O’Reilly Media: Finding Answers in the Content Haystack

Situation

O’Reilly is the go-to learning platform for technology professionals — thousands of expert-authored books, videos, and courses on everything from software development to AI to security. But the way people were using it had shifted. Nearly half of all usage was nonlinear. People weren’t sitting down to read a book. They had a problem at work and needed an answer now. The content was there — buried across thousands of titles. The way to get to it efficiently wasn’t.

Insight

AI could change the access model entirely: from browsing to asking.

Retrieval-Augmented Generation made it possible to turn O’Reilly’s full library into an on-demand, multi-source research tool.

One question, one answer drawn from the best available material, with citations back to the originals. But getting the technology to work was only half the challenge. The harder part was designing a real human-AI collaboration. A chatbot on top of a content library isn’t useful for long. A system that responds with grounded, cited answers and then opens doors — to source material, follow-up questions, related topics, something the user hadn’t thought to ask — that’s a different kind of product.

Transformation

Phillip led the product delivery of O’Reilly Answers. Users ask free-form questions and get accurate, cited summaries drawn entirely from O’Reilly’s curated library — no open web, no unvetted sources. The interface was built around choice and exploration: every response gives users multiple next moves instead of a dead end. Answers started embedded within books, courses, and videos, proved its value, and now sits prominently as a standalone tool on the platform’s main page. Content creators are compensated when their work contributes to a response — one of the first AI tools to do that.

What This Means for Your AI

O’Reilly had a massive content asset that wasn’t reaching its full potential. Adding AI didn’t just improve search — it created an entirely new way to access the library. Static content became dynamic, on-demand, multi-source answers.

That’s the difference between bolting AI onto a product and rethinking the product around what AI makes possible. The business upside is real: more value from existing content, faster time-to-answer for users, and a platform that keeps pace with how professionals actually work today.

If your organization has deep content or expertise that isn’t being fully leveraged, this is what it looks like to unlock it.