infersurge is built on a simple promise: the gateway sees metadata, not content. This page explains exactly what we collect, what we deliberately never keep, and why.
Effective date: October 19, 2026
infersurge is a remote device gateway. It routes tool calls from your AI client — ChatGPT, Claude, VS Code, or any MCP-compatible harness — to machines you own, over an encrypted WebSocket maintained by a small agent running on each device.
To run accounts, meter usage, and keep the service secure, we collect the minimum: your account email, usage counts (how many tool calls succeeded), and a short audit trail of events. That is the entire list.
The content of your work — files, terminals, screens — flows end-to-end between your AI client and your devices. The gateway relays it in memory and stores none of it. See Terms of Service for the contractual side.
The gateway is a zero-content-storage system. Here is the complete, honest inventory:
metadata only
web1)zero content storage
wss://). No inbound ports
are opened on your machines: devices dial out to the gateway.
api.infersurge.com and app.infersurge.com.We do not sell or rent personal data. We run no advertising networks and no third-party analytics scripts on this site. And because we never store your content, your data is not used for model training — there is nothing to train on.
For any privacy question, data request, or deletion request, contact us at mail@infersurge.com. We aim to respond within two business days.