These are the foundational texts early readers rely on to connect phonics to meaning. They are essential and very expensive: the cost of high-quality sets could reach $3000. What’s more, too often, the stories weren’t culturally relevant or instructionally aligned.

Katie Pace Miles, Executive Director of The Reading Institute, didn’t need data to know the problem. Her own son, a kindergartner in Brooklyn public schools, hadn’t brought home a single decodable book.

So Katie did something radical. She built a GPT model to generate decodable books, engineered to match her phonics scope and sequence. Each story would be 90 percent decodable, with characters and plots that reflect the students reading them.

Her GPT worked and she is now building a collection of 200 books and a distribution portal so families and educators everywhere can access them for free.

This isn’t just about literacy, it is about infrastructure. Katie is building a public good using private tools and showing what’s possible when pedagogy meets technology with purpose.