Pusher Channels is a hosted WebSocket relay, and it's a clean way to move messages between clients. Air Pipe covers the same realtime fundamentals — WebSocket pub/sub, presence, history-rewind and retained messages — and adds what a relay can't: the same connection runs your pipeline on each message, and the auth that gates each channel is part of the same config, not a separate endpoint you host.
Pub/sub with wildcard channels, presence, message history-rewind and retained values live in the WebSocket channel API — and MQTT is available from the same config for devices. You get the realtime feature set, plus a second transport a WS-only relay doesn't offer.
A relay moves bytes between clients and stops there. Air Pipe can run your config on each message before it fans out — assert on it, transform it, write it to Postgres or MongoDB, call an upstream API — and gate each channel with your own pipeline logic against the user's JWT claims. The realtime layer and the backend are the same thing.
With Pusher, private and presence channels are authorized by an auth endpoint you stand up and host yourself. In Air Pipe the auth, the channel and the backend are one config: your end-users connect with their own token — a shared secret, your JWKS, or an Air Pipe-issued token — or fully open, with no org key in the browser. There's no separate auth service to run and keep in sync.
With a relay you still bring your own backend, your own auth service and the glue between them. Air Pipe is transport, per-user auth, the compute that runs on each message, and billing in one product — metered in the same request units as everything else you run, which roll over for 12 months.
Run realtime on our managed cloud, or self-host the same single binary on your own infrastructure — nothing-shared either way, with no external relay or Redis to operate. Self-hosted realtime connections aren't billed.
Out of the box, Air Pipe doesn't advertise the edge footprint or published SLAs of a dedicated global realtime network. But extreme scale is a conversation, not a wall: if you need very large connection counts or dedicated capacity, we scale out or provision dedicated infrastructure for you — talk to us. For most products the better fit is realtime that's part of your backend, runs your logic on each message, and self-hosts on infrastructure you own — and that's there from day one.