Use Case: Building a Secure and Controllable AI Solution

Our Focus

Any company that wants to leverage the latest in AI capabilities for its own internal process and business development. Protecting the source data is a must why a public service like ChatGPT is out of the question. The lack of fully secure protection of source data in any public service is another hurdle to overcome.

Background Brief

The latest developments in AI technology have brought completely new possibilities to analyzing and learning from a multitude of scattered source data that may be in any type of format. The large language models behind e.g. OpenAI and ChatGPT have given users completely new dimensions to work with. The main restriction for those is that any data provided to them automatically gets stored in their learning materials. That makes it impossible to use them for sensitive or private data.


in Build, Control and Operations

The main question is – how to start using AI for data that has to be protected at all time. An AI model is just as good (or bad) as the training data going into it and the model works better the more data it is fed with. There are good open-source based tool especially in the GenAI field that work adequately with limited amounts of training data but setting them up in such a way that performance, security and data protection are all on good level is a very difficult thing to do. The algorithms are very resource heavy why a cloud type of environment is preferred. Making sure that data and handling is secure in a public cloud is very hard to do as well. Quite a lot of good AI utilization cases are left untapped as the challenges and restrictions are unsurmountable.

With a Cloudshift21 Cloud CapsuleTM

The Cloudshift21 platform aka the Cloud CapsuleTM makes it possible to protect all aspects i.e. user access and authorization, data handling and protection, controlled location and data spaces and secure communications out of the box. It was built with all of this in mind for the highly sensitive mobile data handling in telco business and has been extensively analyzed and audited to cater to those needs. It was also built for real big data making it possible to run also large training models in a reasonable set of resources. And as it shares the underlying hardware infrastructure across all tasks and users it makes the best use of critical resources like e.g. GPU resources in an AI training scenario. Finally, it is infrastructure agnostic i.e. it can run on anything from big clouds to on-premise data centers and even to EDGE. And the same application runs without modification on all of those. In an AI scenario it would be possible to train and analyze results on cloud and then deploy the needed subset of the application on EDGE as just one example.
A possible scenario for some of the challenges listed above could be that the platform is deployed in own premises on a set of servers adequate for the training material needs. User accounts are created according to the different types or groups of users where e.g. the people feeding the training data has access to that and users of the model can only access the results to cater for their needs. As each user account can have several workspaces created with each of them providing the same level of protection as between user accounts there can be even more fine-grained types of access and authorization.
In short – Cloudshift21 can provide a kind of fully protected local cloud environment that doesn’t require extensive skills in cloud setup nor data handling. It is all part of the fully integrated solution.

Ready to see for yourself?

Sign up to stay

Register for the Capsulized CloudTM open beta

Register for the Capsulized CloudTM
open beta

Beta users get 2 weeks free PLUS early access pricing that's up to 67% off our already radically affordable rates. Leave your info below to secure your spot!

Download an exclusive white paper analyzing the issues within the cloud industry & learn about the technological innovations behind the

Capsulized CloudTM

Capsulized CloudTM