Trained for you, not on you.

For fourteen years, Lama worked alongside authors. Helping them publish. Helping them sell. Helping them build the platforms that connect them to their readers. Through every shift in the industry, the relationship between an author and their book stayed the same. The author wrote it. The book carried their voice.

Then in 2023, The Atlantic exposed something that changed everything.

A dataset called Books3 had been used to train the largest AI models in the world. It contained 183,000 books. Pirated. Scraped. Used without permission. Meta. Bloomberg. Other tech giants. All of them training their AI systems on books authors had spent careers writing, without ever asking.

Authors searched their own names and found their entire catalogues inside the dataset. John Grisham. Jodi Picoult. Margaret Atwood. Stephen King. Authors at every level of the industry discovered their life's work had been used to build the very systems now competing with them.

"The future promised by AI is written with stolen words."

Lawsuits followed. The Authors Guild and dozens of bestselling writers filed class actions against OpenAI, Meta, and others. But the damage was already done. The conversations Lama had been having with authors for years started to carry a new note. Fear, but also frustration. Authors did not want to refuse AI. They wanted AI that respected what they had built.

XanaMind was the first vault, built for the people we were already serving.

On 22 January 2026, ABC News broke a story that captured the same problem in a different form.

A small Australian tour operator had used AI to generate a travel article. The article promised visitors a "peaceful escape" at Weldborough Hot Springs, ranked among the "7 Best Hot Springs Tasmania Experiences for 2026." Tourists began arriving in droves at the remote Tasmanian village, searching for the mineral pools the article had described.

Weldborough has no hot springs. It never has. The article was an AI hallucination. Confident. Detailed. Completely fabricated.

The business that published it lost its reputation overnight. The owner described the damage as "soul-destroying." His admission to ABC News was direct.

"Our AI has messed up completely."

The pattern was the same as the one we had been watching in publishing. AI generating with confidence what it did not actually know. Operating from generic training data instead of verified knowledge. Damaging the very people who had trusted it.