Catholic Startup Develops AI Chatbot Trained on Church Teachings

Longbeard founder Matthew Harvey Sanders created Magisterium AI to answer questions drawn from Roman Catholic doctrines and teachings. The system uses existing Gemini and GPT models trained on more than 32,000 documents spanning 2,000 years. Sanders, 44, said the chatbot was initially intended as a research tool for converts, students, scholars and priests.

Within three weeks of launch, users began submitting nuanced moral and personal questions, prompting development of the new Ephrem model.

Longbeard has raised $3 million in pre-seed and seed funding and is seeking an additional $4 million to $5 million in a Series A round at a $25 million valuation. com and K-TV, invested $500,000 in March 2025. The company reports Magisterium has reached users in 190 countries. Revenue comes from a subscription pro tier and API access, with Hallow listed as a customer.

XIV issued a social encyclical this week that addressed AI's effects on human relationships, institutions and power. The letter criticized AI applications in warfare, political manipulation and image distortion. Anthropic, valued at $965 billion, invited 15 Christian leaders to its San Francisco office in March to discuss chatbot responses to moral and spiritual questions.

Cofounder Chris Olah later spoke in Rome after the encyclical presentation.

Michael Baggot, a Longbeard board member and professor at the Pontifical University of St. Thomas Aquinas, described Magisterium as functioning like a digital librarian. He said Ephrem will direct users to priests when questions exceed its context. Sister Nancy Usselmann, director of Pauline Media Studies in Los Angeles, attended Longbeard's Builders Artificial Intelligence Forum in Rome in November.