topic Re: Use TensorFlow models inside XSOAR automation in Cortex XSOAR Discussions

Use TensorFlow models inside XSOAR automation

Josep — Wed, 11 Jan 2023 17:04:11 GMT

Hello,

We'd like to create our own TensorFlow models to improve the system. The model will be trained and tested outside XSOAR, while the production model will be set inside an automation. The main problem here is whether XSOAR containers could have enough resources to make it work. The other option is to create a server to communicate via API with XSOAR. Obviously, the latter will be more time-consuming to deploy.

Thanks for your help

Re: Use TensorFlow models inside XSOAR automation

arnarayanan — Fri, 20 Jan 2023 14:02:54 GMT

It is difficult to comment without knowing what kind of model is this and how much time does it take to give you an answer;

There are some server configurations to adjust the memory that can be accessed by the container

https://docs-cortex.paloaltonetworks.com/r/Cortex-XSOAR/6.8/Cortex-XSOAR-Administrator-Guide/Configure-the-Memory-Limitation

So, if your model is a file in the image, then you can access it from the automation.