Bring the prompt you'd never paste into ChatGPT.

Most organisations with sensitive data have the same problem: the best AI lives on someone else's servers. Law firms, hospitals, banks, funds and government teams are now buying their own GPUs and running open-weight models in-house, so client and patient data never leaves the building.

Air-Gap Night is a live demo of what that looks like in practice. A private AI runs on a server in the room. Halfway through, we pull the internet cable out. It keeps working.

The evening

  • 6:30pm: doors and drinks

  • 7:00pm: the demo. Fine-tuned open-weight models on local GPUs, drafting, searching and answering over private documents. Then we unplug it.

  • 7:30pm: your prompt, offline. Write the prompt you'd never send to a cloud AI on a card. We run a few, anonymously, on the air-gapped box.

  • 7:50pm: the honest numbers. What on-prem really costs (hardware, power, models, people) versus API bills, and when it isn't worth it.

  • 8:15pm: drinks and networking with London's on-prem and open-model crowd

Who it's for

Anyone responsible for sensitive data and AI: CTOs, heads of AI and data, CIOs and CISOs, innovation and legal teams, researchers, founders and the engineers who'll actually run it. Hosted by Encode Club at Encode Hub, with a live demo from the Saint PrivateAI team.

Spaces are limited and approved by the team. Free to attend.