Skip to main content
Question

NetBrain integration with AI API Proxy

  • June 19, 2026
  • 1 reply
  • 12 views
A
Forum|alt.badge.img

Hi Support Team,

As most of our customers is running in mission critical environment, sometime with air-gapped network, so using public AI LLM like OpenAI or ChatGPT etc. will be a challenge. I understand that NetBrain can also integrate with local deployed AI inferencing servers, however an enterprise grade inferencing server also costs very high.

One possible option is to use AI API proxy servers like LiteLLM Proxy or Apache APISIX or Kong AI Gateway etc., the dual home proxy server could be located in restricted DMZ zone which has limited internet connection to Open AI platforms, while NetBrain can point the AI proxy to the API server.

I am wondering if NetBrain has any similar deployment cases in your customer’s environment. It will be appreciated if you could provide some references or ideas for this type of deployment.

 

Best Regards,

Andong

 

 

      1 reply

      S
      Forum|alt.badge.img
      • Sahilkatira
      • NetBrain Employee
      • June 30, 2026

      Hi Andong,

      Thanks for laying this out so clearly — you've obviously thought the architecture through.

      The design you've described is sound and supported. NetBrain connects to the LLM through any OpenAI-compatible API endpoint, so pointing NetBrain at a proxy or gateway in the DMZ, with that proxy acting as the sole controlled egress to the provider, works as you'd expect. That holds for OpenAI-compatible gateways generally, including the LiteLLM, Apache APISIX, and Kong AI Gateway options you mentioned.

      One expectation worth setting: this fits a restricted/DMZ environment where controlled outbound access to the LLM provider is permitted. For a fully air-gapped network with no external connectivity, you'd still need a locally hosted inference model, since the proxy pattern still depends on an egress path to the provider.

      When you're ready to move toward production, or want to work through the specifics of your environment, I'd recommend opening a case through our standard support process so the team can go deep on the configuration with you.

      Happy to share more detail if it would help.

      Thanks
      Sahil

      Follow NetBrain on LinkedIn

      Terms & ConditionsAccessibility statement