Guomai Culture
Delivered 400 AI reading-companion agents for Guomai Culture, built on a knowledge graph covering 10,000 books.
Client
Guomai Culture is a book publishing and cultural media company, and a client of the AI business transformation line.
The problem
A publisher’s most valuable asset is the copyrighted content itself, but once a book is sold the connection between publisher and reader breaks. Readers who cannot finish a book, cannot follow it, or have nobody to discuss it with represent real demand with no product to absorb it. Making content conversational is the most direct entry point for AI in publishing; the difficulty is scale. A few hundred books can be handled by hand, several thousand require an engineered knowledge base.
What we did
Built a knowledge graph covering 10,000 books as a single content substrate, then delivered 400 AI reading-companion agents on top of it, each tied to specific titles or themes and responsible for guided reading, questions and further reading.
| AI reading-companion agents | 400 |
|---|---|
| Knowledge graph coverage | 10,000 books |
Limits of this case
- Those numbers describe delivery scale, not usage effect. This page does not treat one as the other.
- Delivery at this scale assumes the client owns a large body of structured content with clear rights. Companies without that substrate cannot copy the approach.
- A knowledge graph is a heavy one-off investment that pays back through reuse. A requirement for a few dozen agents does not justify building one first.
- Agent count is not business value. The real test is reader-side usage data.