What actually happens when you press send
Most "send a big file" services are storage lockers. Your file is uploaded to a server, parked there for a few days, and the recipient downloads a copy. That model is why they have size caps, expiry dates, paid tiers and privacy policies — storing and moving other people's data costs real money, so it has to be rationed.
This tool does something different. It opens a direct connection between the two browsers using WebRTC, the same technology behind video calls, and pushes the file down that connection as a stream of chunks. There is no copy on a server because there is no server in the path at all.
What our side does provide is introduction. Two browsers cannot find each other on the open internet unaided — each needs to learn the other's network address and encryption parameters. So the sender writes a small description of itself into a temporary pairing room, and the receiver, armed with the six-character code, reads it and writes back its own. That exchange is a few kilobytes of connection metadata. The moment the two are talking directly, the room is deleted.
Why the code expires and the tab has to stay open
Because the connection is live, both devices have to be present at the same time. This is the one real trade-off of the peer-to-peer approach, and it is worth being plain about: you cannot queue a file up on Monday for someone to collect on Wednesday. There is nowhere for it to wait.
So expiry here means something narrower than it does on an upload service. It is the pairing code that expires, after ten minutes, not a stored file. The code also stops working the instant somebody uses it, which closes the window in which a guessed code could be redeemed. Six characters drawn from a 32-symbol alphabet gives just over a billion combinations, and a given code is alive for only a few minutes.
Keeping the sending tab open is the other requirement. Closing it tears down the connection mid-stream, which is why the page warns you before you navigate away.
When a direct connection is not possible
Two devices behind ordinary home routers can almost always reach each other, using a technique called hole punching: both make outbound connections that their routers then allow replies through. A helper server known as STUN tells each device what its address looks like from outside, which is all the information the technique needs.
Some networks defeat this. Symmetric NAT, which assigns a different external port for every destination, makes the address unpredictable. Strict corporate and university firewalls sometimes block the traffic outright, as do a few mobile carriers. In those cases the usual industry fix is a TURN relay — a server that both devices connect to, which forwards the data between them.
We deliberately do not do that, because a relay would carry every byte of your file through somebody else's machine, which is exactly the thing this tool exists to avoid. A failed connection here is an honest failure rather than a quiet upload. The reliable workaround is to put both devices on the same Wi-Fi network, where a direct connection nearly always succeeds. If you only need a small file moved and the network refuses to cooperate, encoding it another way — as a Base64 string, or into a QR code — is sometimes the quicker route.