How it works

• You pick files, or a whole folder. They stay on your disk.
• You get one link — or a QR code — and send it however you like.
• They open it, and the two browsers find each other and connect.
• The files stream across that connection in small pieces, one after another.
• Both devices check every file matches as it finishes.

A connection between two browsers has to start somewhere. Our server is that introduction: it passes the two sides enough information to find each other on the network, then stops being involved.

It handles connection details and nothing else. No part of your file passes through it, which also means there is no upload step and no waiting for one to finish before the other person can start downloading.

A transfer carries a batch: a handful of files, or a folder with its structure intact. The other side is told the whole list up front — names and sizes, nothing else — and decides once, for all of it.

The files then stream back to back with nothing in between. There is no pause between one file and the next and no round trip to negotiate each one, which is what keeps a folder of two hundred small files from taking longer than the one big file beside it. Each file still gets its own checksum, so a failure names the file it happened to.

The files are read from your disk as they send. There is no copy sitting anywhere else, so if you close the tab, there is nothing left to send from. The same is true in reverse — the other person has to be there to receive them.

This is the real trade-off compared to an upload service. You cannot send a file to someone who is asleep. In exchange, your file never sits on a stranger’s hard drive.

The limit is 100 GB per transfer, counted across every file in it. Neither side ever holds a whole file in memory: bytes are read in 64 KB pieces, sent, and written straight to disk on the other end. A 100 GB transfer uses about as much memory as a 100 MB one.

Writing directly to disk needs a browser feature that not every browser has. If yours cannot do it, the page says so before you start rather than failing partway through.

Some networks — corporate firewalls, a few mobile carriers — will not allow two devices to talk to each other directly. When that happens the connection fails outright and the page tells you. It does not silently reroute your files through somewhere else.

The diagnostics page tests this network without starting a transfer, and says which kinds of path it managed to find. That turns “it will not connect” into something specific enough to act on.