Do you send takedown notices for us?
No. Carriagegraph finds the servers, ranks them by who can act, finds the right contact and writes the complaint with its evidence. Your team chooses what to send, so every notice goes out under your name and your policy.
How do you avoid flagging our own or licensed distribution?
Free-TV platforms and licensed distributors are recognised, and you declare your own hosts once. Streams from them are never counted as findings.
How is evidence kept?
Every capture keeps its frames, the text read on screen, a time stamp, an integrity hash and the network that carried it. Reports bring them together as case files with an evidence register.
Can our AI assistant use it?
Yes. Add Carriagegraph as an MCP connector in Claude, ChatGPT or another assistant. It sees what your role allows, nothing more.
How is this different from an open channel list?
An open list gives a name and a URL that somebody once saw. Carriagegraph tells you whether it answers right now, whether a real player can use it, at what quality, from which host, and since when. Those lists are an input to this product, not a substitute for it.
Could we not build this ourselves?
You could build a crawler in a week. The hard parts are the other five subsystems and the years: identity resolution across hundreds of thousands of raw names, verification at six-figure daily volume inside a fixed cost envelope, classification from evidence rather than labels, and above all a dated record that only exists if you started collecting it before you needed it. Every day the index runs is a day of history a new entrant does not have.
How is this different from my content protection vendor?
A vendor's detection is keyed to the content you enrolled and the places it is pointed at. Carriagegraph watches the whole open ecosystem, including services and channels nobody pointed it at, follows each one to the operation, sponsors and hosts behind it, and keeps the evidence. It can work alongside your vendor or directly with your team.
How do you know a stream really shows the programme?
Because we look. A real browser plays the stream, two frames are taken eight seconds apart, and a vision model judges them, reading any text on screen in any language. On 1 October 2026, 14% of the channels whose streams played showed no programme at all: a notice, a test card or a blank screen.
How do you work within the law?
Carriagegraph observes what is publicly reachable and records what it sees, as evidence for the rights holder. It does not host, redistribute or provide access to any stream, and it never publishes a pirate address: the public explorer shows official destinations only. Complaints go out only when the rights holder sends them. How this applies in your jurisdiction is a question for your counsel, and we are glad to walk them through the method.
How fresh is the data?
Endpoints are re-verified on a rolling schedule that adapts to each one: reachable feeds every few hours, the rest on a backoff that lengthens as they stay dark. Several million verifications a month. Every observation carries its own timestamp, so you always know the age of what you are reading.
Can I get history for a channel we never monitored?
Usually yes. The index is not customer-scoped, so coverage of a channel begins when we first observed it, not when you became a customer.
What does it cost?
Every engagement is quoted, after the first report, because the right scope only becomes clear once you have seen where your channels actually appear. The first report is free. A pilot has a fixed fee, agreed with you and credited against your first year.
Can we sign up online?
No. Every customer starts with a conversation and a carriage report. Once the scope and the quote are agreed, we open accounts for the people you name, with the roles you choose.