Snapshot · This Week in Tech (Audio)
The Vatican built its own private AI model entirely from scratch, trained exclusively on several hundred years of Church documents, with no outside data and no internet access.
Where this was said
At 1:32:03 · chapter starts 1:27:10
The episode's most extraordinary segment unfolds as Fr. Robert Ballecer describes the Vatican's private AI project. Built completely from scratch — not fine-tuned from any public model — it was trained exclusively on several hundred years of digitized Catholic Church documents. Its primary use cases are document retrieval and synthesis for sensitive canonical cases that cannot legally be transmitted over the internet, and real-time simultaneous translation across 19 to 20 languages during Vatican meetings. It speaks Latin because that was one of the very first capabilities the team trained — necessary to translate older source documents. The model is specifically designed to say 'I cannot answer that' when confidence drops below an acceptable threshold, eliminating hallucinations by design rather than guardrails [1] — Fr. Robert Ballecer "The Vatican developed a completely private AI model trained from scratch on several hundred years of Church documents. It speaks Latin, han…" 1:30:20 . The hardware is an Acer Veritron running NVIDIA's Blackwell GV10 chip with 128GB of memory and 4TB of storage. Spanish regional accents are proving hardest — the language itself is easy, but distinguishing Castilian from Colombian from Mexican from Bolivian remains the system's biggest challenge. The team would love a Latin podcast.
The Vatican developed a completely private AI model trained from scratch on several hundred years of Church documents. It speaks Latin, handles 20-language real-time translation, and is used for sensitive canonical cases that can never legally touch the internet. Spanish regional accents are proving the hardest to crack.
The Vatican's private AI model performs real-time simultaneous translation across 19–20 languages during internal meetings, with Spanish dialects proving the most challenging.
Bhanu and his team built approximately 50 free tools to attract search traffic, each linked back to SiteGPT.
With AI coding tools like Cursor, Bhanu can now create a new free marketing tool in less than 5 minutes by referencing existing tools.
Bhanu filters Ahrefs keyword results to show only those with a keyword difficulty below 10, making them realistic ranking targets for any decent website.
Bhanu sets a minimum search volume of 1,000 monthly searches when selecting keywords to target with free tools.
PropGPT averaged 20 downloads per day right after launching on the App Store through influencer marketing.
Eyal and Yali shut down all marketing and spent 4 months completely rebuilding PropGPT from scratch.
PropGPT has accumulated over 40,000 total downloads since launch.
PropGPT's large language model (AI) operating costs are just $20 per month, and the cost is continually falling.
Ad-based monetization works well for game apps where users spend extended time in-session, as seen with Grid and Wordle.
We use essential and analytics cookies to run Vuci. To understand how the site is used: Privacy Policy.
Install Vuci on your phone
Add it to your home screen for a faster, app-like experience.
Install Vuci on your phone
Tap the Share button, then “Add to Home Screen”.
A new version is available
Reload to get the latest Vuci.