Local AI Transcription – Traffic Flow

[morefon]

Hi everyone,

we are currently testing Local AI Transcription together with 3CX and would like to better understand the network traffic flow in different deployment scenarios.

Current test setup
• 3CX hosted in a private cloud
• Local AI transcription server running on Google Cloud / AWS
• Both systems have different public WAN IP addresses

For testing purposes, the transcription traffic currently goes over the public WAN, which is fine for us at this stage.

Target production setup

For a future productized offering, our goal would be to:
• host the Local AI transcription servers in the same data center as our own 3CX systems
• ideally avoid unnecessary WAN exposure
• clearly explain the data flow to customers and data protection officers

Questions
1. Private networking / VPN
• Can Local AI transcription be used over a site-to-site VPN between different private IP subnets?
• Will all audio and transcription data remain inside the VPN tunnel in that case?
2. Same data center / same LAN
• Is it supported to run 3CX and the Local AI transcription server in the same data center or LAN, communicating only via private IPs?
• In this scenario, does any transcription-related traffic leave the local network at all?

The main objective is to ensure a clean, well-documented data flow that can be easily explained to customers, auditors, and DPOs.

Thanks in advance for clarification

Best regards

 

[nikolascx]

Hello Dear @observer_at2

Yes you can run 3CX + Local AI Transcription either:
over a site-to-site VPN (private subnets), or
in the same DC / same LAN using private IPs only

…and in both cases, the audio + transcription traffic stays on the private path (VPN tunnel or LAN).
Nothing needs to traverse the public WAN as long as you point 3CX to the transcription server using its private address and the routing/firewall rules allow it.