Kevin Surace walks through GENIE's deployment options: public cloud or on-premises behind your firewall. It runs on H100 servers, encrypts all data end-to-end, and doesn't learn across sessions or share data with the vendor. That matters for banking, healthcare, and insurance.
Questions this video answers
An H100 server (one or multiple H100s).
"100% encrypted" and "100% secure"; vendor does not see or share the data and the system does not learn from it beyond the application session.
Chapters
Talk a little bit about where it lives. GENIE lives, I'm assuming, in the cloud somewhere. Genie lives in the cloud, but can also live behind your firewall. You do need a special GPU server and anything from an. If I was running hypersensitive secure application test cases, that still all works. I just. Absolutely. Private cloud, you can put in your private cloud, you can put it behind your firewall. You need a, you know, a specific type of a 100 or multiple H100 server. These are widely available today. They are out there. They are not inexpensive because they're meant to do AI, but they are very powerful. And depending on the way we've written this, it's all private. Those models reside on that server, what we call the GENIE server. And there's an API that automatically gets called from AIQ that sends the information there and brings it back all secure, 100% encrypted, 100% secure. We don't see it, it's not shared, and the system doesn't learn from it, except for your application. Right. So anything that it learns is only for use during that session with your application under test, and then after that it's just all go, no need to use it. Okay, so like, the digital twin is only used for that purpose. It's not like it gets sent to us, it gets sent anywhere else. Everything's private. And this is pretty unique that you can now deploy this directly beyond your firewall. But our banking, customers, financial services, healthcare, pharma insurance, that's all behind their firewall or in their private cloud. Like, that's not inarguable. That that's a given. And so you have to do that. They have to use private models, they have to show that they're private models. We fully support that.
